Hawes - Consider using a GFI outlet at your power strip just in case. Most mishaps in life make you stronger....except for electricity. It just kills you.
He's doing it the smart way by interrupting the power to the socket. And now that he is using a SSR he is insulating himself from liability that much more. He needs to have a big fat warning regarding using his device on appliances that contain their own microprocessor though. Your point about GFIs (if not already installed) is a good idea. And as an electrician who started out his journey in 1975, when it comes to getting popped I'd have to say that at worst it only stung a whole hell of a lot. Nevertheless, I still wire things hot, it was how we did it back then.
You have essentially created a very expensive light dimmer, which is why it buzzes, as you have learned. The triac is burning itself out because you are turning it on in the middle of the cycle, thus it needs to dissipate the switching current at that point and generates a huge amount of harmonics in doing so. Your second idea is on point, and you can use a slightly larger triac (80% over rated) with a heat sink. You can still PWM, but you need to think about doing it over something like N cycles. If N = 100, and you want a 50% duty cycle, on for 50 half cycles and off for the next 50 cycles. The temp won’t vary that fast, so even using 256 over 2 seconds is perfectly fine. It will dissipate much less heat as well.
Note that chopping the input may mess up ovens that have timer electronics. Fan ovens probably have better uniformity, but also may be unhappy with chopped mains
Yep, the oven I'm using for reflows (which is a convection oven, basically the bigger version of the Black and Decker Stephen shows) _DOES NOT_ like chopped AC, so what I'm doing is using a programmable industrial temp controller that has a configurable dwell time, which I have set to 10 seconds, and the oven is set to full temp and no shutoff time (timer disabled). The controller switches mains to the oven but does so in 10-second bursts instead of playing with the mains waveform.
Yes, adding a fan will help a ton with uniformity. When I made mine, the propagation of the heat made using faster than 0.5s pwm periods not necessary.
Yeah, if the project is to avoid ripping out any existing controller, the oven chosen has to be completely "mechanical", just switches, and a timer based on clockwork.
Bold choice to leave your board with an untested mains circuit plugged into the USB of your laptop haha. Great project, really interested in the result. I feel like since heat is such a slow moving thing, you don't really need that fast of a PWM signal. You could probably get away with switching 2-4 times a second and still get super precise control. Most older microwaves literally have a 5 second duty cycle.
Yeah fr he's going wicked overkill with that control scheme imo. I once had to build a couple dirt cheap hotplates for a chemistry experiment and I just straight up hooked a coffee maker heating element to a ssr and controlled it with a free arduino pid library on a nano, using cheap thermistors I had lying around for feedback. The control there was setup for mechanical relays using on time 'windows.' They worked exactly how I wanted, I even ripped a processing sketch off the internet to graph the curves for tuning the pid loops, once tuned in they were rock solid.
Naw he's using an optocoupler and has huge creepage and clearance between mains side and ELV side. No problem leaving it connected to a laptop. Also, bang-bang control (like pulse density modulation) at zero-crossings is the right approach and doesn't force you to make reaction time so slow ("2-4 times a second")
@@NavinF It may be the right approach for some things, but an oven doesn't need that level of control. 2-4 times a second is plenty (by an order of magnitude, really) for an oven and a heating curve. PID control and you're good to go.
That is is a wicked project Stephen! An "all in one box" to plug a toaster into was something I was looking at for a long time, but as you now see, the SSR is mandatory, and you'll likely find an SSR mounted to the main PCB still an issue (I did, I couldn't find anything that didn't still require active cooling on the PCB) which is why I landed on an external SSR requirement. They still get quite hot, and I always recommend folks don't even mount them inside their toaster without some thermal shielding as mounting a device that gets hot to a unit that is designed to generate a lot of heat.. um, danger danger Will Robinson! hahaha. I look forward to seeing your next steps on this, and thanks for the nice shoutout at the start and the RMP link in the description!
Yeah, you need a zero-crossing SSR. There are a ton of open-source reflow oven controller projects to adapt. The AQ-A series of SSR's from Panasonic seem to be popular. a 15 amp, 75-250VAC model is about 20 bucks for onesies, 15 bucks in quantity. The heat sink can sometimes be more expensive than the part, they need beefy ones.
Im sure it's been said before in the comments, but I'm too lazy to scan. The hysteresis of a heating element (not to mention the oven as a whole) is _Long_ , so PID controlling at mains frequency might be overkill and harder than using bang-bang controller with an allowable temperature range. Just imagine trying to control your home HVAC with both methods and my comment will make make more sense.
I'm using a PID controller with a configurable dwell time set to 10 seconds because of this issue - the quartz IR tubes in these ovens don't do chopped AC reliably/predictably, and power cycling at any shorter a timeframe than about 10 seconds doesn't get you that sweet soak/bake curve you need for effective reflow.
@@OddlyIncredible I always thought the mercury switch thermostat was simple genius. The mass of the mercury swaying from one side to the other set the dwell.
Hey :) There are optocouplers with diodes both ways on the inputs so you can detect zero-corssing without external diode bridge (4 diodes). Like LTV-814. And yeah, it is better to regulate heaters by switching on / skipping a whole wave halves instead of chopping the wave (you can count the waves by an MCU). That sound you heard (like a PWM sound as you called it) was the heater hit by a sharp front of the wave. The closer to the wave peak you chop it, the louder the sound. I am driving a 400-Watt heater plate by BT138 (TO-220 package) without any heatsinks and it is not even warm...
A tip is to connect a light bulb in series when you test electronics with 230 or 120 volts for the first time. If there is a shortcut the light bulb will light up and no danger.
When i was in college I liked to work on Hi-Fi gear. One day I was working on the power supply for an amp, and I had it all bare on my bench, plugged into mains running a test. I thought I was being safe, but while I was testing i was on my other bench using the computer. My buddy walks into the room, and as i turn to him he says "oh that's a beefy looking transformer" and he picks it up! I almost had a heart attack as i yelled at him to drop it. It's cool that you built a box so you don't have to handle the mains side of the board, but it's not just you that you have to worry about. My buddy was fine, he managed to out it down without touching copper, but that was a lesson i wouldn't forget.
First reflow oven was a Hamilton Beach I picked up on the side of the road on trash day - full blast with multimeter TC to run up to 247-250 w/ GC10 paste made a lot of boards, just had to watch it close to turn off and open the door (wish I had one of the diy kits with the door opener puller to reduce chance of burning boards). Last year I went with a T-937M reflow oven in their 120v variant. The 220v variant has its issues but may have worked. Either case, running high current 120 or 220 isnt great for most users :/ so this project is definitely going to help the community! Best feature on the 937M for me is the exhaust and intake fan for unattended cooldown for sure. Duct for the exhaust is great to get fumes out, and the intake fan also runs for temp regulation and circulation fan. Oven in the unconditioned garage so 3000W heating keeps up in the winter time :)
Honestly, the best 'don't touch anything inside' might just be to pop the temperature control knob off and mount a servo with double-sided tape. You won't have to touch mains OR the oven itself, and switching ovens would be a matter of five minutes for moving the hardware over followed by recalibrating everything. (That being said, a simple relay to turn the oven on and off would also be handy.)
We did something like this, back in 1978, to control a 10 kW arc lamp. Switching full AC (no rectifier) OFF at the desired point causes less dissipation in the triac. (And MIGHT clean up the noise.) As I recall (it's been a few months, right?) we used a diac to drive the triac. Again, from really failing memory, I think we had an RC snubber across the output. (No need for it in our application, it was industrial - but you may want to use one.)
I rebuilt a controller board for my kitchen oven. I went through exactly the same process with zero crossing detector and triac, finding out i needed a ton of cooling and had loads of noise then ending up with SCR. It turns out that you don't need so much fine control over the on / off period. Seconds of time are good enough and the tuning can be done with PID. Reducing the amount of switching also lengthens the life span of the SCR.
SSRs are also just this triac circuit with a zero crossing controller chip(something like the MOC3041M), potted and on a heatsink. There's nothing magical about them. But yes, I'd recommend that you buy a pre-made unit if you want to avoid the hassle of bringing mains onto your PCBs, which can complicate certification.
In general switching elements (and the things they power) also don't like to switch in and out in the middle of the sine wave. A better approach might be to apply power for a percentage of counts instead of a percentage of every count. Very easy change since you already have a microcontroller. These were literally the next words you spoke after I wrote my comment lol.
Nothing wrong with the system he's using. Phase cut dimming has been around for decades and is a well tested and reliable method for controlling AC power to resistive loads. It's used in almost every household lighting dimmer ever made and is used for theatrical lighting into the 10s of kW range. The only real downside is it's noisy, but that's not that difficult to filter out with a basic LPF.
But the temperature of a thick heater follows switching cycles much more slowly than that of a thin filament in a lightbulb. In order to control the light intensity, you have to switch quickly and often. I think, for a heating element, I wouldn't use phase control either. 🤷♂@@JaenEngineering
@@JaenEngineering phase cut dimming is used when switching cycles will not achieve the desired result, largely lighting as you have pointed out. When you don't need such tight and constant control of the power using cyclic switching is much easier on the components as if done at zero cross they are switching no current, and instead of switching on every voltage cycle they only switch for whatever duty cycle is required. All switching elements have a limited number of times they can switch state whether mechanical or solid state. That number may be trillions of cycles but there is a limit. By using switching cycles the highest number of switching events will be at 50% power meaning that at WORST case cyclic switching would double the lifespan of the components before you even take into account the strain of switching under load. In order to switch under load the switching elements have to be more robust and more expensive. Yes you can use phase cut dimming, but if you already have everything needed to implement cyclic switching at zero cross, it saves you money in components and extends the life of your product, why would you not use it?
I built my reflow oven some years ago using the Whizoo Controleo3 full kit. It was a bit expensive, but I liked their due to coming with everything but the oven for a full overkill build , separate control for each heating element, and auto door opening for cool down. The benefit of doing the overkill build with full insulation and heat reflection is the oven being able to easily follow profiles. One of my projects had some components with some pretty strict reflow curves, and the oven was able to do it on the first try. You absolutely want an oven with top and bottom quartz heating elements. The oven I went with was the BLACK+DECKER TO1313SBD for a whole $30 shipped. I wouldn't reinvent the SCR.... Properly rated, commercial available SCRs are pretty cheap from Digikey or Mouser.
Many years ago I did a lot of design work with triacs controlling halogen projector lamps. You really have to be careful with dv/dt across the triac as well as conductive noise suppression to keep the FCC happy. Using an SSR that triggers 'on' near the zero crossing as well as 'off' at zero crossing is the best way to go IMHO.
I built a Controleo3 reflow oven using Whizoo's kit and a Black & Decker toaster oven. It works really well, I've made several boards in it. I just got my Lumen PNP delivered a couple weeks ago and got it all configured and calibrated, successfully ran the Functional Test Placement board. Can't wait to get some of my own boards placed with the Lumen and reflowed in the oven. Up until now I've been manually placing parts with an AirPick and a microscope, so glad I'm not going to have to do that anymore.
i ve build such a control for an oven, a ssr is the way to go, they are super simple to use. it is a "wave packet control" - if this is a good translation from the german word "Wellenpacketsteuerung" . the needed pid controller is super simple to find good values for i and d because the system is very sluggish. their allhough a lot of ready to use and highly advanced open source projects :)
Having made a few reflow ovens like this for myself I used an ssr and didn’t worry about cycle switching as you can go pretty slow. I cycle twice a second and the pid works like a charm the heating curves are close to perfect. Good luck with your project.
When i made mine i found that very short pwm periods work just as well as many second long ones because of the masive thermal mass you are heating up. You also cant just have one temperature probe as the temperature is not even and takes a while to propagate to the area of the board. The heating elements also emits large amounts of IR which makes non-reflective objects heat faster such as fr-4.
Good work so far. Since you will have to redesign, might want to add a power converter so the MPU is on when it is plugged in to power. The switching may be better suited for Solid State relays as they do come in higher amp capabilities and your zero crossing circuit would still be able to detect when to turn on. The only issue I see with that option for switching is the cut on and off timing which may be able to be compensated for in programming, possibly adding a thermal sensor to assist in " auto programming " the timing for set temperatures and change the timing of the relay.
@steven, great work! One comment on the mains circuits, you should use multiple resistors in series on the high voltage side. You'll see in the data sheets that they have working voltages typically maxing out at about 200v. Keep in mind that 110v is RMS so peak is higher, plus you need to allow for surge voltages too. If you ever go down the rabbit hole of certification with this you'll be forced to learn all of this in detail, like I had to! Ps it was great meeting you in Munich 😊
Very cool project. Please be careful with these open AC flats. I have also converted a toaster oven. Experience has shown that these things are so slow that it is actually enough to switch the power on/off with an SSR. So I was able to solve the whole 230V switching with an SSR, which I switch with a FET. These ovens are so slow that you have to switch on/off relatively slowly to keep the temperature +/- approximately. Thus, the active line control is not necessary at all, but an on / off is enough in this case. Am curious about the next video.
I have had few problems with making small quantities on the stock T962 running lead free profile 3 on leaded solder. No mods at all. Just lift the board off the base with a couple shims to avoid cold spots. But I bought a 6-zone conveyorized oven for actual small scale production. Use time proportioning with a 10-20 second cycle time and zero-crossing SSR (isolation provided by the SSR so everything on your board is low voltage). Heat sink fairly large (there are design curves from the heat sink manufacturers) for 10W or add a noisy fan to save money, weight and size. Beware that a short to the mains can not only electrocute you but could blow the **** out of your computer via the USB connection to earth. USB isolators are available if you are belt-and-suspenders cautious.
Also, you can use the zero crossing detection to turn relays on/off without arcing, to do this you need to determine the on-delay and off-delay of the relays. Then you use the zero crossing input signal to start a timer, when the timer elapses the output channel of the timer turns on the relay just before the start of the next zero crossing event, the relay then should activate at 0V which will then not cause arcing on the contacts. to turn off the relay do the same thing, but adjust the timer to take account how long the relay takes to turn off. turning on a relay is solenoid activated, turning off a relay is usually via a return spring, both on/off actions take different amount of time to complete. I use zero crossing activated relays to control bar heater elements. and i use an SSR to control power to a 2kw halogen lamp. the bar heaters only come on when the different between the target temperature and actual temperature is large, this avoids excessive switching. the halogen lamp reacts MUCH quicker, so an SSR is used to control the halogent lamp. combo microwave convection ovens are the best as they have turntables and convection fans and are the most easy to retofit, but not as simple to control... also, microwaves have excellent interlocking safety facilities (switches) which can be re-purposed to automatically kill the power to heater elements if the door is opened. the biggest issue you'll have with using a toaster oven is the downward slope of the reflow profile, i've seen people use aircon butterfly valves and an inline blower to get the oven to cool down quickly. toaster ovens are also terrible at keeping the heat inside them. i used rockwool which is use for fireplace insulation around mine, made a huuge difference.
When developing a product with mains, it helps to cover the high voltage section of your board with a insulator, taping a thick piece of acrylic for instance. I also recommend checking if you're plugged into a circuit with an RCD/GFCI. It prevents you from getting zapped when touching the phase (though not when touching phase and neutral)
*Yes reflow should be solved. And there should be great SMT protoboards also. And pick and place with a robot arm, a vac to pick up the components, and OCR to read the vals, and a shaker to shake the components upright should also be a community project. Post a video comparing that reflow method to the sand in a pan on a stove method. Thanks in advance.*
There is a simple solution to even out the hot/cold areas in the T-962 oven. What you need is a fan to agitate the air inside. This oven is a bit cramped, but if you drill a hole from the top, run a shaft through it and attach a DC motor controlled by PWM, or a stepper, you can control the rotation speed. Inside the oven what you need it a propeller, something basic and made of a metal that can withstand the max temperature temperature the oven can reach. I like to use a minimum of 25% above the rated value as a safety factor. A twisted strip of metal with a hole in the center should do, you don't need something fancy. The bearings need to be able to handle the temperature as well, so make sure you get some that will be able to handle the heat. This won't require a high RPM, in fact, very high RPM's may generate so much draft that it could displace the components from their desired locations and it won't make your oven as homogeneous as it can be. When you implement this the oven's characteristics will change, so for very delicate processes, it could be a challenge to get it just right. Firmware changes won't do much. You can get the oven to be homogeneous, even if you don't make any mods, but for that you need to lower the climb rate so much that the heating curves will be dramatically different, and could render the process useless for your desired purposes. The reason why you can't get it to perform any better has to do with the heating elements and how they are placed in the oven. If you had 4 of them, one in each corner, and were able to control them individually, you would have much better results. This also means 4 thermocouples at the very least, ideally 6. One close to each heating element, one for the PCB and one for the air inside the oven. The air thermocouple will govern the process temperatura (process value), the PCB thermocouple will be your failsafe (it will cut off the heating if it overshoots the setpoint), and the 4 in the corners would determine which heating elements need to provide more or less heat to even out the temperatures in order to keep the oven under tight control. In theory you can pass on the fan with this type of control. It will make the oven more efficient and quieter, and your process will benefit greatly. However, you will need to spend more money in heating elements, implement 4 PID's and do a bit of head scratching and possibly melt your retinas a bit learning how this works. It's up to you to decide which solution you want. One is messy, complex and expensive, but as a tighter control and is more gratifying, while the other is way simpler and it could just be enough for you. If you think about it, it's probably better to just get an old microwave oven chassis and build it from scratch. That way you can implement everything, and it's a neat project. Plus, you will get a lot more clearance, the T962 is quite cramped. Not a bad desktop solution, just not good enough for SMD soldering. You get the PCB's charred in the center and the solder may not even melt at the edges.The control is limited to one thermocouple which tracks the process value, and the heating elements (they are 2 if memory doesn't fail) seem to be connected in parallel, meaning 1 output only. A bit of a caveman approach, if you ask me. Not hot enough? Club it a few more times, it will get there. Another solution could be a multi-zone hotplate. You can get a steel plate, divide the plate in 9 zones like in a tic tac toe board, and control each of them individually. Whatever suits you best. The control can be done using MCU's like the arduino, raspberry pie or ESP32, they are not that difficult, and languages such as microPython are a good option for a slow process like temperature control. If you want to go the extra mile and be able to provide rapid cooling, you can always run a copper tube welded to the metal plate and flowing from and to a tank. To avoid getting the tank too hot, you may require a mini cooling tower. It's visually fantastic and very efficient, but not a good choice for a lab as the resulting condensation can cause mold and it's also not good for electrical/electronic devices that cohabitate in the same room. You will also need to compensate for the lost water. This is really over the top, I just thought that it would be interesting to add for the sake of providing a bit more info, and people can look it up if they so wish. Keep having fun kid, you are doing a nice work. Rolling up your sleeves and having a go at what you want is a great example for others to follow. As long as people play it safe, the worst thing that can happen is a failure. If they don't, well then they can burn down the house and the entire family will have to share the basement. Not my idea of fun.
I basically built what you're up to but for incubating eggs with a standard light bulb instead of reflowing solder paste. The right PID coefficients have been the most tricky part here. For obvious reasons I could not tolerate any overshoot whatsoever. I need however a target temperature as smooth as can be in a range of less than 0.2°C. To top things up the ramp up should be quite fast. So, 5 minutes max to 35.7°C, no overshoot, keep it there, little wiggle room of E~0.2°C. And, yes, make it robust against disturbances, no big over- or undershoots there either. Hard to hit those targets combined. Up to now, there's always been tradeoffs, haven't been able to meet all criteria at once, yet. Top prio is E and overshoot. Undershoot and ramp-up is least critical. With reflow ovens it's maybe less of an ordeal. Keep us in the loop, I'm curious about your reflow profile confidence level.
What temperature sensor arrangement are you using? Type of sensor, placement and so on? And does your chamber have a method to circulate the air (like a fan)?
@@Graham_Wideman K-type (for the PID loop) and an SI7021-type sensor (via esp32, tasmota) for verification. Both between and at the top of the eggs. Third, 7021-type, sensor at incubator ceiling. Fan to distribute heat/prevent hot spots. Incubator is isolated by 4cm styro. No window but an esp32-cam. Dimmable (PWM) LED strip (still flickers a bit at the set frequency). Fan does not do it's job as intended yet. Still very distinct heat strata (astonishingly so). WIP.
Very nice to see you doing reseach and development for a new product! BTW I'm guessing that Joule may be a provisional name, but be aware that there's a sous vide thing called Joule as well (Interesting how they got a trademark on a unit)
Not only that, but getting a trademark and having that trademark hold up when contested are two _entirely_ different things. Trademarking a measurement unit name (especially one that's an established and clearly defined standard unit) is likely to not be enforceable in any practical sense, regardless of any assumptions of distinction from a marketing perspective. Nobody will confuse a fruit for a cellphone, but a glorified water heater and a unit of measurement for electrical power have too much overlap in too narrow a cognitive space. (Disclaimer: IANAL but I've had to deal with IP rights, and there's a LOT of interesting shenanigans involving trademarks.)
@@OddlyIncredible Apple trademarjed Newton, then there's Tesla, quite a few companies named Henry, and no doubt more, considering many units are named after people, as is Joule. I don't think that having your would-be trade name already in use for a unit really presents much of an obstacle.
MOC3031 is a zero crossing capable driver, you were really close with the MOC3022, I might be mistaken, but part of the issue could be if you are PWM'ing anywhere in the middle of the sine waves, you are snapping through a large inefficient linear region to get saturated each time, just like with transistors it's not somewhere you want to be for long at all, pretty sure that's another reason zero crossing is often more preferred. I made a 555 drive a MOC3031 for a spot welder controller I designed. With the triacs, pay attention to the ratings and do the dissipation calculations(as you've already realized), also some are available in plastic isolated packages, others will possibly make a heatsink go live.
The way I implemented something similar for a soldering iron without an expensive SSR and ruining the power factor+EMC is to use back to back FETs with a photovoltaic gate driver. To convert the PID controller's output to pulses, I used a RNG and compared with the PID output every interrupt to decide whether to let the upcoming pulse through or not. On average this works quite well since you're doing it at 100-120Hz and the time constant of everything else is in the 10s of seconds Also definitely make sure you're using an optocoupler to detect zero crossings, having a fully isolated controller is a really good idea for safety especially if you want a human pressing buttons on that board. Make sure you respect creepage and clearance requirements for your voltage level too across the isolation barrier. AC optocouplers are the simplest implementation for this imo
Stephen: My top suggestion is this: Assuming you want to do better than just wafting hot air over a board until the solder melts, your task is to get the oven to perform a series of heating steps to get the PCB, specifically the solder, pads and component leads, to follow a thermal profile, for example that published in component datasheets. Many oven mod kits attempt to get the temperature feedback from a thermocouple just floating in the air, perhaps above the board(s). But it turns out that's a pretty poor approximation to what the board is experiencing. So at some point you will need some thermocouples mounted to PCBs, and thermally coupled to pads (maybe with high-temp glue), that you can place in the oven to test actual performance, including in different locations in the oven. You can even do what I ended up doing -- use a small PCB with thermocouple attached to act as the feedback sensor, and it sits on the tray next to the board(s) being reflowed. This is especially useful for lead-free solder where the margin between good soldering temperature and burning the board is relatively less than for tin-lead solder. Several thermocouples-on-PCB is the approach used for profiling commercial conveyor reflow ovens. (These are usually connected to a data logger inside an insulated box that accompanies the test board through the oven.) Anyhow, you're going to need a few thermocouples, and it's useful to get something like an inexpensive 4-channel handheld thermocouple "meter" with logging. I got a slightly upmarket PerfectPrime TC0520, which can also output data to PC for plotting. Pretty convenient for a variety of tasks. We even used it with 4 long-lead thermocouples attached to a PCB to do some basic profiling on an industrial conveyor reflow oven we work with. Anyhow, bottom line is that rather than puzzle over vaguely better or worse soldered boards, you'll want some ground truth, which you will get closer to with thermocouples mounted to PCBs.
I've worked on similar circuits in the past (mains appliances controlled by a microcontroller for home automation, though at lower currents and without PWM), so I have a few notes/suggestions: You explained phase angle control (/phase cutting), but also mentioned that full-wave switching might be better for EMI. As far as I know, there are also power factor specifications, at least for larger loads, which may require lots of filtering if using phase cutting. Generally, switching near zero-crossings is more efficient anyway (for resistive loads), so this seems like an easy choice, also given that the oven's response time is dozens of seconds at least, so you likely don't need the faster/more granular control of phase cutting. Most solid-state relays I'm aware of also use triacs/thyristors for switching, and hence share the ~1V drop and high power dissipation. Their main advantage is ease of use and low part count, but since you're making a custom PCB anyway, it's not much extra effort to implement that functionality yourself (or get a zero-crossing controller and a separate triac). There are plenty of triacs available that are beefier than the one you used, as the manufacturers know they need lots of heatsinking, though you'll likely need to go THT (maybe TO-220 or TO-3) to actually dissipate dozens of watts. Alternatively, you can use high-voltage MOSFETs for switching AC, either using two of them back-to-back or by shunting the DC outputs of a full-bridge rectifier. These offer generally lower voltage drops and hence power dissipation, but they are likely less reliable (due to transient overvoltages, see also below) and are harder to drive (no automatic shutoff on zero crossing, needs correct polarity of gate voltage), so many people seem to recommend against this. If the ovens can run on DC though, there also the option to rectify and zero-voltage switch, as used by a soldering station at th-cam.com/video/erKCA71q7cg/w-d-xo.html : you can rectify but not buffer mains, so you get unidirectional ~110V RMS, but with regular zero-crossings, so a single MOSFET (with low on-resistance and hence continuous losses) can be used. Then by switching near those crossings ("zero-voltage switching") at low voltage and current, you can get by with very low switching losses. Finally, before this becomes an actual product, I'd recommend you carefully test its EMI *resistance*, as having mains on the same PCB as a microcontroller can be interference galore. I'm currently planning another revision of my project, because I've noticed that switching off fluorescent lights in an adjacent room sometimes causes signal glitches in the low-voltage section, and this will likely only get worse with the larger currents and frequent switching you're planning to implement. (Disclaimer: While I'm studying EE, this is not my speciality, and I'm by no means an expert. Please take all the above with a grain of salt)
Indeed running a high wattage load at those kinds of power factors are quite "rude" to the utility company... With regard to regular high voltage mosfets the i^2*r quickly (I has comparing @16A and IGBTs not triacs) became as bad as the IGBTs, good (expensive) GaN and SiC did however beat the IGBTs. I'm inclined to believe that here a triac is probably just fine unless Stephen really wants an efficient controller. Diode rectification comes with its own "horrible" voltage drop.
@@TalpaDK Right, my low-voltage habits are showing - of course IGBTs are an option at mains voltages, and I did forget the bridge rectifier voltage drop. Regarding I^2*R losses, the video mentions 10A, so I estimated that
@@hadinossanosam4459 reflow ovens are often >= 1300W so I would expect a bit more than 10A for a US installation, but yes a EPC2304ENGRT with its Ron of ~3.1mOhm would be way easier to cool even @ 13A (523mW * 2). Is also not your garden variant Si mosfet but actually one of them shiny GaN ones (and a bit on the "expensive" side). Stephen would probably have to spring for something even fancier, as he would probably also want to target 230V countries which would push the voltage requirement up to 325V+margins. I also hope that he is considering adding a MOV and a fuse to his setup for some basic transient protection regardless of his choice of switching device.
Did very similar project as my bachelor thesis (using hot plate additionally cooled by air pump for air mattress). I was surprised how small triac you used couse remember how much mine got really hot - few minutes later and now I see you also faced that problem :) Good job though!
Actually, there is a product that satisfies all of your requirements - the BetaLayout Reflow Controller V3 Pro. I am working with it for years now without issues. Just a little fiddly to place the temperature sensor, but the results always turn out beautifully.
Using the S.S.R , both types are available...Phase control type and Z.C.D. type...with the zcd it's controlled as how you explained .. that's known as " burst firing mode"... Phase control can give tight temp. control..with a good ckt. design...and input voltage fluctuations and " sine to linear ramp" linearisation circuit can be implemented to give a linear temp.rise over the entire temperature range.... for best results with phase control, one can initialize a burst mode followed with phase control for temperature tracking accuracy.
Concur on switching to an SSR. It's still a proportional control system where the power to the heaters is proportional to the multi-cycle on-off time. Actually, this makes more sense as the thermal reaction time of the heating elements is actually quite slow as you've shown by the full-on ramping time experiments at the beginning. In effect, you're modulating reduced heating rates where that test curve is the maximum achievable heating rate given the existing heating elements, oven insulation, oven thermal mass and thermal leakage through the door edges of the unmodified oven.
12:38 and here is where you probably should have covered the top side of the board around the AC section, because all that exposed area is ripe for finger touching while its live.
One more thought: There are solutions to your problem out there. Someone else mentioned a PID controller, which is true (but I don't know if they handle the kind of curves you're trying to follow). What I thought of, is heat treating ovens for knifemaking and metal working. Different temp ranges, but they also handle fairly complex heating curves that are user-programmable.
Looking forward to seeing how this project progresses. Always interesting to see different approaches, though my gut tells me that the thermal lag of the heating elements is so long, compared to that of your control loop, that you are not going to see any better control than your traditional bang-bang controller that is typically used. (on and off for long periods of many cycles) Certainly sub-cycle control, with the triac won't gain you anything here. Good luck with it, interested to see how it goes.
hum...a few thought about the slow reaction curve i assuming the heating elements are some kind of threded/induction elements ...when on a magnetic flow builds up (magnetic flux) ..when thay are rapidly turned off...the element(inductor) will react by reversing the flow of electrons.. aka puch back effect, and since thay cant flow back trough the turned off triac ther will be residual energy left when the next switch on comes... solution: ...u add an inverter to the triac gate signal, connected to a shunt triac and a diode connected to a large capacitor with a discharge resistor to slowly drain it safely (no short) ...this should increase the responsivness of the owan itself... ...to test, u should check with a lage capacitor and no way of discharge it how what voltage u get after some time so u know of much excess puchback u need to dispose... ..u can checkup puchback or flyback effect to understand this better...
I suggest you take a look at driving soldering irons or other open source reflow oven projects, Soldering irons are driven by 24V (AC OR DC) typically but usually driven by 2 SSR's or solid-state relays. I would say this is likely to be the best approach. Keep in mind resistive heaters don't care whether you are driving them with AC or DC so you could potentionally convert it to use DC and use MOSFET's. I would just do some research if the heating element used can actually work on DC.
I've done some of this in the past. Maybe not applicable in this instance but, when driving inductive loads, you should switch complete mains cycles (I divided the zero-crossing pulses by 2 and worked off one edge). Also, rather than n cycles on and n cycles off you can provide small bursts of several cycles on to match the dynamics of the halogen heater elements.
You are making something really similar to the Beta Layout Home-Office SMD Kit that Dave Jones made a video about almost 10 years ago. We've had that in the office for several years now and its awesome ! Best way to get a cheap reflow over I've found so far. As far as mains control on the heater goes - the best cheap way i know of controlling those currents is with a beefier TRIAC, in like TO-220 or bigger case plus heatsink coupled with a zero-cross triac driver. Something like the MOC3043SM, that is extremely popular. Its a way cheaper solution compared to a solid state relay. The way i like to use this combo with a regulator (PID or other) is by just turning the cycles in groups of 20. 1 cycle is 5% power, 2 is 10%, etc, etc. If i want finer control i can get up to groups of 40 cycles. Since heaters are VERY slow, the end result is the same as if I'd used per-wave chopping. I picked 20 since I'm in the EU, we have 50Hz mains here so 20 cycles is 0.5S. 40 is 1S.
I'd like to note that the tab on many triacs is live. You must provide proper ground isolation because the heat sink can become line potential. You can get isolated tab triacs, so be observant. Be careful of unproven sources for SSR's. There are literally billions of fakes out there. Say they're rated for X amps, when investigated have internals rated for 1/3 of requirement.
An SSR will work great. We used one with an off the shelf PID controller to control a crock pot a few years ago. You still need a heat sink, and you need to wire the center screw post to ground (a lot of guides don't tell you that). But seriously, you don't need a micro controller or anything for this, unless you want a fancy ramp or a timer or something. If you just want to put something at an exact temperature, an off the shelf PID/SSR kit will work fine.
A fancy ramp is exactly what Stephen wants to do, as specified in component datasheets. The general principle is to ramp the board (and components) up to a plateau below solder-melting temperature, let it sit there briefly so that all of board and components get close to that temperature, then "sprint" up to a little over melting temperature, melt for a limited time, then a controlled descent back to ambient. The idea is for the time above solder melting temperature should be long enough but not too long. That much precision is not necessary for making a few hobby boards, but it is important for producing larger quantities where you're confident that the components have accumulated minimum damage from the soldering process.
@@Graham_Widemanto go that far , calls for a contoured temperature profile according to the process possible in this small oven.. but here everything is manually labour intensive..hence a bit of overkill.
@@Graham_Wideman then he can do it with an SSR, a mosfet, and an Arduino. You can prototype this with a RAMPS 1.x board, no soldering required. Or a solderless breadboard would work too. It's simple to do.
@@BrainSlugs83 Your suggestion lacks a graphical display, which of course could be added. But for Stephen, it's probably just as quick to knock out a PCB as it is to mess around with breadboards, or getting set up for Arduino and RAMPS etc. Instead, he gets a system with an MCU he's familiar with and already has a suite of libraries for. So sure, all sorts of things can be done with Arduino, doesn't mean that path is simplest for everyone.
Basically it’s a phase cut dimmer your making, I could suggest an FL5160 it is controlled by 0-5v signal, and uses 2 IGBT which you would need to mount to a heat sink.
I think an SSR is the best solution to this, just be sure to get a genuine one from a reputable manufacturer. There's lots of cheap clones out there if you don't buy from the usual suppliers. Also, keep in mind that SSR's tend to fail CLOSED if/when they do, so some form of thermal runaway protection should be implemented if this is to be a product. Loving the progress so far though, this is an awesome project!
Triacs to control 15 amps come in TO3P or TO247 packages. Some TO220 triacs can take the load but you must have the heat sink to back it up. Also check the ratings on the IEC power connector and cords. Many of them are not rated for 15 amps. 14AWG (insert metric equivalent here) wire minimum. Elsewhere in the world, where ~240v power is standard, current is half.
You need to pick up an isolation transformer. This way if you ground yourself while grabbing high voltage you don't make a circuit. Also, I have built 1500W heaters with crydon SSR's that work great. I also recommend adding a thermal fuse to prevent the oven from cooking itself if something goes awry on the control board. I have taken the heater product through FCC and CSA ISO601010 testing if you need help or advice.
You can always run multiple Triacs in parallel, spreading out the current across multiple components. That doesn't help with your EMC problem but it would prevent the parts from overheating.
Equal current sharing in 2 or more triacs is difficult to achieve....a better option is two anti- parallel SCR's each handling one half of the waveform...control ,even from an opto coupled triac is easy with just a polarity steering diode to each scr gate... SCR's are much more robust and have better & higher noise immunity than triacs... This scr config. has been a standard in high current applications.
Years ago there was a product called the Reflowster. It was a controller that turned a toaster oven into a reflow oven with zero modification required. I still have one.
Perhaps they ran into issues with people using newer, more expensive toaster ovens that have electronic controls built-in. The Reflowster really needs a dead-simple toaster oven with physical thermostat that won't be screwed up by power cycling.
You can't go without Zero-crossing nowadays. In you case, I would go with MOC3083M + BTA16-800, through hole so you can find cheap heatsinks, and eventually a good old relay to switch the load on for extended time like when you are heating full power. Heat sink may not even be needed with the relay.
Definitely not an enginner but just a thought, if you just make sure the time it is on for or off for is a multiple of 1/60 of a second (or 1/50 here in the UK), you will always get a consistent amount of power. I feel like something like this is operating over the sort of timescale where being on for 5s then off for 5s would be fine (I think I saw someone in the comments already saying this is about the speed that microwaves run). I also think if the toaster ovens have anything else going on, 'dimming' the current might cause problems that switching for longer sections might not. Also I like the idea of putting an IEC connector on this. I think that is one of the more universal connectors common here in europe as well as the US. I love the idea though, I'd like to get an oven for the lab I work in but there is no way I'm getting a reflow master past the electrical inspections that we have to pass (as much as I know they are safe and solid). Whenever I have students that want to work with mains, I point them to those off the shelf smart plugs and have them write some network code to control it rather than doing anything with the wires... or glue a servo to the switch on a power strip! Anyhow just some thoughts hope something is useful here! Keep up the good work!
You should have still have pretty good granularity with the zero crossing SSR approach. Since you're in the US, you've got a zero crossing 120 times a second. If working over a period of 128 zero crossings, that's 7 bits of resolution with an update rate of once every 1.067 seconds. You just need to have a counter register increment by 1 every zero crossing and reset at 127, have a compare that increments by powers of two. And for each bit in your desired output power value, you either turn it on or of for a length of time that is equal to the value of that bit. So if you're trying to output a power 105 out of 127: 105 is 1101001, so you'd be on for 1 zero crossing, off for 2, off for 4, on for 8, off for 16, on for 32 and on for 64. If you wanted the full 8 bits of resolution, count up to 255 instead and your update rate becomes 2.13 seconds. With the right PID tuning, you could get pretty accurate temp control.
Just a suggestion from a UX standpoint, I would ditch the buttons and use an encoder for the final product. Just like printers do it, I bet because of 3d printers you could get really good deals on 16x2 lcds and encoders from china. They are also more accessible parts for anyone building the device.
This video reminded me of the Keith Wakeham's Maelstrom fan project. He made a device to switch the mains voltage of his cooling fan for indoor cycling so that it would blow harder as your heartrate went up or pushed more power. He has a series of videos on the development on TH-cam. Looking back he used a RobotDYN PWM AC light dimmer module with zero crossing detection just like you designed but the triac on that board has a fairly substantial heat sink. The prototype was very good but the project was cancelled in beta due to the difficulties and cost of getting the device certified. In his final update videos he talks about how the certification would cost tens of thousands of dollars. Good luck, I really hope this works out but I can see why most oven controllers are more diy.
Quartz and resistive elements have a curve, so it might be worth just expanding the duty cycle to a few seconds of ontime, instead of rapid pulsing. Marlin etc have been doing PID tuning for years, I wonder if that code might run better than rapid switching.
I have a background with theatrical dimmers for tungsten lights. It isn't uncommon to see triac based dimmers in the 1KW - 2KW range, but IGBT based dimmers are becoming more common. The main difference being which half of the wave is cut. IGBTs you want to turn on at zero crossing to limit the current flow at that point, but turn off in the middle of the wave. The triacs work as you described and turn off only at zero crossing. A good design incorporating a triac or IGBT shouldn't produce that much electrical noise when controlling a purely resistive load, and in the environments where theatrical dimmers are used, there are typically many thousands of watts of dimmed lights in use in the same building and powered from the same pad mount transformers (those big green deals the power company installs with underground 7200v running to your property). We never had any noise issues that could be traced back to our dimmers with nearly 100KW total load, even with over a hundred audio channels - more than a dozen of which were UHF wireless - and multiple guitars (which are great at picking up stray electrical noise).
From the diagram it looks like you're switching the neutral and passing the hot straight through? In DC I know it's standard to switch on the ground, but afaik with mains you really want to be switching the hot and keeping the neutral unbroken to minimize likelihood of someone getting shocked
Ok, that was hard to see until the end. so mutch stuff I wanted to throw in. first things first: to use a triac you need a snubber (cap in series) so the triac can't be hold open and a filter for the triac (so he switches smove). Than, like with solid state relays (they work with triacs internally) you need some kind of falt protection, they normally stay open when failing. Now to the emi... that's a real problem. I recommend the Würth Elektronik design kits for emi noise reduction (but I think there will be some american stuff to, I'm german) BUT you will need a spectrum analyzer to know where to go for. after all triacs a great for mains control in high precision applications, I use them in the development environmental chambers to control heat and humidity. PS: use another triac, one that does not have mains on its backplate :)
Also look into the practical heat output cycle on the quartz - there is a lower practical limit to how short of a time slice results in practical heat. I've just defaulted to 15 seconds to be perfectly safe and accurate enough. Maybe after a ramp up to heat soak you can go shorter.
Another approach I have used to control a heating element is the opposite of the leading edge triac control, trailing edge control. There's a really great trailing edge controller ic 'FL5150' that drives a pair of MOSFETs with just a plain DC control signal. Although I suspect this would have the same thermal problems as the triac at those high currents.
Trailing edge ..so one always turns the switch on at zero voltage..so no inrush/ turn on spikes..and phase controlled turn off can be a soft turn off type so the EMI/ EMC compliance can be met easily.
Interesting idea to control the temperature. Curious to see how it performs. I went a different route with mine and used an oven with multiple heating elements. I have a short video on my channel with a link to more details. I can control the ramp up speed by only turning on the top two for example. I find this is good enough control, but I guess one down side of this approach is cold/hot spots. Mine is fully insulated, but this meant I wasn't able to cool the oven fast enough if I overshoot so I added an exhaust fan. From my understanding, this isn't recommended most professional reflow specs, but I'm not doing production work so these trade offs were okay for me. Anyway, looking forward to the next video.
Zero cross, run a pid loop that can only output at 5hz. Buy 2 toaster ovens and put both heating elements in one unit. Put a blowfan to counteract overshoot acne cool at the end. I built one and had to do the above to make it perfect
Awesome design, but I saw that small triac frying a mile away, considering the current that oven draws. Definitely look at a SSR rated well above the current rating of your oven. If you build this for sale, and want a UL listing, that's very admirable. So over engineer it and put warnings everywhere. Also, put thermal fuses on your SSR to avoid fires. Your approach to pulse the AC sinewave coming in from the mains is quite interresting. I have been an electronics technician for over 40 years trained by the US Navy. We have 100 KW radio transmitters that autotune. Meaning, the tank circuits of the different stages of amplification tune automatically. They use the AC sine wave much like you did here, except we also look at the polarity of the AC sine wave coming in to drive a DC motor. Polarity sensing lets us move the DC motor in the correct direction to change either the inductance of a coil or a variable capacitor. I AM Very interested in following this project as I am doing the same thing you are... controlling the temperature of an oven. I am very good with hardware but I cannot code at all. I am using a 32 bit ESP32 with WiFi and BlueTooth to accurately control the temperature of a cheap, commercially available meat smoker. I have subscribed to your channel just to see how you progress with this project. You might consider making a Meat Smoker controller with this same circuit, but more robust and able to handle the heater element currents but please make sure you include PID. Also, a wifi or bluetooth connection to a mobile phone app. A meat smoker controller might be more commercially viable for you to make. Much more demand than a flow oven. Do both. I'll buy your meat smoker controller!
I have been talking with Seon of Unexpected Maker, who is reference in this video. I have just purchased his Reflow Oven Pro, or whatever it's called. He seriously warns that using his device to control my meat smoker oven is outside of it's use-case. I must insulate my meat smoker to allow that controller to have a chance to work. I will make a video of how I incase my $150 meat smoker with ceramic fiber "KAOWOOL" and an outside frame of some sort to insulate my smoker. Then use Seon's Reflow oven Pro controller to make it it work, if I can make it work. Not a fault on Seon's controller if it doesn't work. It's a reflow oven controller. NOT a smoker oven controller. But if it DOES work! Yum-yum! Great food with little baby-sitting of the temperatures in the future!
Inspiring work! A few words of caution and some suggestions: if your product touches the main lines (120v / 240v) it must be UL certified. Otherwise you are exposing yourself and your company can be held liable for any fire incidents or electric shock etc. Alternatively I suggest you source a high power smart dimmer that is already UL certified and only send control commands to it....
I've thought about this too. Unfortunately you will still need plenty of certification for anything with mains, and every country is different. The US is a bit special in that you don't need to be licensed to fiddle with mains. A lot of other countries require tou to be a licensed electrician to work with LV. The best way around this is to keep it ELV. Find a solid DC supply and use that. Then on the toaster you need to go simpler - look into RV/Caravan, Truck, or Marine (boat) electric ovens that run from 12v or 24v is my suggestion. Though with ELV being up to 50v AC or 120v DC, if you can find something that runs nearer those limits you can cut way down on current.
Sorta like 3D printer heatbed controls - those are usually 12VDC/24VDC at a few hundred watts and use simpler power-MOSFET PWM control and thermistor temp sensing. Unfortunately this would mean your reflow area would probably be a small hotplate and not a big enough thing to process a whole 300mm^2-ish PCB panel. I'm building a transformer winding machine ATM, and instead of integrating mains PSUs I've stuck a barrel socket in the back and it runs on a 12V 5A wall wart because screw having to spend $125K and about six months of back-and-forth (as of 2018 when I last checked into this) getting RFI/EMI and mains safety certifications.
@@OddlyIncredible Yep, you can get off the shelf ovens that are up to about 500W from my quick googling. Unfortunately they are a lot more expensive than an ordinary toaster oven. That said, you could also try modifying a toaster oven that runs on 110v AC to run on 110V DC. Since the elements are resistive, it should be able to handle it without too much issue. The biggest problem is switching DC instead of AC with relays. Pwming that much power will be very inefficient, unless your duty cycle is very low - which is basically a variation on the classic bang-bang control.
I agree that for people like us used to buying most components for cents (or at most single-digit dollars) per part, it's definitely painful to consider buying something as expensive as an SSR for a project like this, but you're correct that a zero-crossing one dramatically simplifies your design in a whole bunch of ways. It's probably the way to go here.
You could use two MOSFETs to switch whole halfwaves with your zero crossing detection. But you would need pretty low Rdson to make a significant reduction in conduction losses. Like 10mΩ for 1W per FET losses. Basically designing an ZCD-SSR by yourself. This would be super neat and I should definetly at some point try it out, BUT for this project you will be far better of keeping the mains of the PCB and using SSRs designed and already certified(!) to handle the current required. My Power Electronic heart bleeds electrons typing this, but after some thought it seems the best compromise. Or you add more heat sinking capability, but thats a whole other can of worms with case isolation and so on.
Back-to-back power MOSFETs and a zero-crossing photodiode gate driver makes for a great and super-efficient SSR. Made a few of those for projects and they're solid performers.
@@OddlyIncredible For 60 Hz or 120 zcd's and two back to back igbt's or mosfets... (N ch.. g-d-s in " series " with another as s-d-g) ..its not necessary to have any zcd based triggering. This is reduced complexity at the cost of slightly higher switching losses..if one considers a switched pwm of around just 240 to 500 Hz or thereabouts...this can provide sufficiently good regulation..
@@analoghardwaretops3976 Yep, that would give pretty robust control. However, that having been said, if he's using a conventional heater element or quarts IR tube oven, he doesn't even need that because the total thermal hysteresis of these toaster/convection ovens is _crazy_ high - for example, I have my oven controller set up with a _10 second_ dwell time. All he'd really need to do is drive the heater at a 10s "on" time and 50% duty cycle to bring it to soak temp, hang there for a minute or so, go full power to reflow, then switch off and let the oven coast back down from there - no real need in this application to do waveform level control.
My T962A has been great. I did the mods and custom firmware. I soak it at 60 for a few minutes first, and try to avoid putting anything right at the edges of the tray.
@@gregstinkertown you might know this, even if you rework outside a heat bed is recommended. If you do this inside this oven you stress the components unnecessarily. It might work but for sure it's not good practice. Different solder has different melting points, this oven is definitely on the lower part of supporting various pastes. I would never put TG140 into a single sided heating oven only..
ZC detecting SSR is def a way to go. We use Omrons to PID the beds in VORON printers. I’ve had one PID a 1.5kW heater with a heat sink attached. Pro-tip: Set your PID cycle to whatever your mains frequency is to avoid flickering lights 😊
what do you mean by "Set your PID cycle to whatever your mains frequency is to avoid flickering lights". If you have a ZC detecting SSR, it's already switching at some integral multiple of the mains period, ie: at a fraction of the mains frequency.
@@Graham_Wideman In Klipper, the default PID frequency is order of magnitude lower than 60 or 50 Hz, so you end up switching on every Nth zero cross, and that makes your light flicker. The proper way to fix it is not having lights on the same circuit as outlets but that requires fixing your house wiring.
You can do the 0 crossing chopping with the schematic that you have, just use a to220 triac, 1kW of resistive load shouldn't be too hard. For this aplicacion seems like the better alternative, it's the same as the ss relay, cheaper and smaller as you can control the heat sink for the application, and pretty much any small one would do on a to220 or similar size package. Chopping mid wave is (or was) useful for light dimmers, as missing half cycle would make for bad flicker. Other option would be to actually use pwm, if it's much faster than mains the problem you mentioned wouldn't affect the power calculations, and the higher frequency would make for smaller filtering components, much more reasonable than a huge inductor you'd need for the triac half wave. (for pwm a transistor shorting the DC side of a full wave rectifier is used, choose your poison, mosfet or IGBTs are usually the options) Good luck with the project!!
10:24 Love how you're getting all excited about triacs... For us old folks that used to go out clubbing in the 70s and 80s, these are indelibly linked to lights flashing to the beat of the music.
50Hz mains have 20mS period, you don't need to be quicker for heating. For lighting it do matter, but not for heater. There are some MOC optotiracs whit built-in zero crossing. I like tht and alu heatsinks for power stuff. You get better cooling with no electrical tab isolation, but then you have a chunk of metal at mains live :) Check for example BTA16 and BTB16 triacs
The real problem is not the controller but the oven it self. It's too larger for the task with too much thermal mass given the limited power input due to mains limitations. I found the quarts tubes tended to burn things before reflow was completed. What is needed is a smaller lower mass type oven cavity that can heat and cool fast, allowing the PID to work correctly. One thought I had was to control the temp with some sort of servo driven air blend door, that could select between a reserve of heated air and cooling air. The heated for the hot air source could be stored in some preheated thermal mass that gets charged prior to starting the the cycle. You could heat charge the thermal mass over a period of time even allowing the use of low voltage input power eliminating the need to deal with mains voltages. Think of a 3D printer with low voltage heated bed and hot end.
You can use the fact that you are not switching a lightbulb to your advantage - the switching is what is heating - but there is absolutely no reason to switch it 100 times a second - probably 10 times a second is more than enough probably even just switch it after measurable temperature change.
The first strategy I can think of is getting an off-the-shelf solid state relay and using delta sigma modulation to determine which half waves to let through and which to block. Maybe running the modulator at half frequency so it works on full waves would be better for EMC, that needs testing.
Using sigma delta is essentially picking particular constants for a PID algorithm. It's not clear that those constants are necessarily optimum, since they don't incorporate any knowledge of the thermal momentum of the oven.
Triacs are usually paired with a DIAC and capacitor. They are capable of handling fairly high current considering they're frequently used on power tools.
PWMing is ... noisy from a stability POV and even more so at these power levels. Me, I'd approach it differently and focus on adjusting the AC voltage amplitude being delivered to the load to control power output. Thus, the AC output power waveform is stable and you're just modulating its amplitude.
Use triac and a zero crossing opto isolator and then slow your "pwm" down to a second or two. the power blowing your triac is due to you switching at full voltage rather than on the crossing. A heat sink might be a good idea too even with zero crossing.
IGBT and a photovoltaic isolator (works like a photocoupler, but instead of having a transistor output it provides voltage -- enough to power an IGBT) Zero crossing you can do in software
I still have one of your "Atlas Labs" business cards and i'm so happy to see how far you've taken it with Lumen and now this. You're doing good stuff and I can't wait to see the next product line.
Resistive heaters like that have a response time of more than a second, a normal relay or solid state relay switching on or off once a second would be more than fast enough, no reason to overcomplicate stuff.
Yeah, work with those triac in the past and it can defenately do what you are asking of it (when you select the right one) but it will always need cooling, you can add a big ass heat sink on the back or you can use a silicon thermal pad and let the heat sink in the metal inclosure but unfortunately the toaster casing seems to thin for that.
So we actually have a device that does exactly what you are designing- a reflow controller that you plug a toaster oven into. One thing that differs between it and your design is the rate at which it turns on and off. The one we have switches once every second at most. At that frequency you don’t exactly need to worry about the zero crossing issue. I can find out the name of the controller if you are interested!
@@Graham_Wideman the toaster we use has very little modification to it. We added a small layer of insulation and added a thermistor probe. The controller has an outlet you plug the toaster into. I'll ask what model the controller is
I'm not really sure how accurate you need the control to be but deciding to switch power every 0.25s or so would mean that you don't need to evaluate the phase of your ac. I don't really understand why you need the heater control to be at such a high frequency.
100% chopping at zero crossing is so much easier, less problematic and give you enough control for what you try to do. But also would be interesting to play with refractive bricks and Nichrome wire (heating element) to build a fully automated p&p + reflow system on rails
Hawes - Consider using a GFI outlet at your power strip just in case. Most mishaps in life make you stronger....except for electricity. It just kills you.
Well said
There should at the very least be a circuit breaker or fuse on the AC mains input on his controller.
Maybe afci as well
He's doing it the smart way by interrupting the power to the socket. And now that he is using a SSR he is insulating himself from liability that much more. He needs to have a big fat warning regarding using his device on appliances that contain their own microprocessor though. Your point about GFIs (if not already installed) is a good idea.
And as an electrician who started out his journey in 1975, when it comes to getting popped I'd have to say that at worst it only stung a whole hell of a lot. Nevertheless, I still wire things hot, it was how we did it back then.
GFCI/RCDs are mandatory and centrally installed in Europe since decades. I wonder if at least some US states require the same?
You have essentially created a very expensive light dimmer, which is why it buzzes, as you have learned. The triac is burning itself out because you are turning it on in the middle of the cycle, thus it needs to dissipate the switching current at that point and generates a huge amount of harmonics in doing so.
Your second idea is on point, and you can use a slightly larger triac (80% over rated) with a heat sink. You can still PWM, but you need to think about doing it over something like N cycles. If N = 100, and you want a 50% duty cycle, on for 50 half cycles and off for the next 50 cycles. The temp won’t vary that fast, so even using 256 over 2 seconds is perfectly fine. It will dissipate much less heat as well.
Note that chopping the input may mess up ovens that have timer electronics.
Fan ovens probably have better uniformity, but also may be unhappy with chopped mains
Yep, the oven I'm using for reflows (which is a convection oven, basically the bigger version of the Black and Decker Stephen shows) _DOES NOT_ like chopped AC, so what I'm doing is using a programmable industrial temp controller that has a configurable dwell time, which I have set to 10 seconds, and the oven is set to full temp and no shutoff time (timer disabled). The controller switches mains to the oven but does so in 10-second bursts instead of playing with the mains waveform.
Yes, adding a fan will help a ton with uniformity. When I made mine, the propagation of the heat made using faster than 0.5s pwm periods not necessary.
Yeah, if the project is to avoid ripping out any existing controller, the oven chosen has to be completely "mechanical", just switches, and a timer based on clockwork.
everything has microchip in it somewhere, even in the most simplest things there have manged get one in somewhere?
@@dh2032 Nah, you can still get entirely electromechanical toaster ovens - the one I'm using for my reflow setup is that way.
Bold choice to leave your board with an untested mains circuit plugged into the USB of your laptop haha. Great project, really interested in the result.
I feel like since heat is such a slow moving thing, you don't really need that fast of a PWM signal. You could probably get away with switching 2-4 times a second and still get super precise control. Most older microwaves literally have a 5 second duty cycle.
Yeah fr he's going wicked overkill with that control scheme imo. I once had to build a couple dirt cheap hotplates for a chemistry experiment and I just straight up hooked a coffee maker heating element to a ssr and controlled it with a free arduino pid library on a nano, using cheap thermistors I had lying around for feedback. The control there was setup for mechanical relays using on time 'windows.' They worked exactly how I wanted, I even ripped a processing sketch off the internet to graph the curves for tuning the pid loops, once tuned in they were rock solid.
Naw he's using an optocoupler and has huge creepage and clearance between mains side and ELV side. No problem leaving it connected to a laptop. Also, bang-bang control (like pulse density modulation) at zero-crossings is the right approach and doesn't force you to make reaction time so slow ("2-4 times a second")
@@NavinF It may be the right approach for some things, but an oven doesn't need that level of control. 2-4 times a second is plenty (by an order of magnitude, really) for an oven and a heating curve. PID control and you're good to go.
@@honkhonk165 Totally depends on the oven. You can do a whole reflow cycle in a few seconds. Lotta thermal mass to heat up
@@honkhonk165approach
That is is a wicked project Stephen! An "all in one box" to plug a toaster into was something I was looking at for a long time, but as you now see, the SSR is mandatory, and you'll likely find an SSR mounted to the main PCB still an issue (I did, I couldn't find anything that didn't still require active cooling on the PCB) which is why I landed on an external SSR requirement. They still get quite hot, and I always recommend folks don't even mount them inside their toaster without some thermal shielding as mounting a device that gets hot to a unit that is designed to generate a lot of heat.. um, danger danger Will Robinson! hahaha.
I look forward to seeing your next steps on this, and thanks for the nice shoutout at the start and the RMP link in the description!
Yeah, you need a zero-crossing SSR. There are a ton of open-source reflow oven controller projects to adapt. The AQ-A series of SSR's from Panasonic seem to be popular. a 15 amp, 75-250VAC model is about 20 bucks for onesies, 15 bucks in quantity. The heat sink can sometimes be more expensive than the part, they need beefy ones.
I agree. Get a name brand SSR that can handle the load. You'll pay a little more. The cheaper Chinese ones are hit and miss. some work, some don't.
Crydom ssrs are better
Yeah he needs to freeze his nuts. Your 17 subs TELL YOU TOO!!!!!!!!!!!
Nice. be careful around those exposed mains pads. Hope that quartz heater will hold up to rapid pulsed current.
The faster it switches, the better. After all, it normally runs on 120 pulses per second (60Hz is 2 full voltage cycles).
Im sure it's been said before in the comments, but I'm too lazy to scan. The hysteresis of a heating element (not to mention the oven as a whole) is _Long_ , so PID controlling at mains frequency might be overkill and harder than using bang-bang controller with an allowable temperature range. Just imagine trying to control your home HVAC with both methods and my comment will make make more sense.
I'm using a PID controller with a configurable dwell time set to 10 seconds because of this issue - the quartz IR tubes in these ovens don't do chopped AC reliably/predictably, and power cycling at any shorter a timeframe than about 10 seconds doesn't get you that sweet soak/bake curve you need for effective reflow.
@@OddlyIncredible I always thought the mercury switch thermostat was simple genius. The mass of the mercury swaying from one side to the other set the dwell.
Hey :)
There are optocouplers with diodes both ways on the inputs so you can detect zero-corssing without external diode bridge (4 diodes). Like LTV-814.
And yeah, it is better to regulate heaters by switching on / skipping a whole wave halves instead of chopping the wave (you can count the waves by an MCU).
That sound you heard (like a PWM sound as you called it) was the heater hit by a sharp front of the wave. The closer to the wave peak you chop it, the louder the sound.
I am driving a 400-Watt heater plate by BT138 (TO-220 package) without any heatsinks and it is not even warm...
A tip is to connect a light bulb in series when you test electronics with 230 or 120 volts for the first time. If there is a shortcut the light bulb will light up and no danger.
Yep it saved me from blowing many circuits. Once tripped my breaker of the whole house because of a shorted recitifer
@@309electronics5and that's why we have breakers 😅
When i was in college I liked to work on Hi-Fi gear. One day I was working on the power supply for an amp, and I had it all bare on my bench, plugged into mains running a test. I thought I was being safe, but while I was testing i was on my other bench using the computer. My buddy walks into the room, and as i turn to him he says "oh that's a beefy looking transformer" and he picks it up! I almost had a heart attack as i yelled at him to drop it.
It's cool that you built a box so you don't have to handle the mains side of the board, but it's not just you that you have to worry about. My buddy was fine, he managed to out it down without touching copper, but that was a lesson i wouldn't forget.
First reflow oven was a Hamilton Beach I picked up on the side of the road on trash day - full blast with multimeter TC to run up to 247-250 w/ GC10 paste made a lot of boards, just had to watch it close to turn off and open the door (wish I had one of the diy kits with the door opener puller to reduce chance of burning boards). Last year I went with a T-937M reflow oven in their 120v variant. The 220v variant has its issues but may have worked. Either case, running high current 120 or 220 isnt great for most users :/ so this project is definitely going to help the community! Best feature on the 937M for me is the exhaust and intake fan for unattended cooldown for sure. Duct for the exhaust is great to get fumes out, and the intake fan also runs for temp regulation and circulation fan. Oven in the unconditioned garage so 3000W heating keeps up in the winter time :)
Honestly, the best 'don't touch anything inside' might just be to pop the temperature control knob off and mount a servo with double-sided tape. You won't have to touch mains OR the oven itself, and switching ovens would be a matter of five minutes for moving the hardware over followed by recalibrating everything. (That being said, a simple relay to turn the oven on and off would also be handy.)
We did something like this, back in 1978, to control a 10 kW arc lamp. Switching full AC (no rectifier) OFF at the desired point causes less dissipation in the triac. (And MIGHT clean up the noise.) As I recall (it's been a few months, right?) we used a diac to drive the triac. Again, from really failing memory, I think we had an RC snubber across the output. (No need for it in our application, it was industrial - but you may want to use one.)
I rebuilt a controller board for my kitchen oven. I went through exactly the same process with zero crossing detector and triac, finding out i needed a ton of cooling and had loads of noise then ending up with SCR.
It turns out that you don't need so much fine control over the on / off period. Seconds of time are good enough and the tuning can be done with PID.
Reducing the amount of switching also lengthens the life span of the SCR.
SSRs are also just this triac circuit with a zero crossing controller chip(something like the MOC3041M), potted and on a heatsink.
There's nothing magical about them. But yes, I'd recommend that you buy a pre-made unit if you want to avoid the hassle of bringing mains onto your PCBs, which can complicate certification.
In general switching elements (and the things they power) also don't like to switch in and out in the middle of the sine wave. A better approach might be to apply power for a percentage of counts instead of a percentage of every count. Very easy change since you already have a microcontroller.
These were literally the next words you spoke after I wrote my comment lol.
Nothing wrong with the system he's using. Phase cut dimming has been around for decades and is a well tested and reliable method for controlling AC power to resistive loads. It's used in almost every household lighting dimmer ever made and is used for theatrical lighting into the 10s of kW range. The only real downside is it's noisy, but that's not that difficult to filter out with a basic LPF.
But the temperature of a thick heater follows switching cycles much more slowly than that of a thin filament in a lightbulb. In order to control the light intensity, you have to switch quickly and often. I think, for a heating element, I wouldn't use phase control either. 🤷♂@@JaenEngineering
@@JaenEngineering phase cut dimming is used when switching cycles will not achieve the desired result, largely lighting as you have pointed out. When you don't need such tight and constant control of the power using cyclic switching is much easier on the components as if done at zero cross they are switching no current, and instead of switching on every voltage cycle they only switch for whatever duty cycle is required. All switching elements have a limited number of times they can switch state whether mechanical or solid state. That number may be trillions of cycles but there is a limit. By using switching cycles the highest number of switching events will be at 50% power meaning that at WORST case cyclic switching would double the lifespan of the components before you even take into account the strain of switching under load. In order to switch under load the switching elements have to be more robust and more expensive. Yes you can use phase cut dimming, but if you already have everything needed to implement cyclic switching at zero cross, it saves you money in components and extends the life of your product, why would you not use it?
@@jurgenaddicks1634 So... more thermal inertia means slower change from switching. So why is phase control a problem with a thicker element?
I built my reflow oven some years ago using the Whizoo Controleo3 full kit. It was a bit expensive, but I liked their due to coming with everything but the oven for a full overkill build , separate control for each heating element, and auto door opening for cool down.
The benefit of doing the overkill build with full insulation and heat reflection is the oven being able to easily follow profiles. One of my projects had some components with some pretty strict reflow curves, and the oven was able to do it on the first try.
You absolutely want an oven with top and bottom quartz heating elements. The oven I went with was the BLACK+DECKER TO1313SBD for a whole $30 shipped.
I wouldn't reinvent the SCR.... Properly rated, commercial available SCRs are pretty cheap from Digikey or Mouser.
Well done, Stephen! Really enjoyed watching this video where you shared your passion and knowledge with us.❤️
Many years ago I did a lot of design work with triacs controlling halogen projector lamps. You really have to be careful with dv/dt across the triac as well as conductive noise suppression to keep the FCC happy. Using an SSR that triggers 'on' near the zero crossing as well as 'off' at zero crossing is the best way to go IMHO.
I built a Controleo3 reflow oven using Whizoo's kit and a Black & Decker toaster oven. It works really well, I've made several boards in it. I just got my Lumen PNP delivered a couple weeks ago and got it all configured and calibrated, successfully ran the Functional Test Placement board. Can't wait to get some of my own boards placed with the Lumen and reflowed in the oven. Up until now I've been manually placing parts with an AirPick and a microscope, so glad I'm not going to have to do that anymore.
i ve build such a control for an oven, a ssr is the way to go, they are super simple to use. it is a "wave packet control" - if this is a good translation from the german word "Wellenpacketsteuerung" . the needed pid controller is super simple to find good values for i and d because the system is very sluggish. their allhough a lot of ready to use and highly advanced open source projects :)
Having made a few reflow ovens like this for myself I used an ssr and didn’t worry about cycle switching as you can go pretty slow. I cycle twice a second and the pid works like a charm the heating curves are close to perfect. Good luck with your project.
When i made mine i found that very short pwm periods work just as well as many second long ones because of the masive thermal mass you are heating up. You also cant just have one temperature probe as the temperature is not even and takes a while to propagate to the area of the board. The heating elements also emits large amounts of IR which makes non-reflective objects heat faster such as fr-4.
Good work so far. Since you will have to redesign, might want to add a power converter so the MPU is on when it is plugged in to power. The switching may be better suited for Solid State relays as they do come in higher amp capabilities and your zero crossing circuit would still be able to detect when to turn on. The only issue I see with that option for switching is the cut on and off timing which may be able to be compensated for in programming, possibly adding a thermal sensor to assist in " auto programming " the timing for set temperatures and change the timing of the relay.
@steven, great work! One comment on the mains circuits, you should use multiple resistors in series on the high voltage side. You'll see in the data sheets that they have working voltages typically maxing out at about 200v. Keep in mind that 110v is RMS so peak is higher, plus you need to allow for surge voltages too. If you ever go down the rabbit hole of certification with this you'll be forced to learn all of this in detail, like I had to!
Ps it was great meeting you in Munich 😊
Very cool project. Please be careful with these open AC flats. I have also converted a toaster oven. Experience has shown that these things are so slow that it is actually enough to switch the power on/off with an SSR. So I was able to solve the whole 230V switching with an SSR, which I switch with a FET. These ovens are so slow that you have to switch on/off relatively slowly to keep the temperature +/- approximately. Thus, the active line control is not necessary at all, but an on / off is enough in this case. Am curious about the next video.
I have had few problems with making small quantities on the stock T962 running lead free profile 3 on leaded solder. No mods at all. Just lift the board off the base with a couple shims to avoid cold spots. But I bought a 6-zone conveyorized oven for actual small scale production. Use time proportioning with a 10-20 second cycle time and zero-crossing SSR (isolation provided by the SSR so everything on your board is low voltage). Heat sink fairly large (there are design curves from the heat sink manufacturers) for 10W or add a noisy fan to save money, weight and size. Beware that a short to the mains can not only electrocute you but could blow the **** out of your computer via the USB connection to earth. USB isolators are available if you are belt-and-suspenders cautious.
Also, you can use the zero crossing detection to turn relays on/off without arcing, to do this you need to determine the on-delay and off-delay of the relays. Then you use the zero crossing input signal to start a timer, when the timer elapses the output channel of the timer turns on the relay just before the start of the next zero crossing event, the relay then should activate at 0V which will then not cause arcing on the contacts. to turn off the relay do the same thing, but adjust the timer to take account how long the relay takes to turn off. turning on a relay is solenoid activated, turning off a relay is usually via a return spring, both on/off actions take different amount of time to complete.
I use zero crossing activated relays to control bar heater elements. and i use an SSR to control power to a 2kw halogen lamp. the bar heaters only come on when the different between the target temperature and actual temperature is large, this avoids excessive switching. the halogen lamp reacts MUCH quicker, so an SSR is used to control the halogent lamp.
combo microwave convection ovens are the best as they have turntables and convection fans and are the most easy to retofit, but not as simple to control...
also, microwaves have excellent interlocking safety facilities (switches) which can be re-purposed to automatically kill the power to heater elements if the door is opened.
the biggest issue you'll have with using a toaster oven is the downward slope of the reflow profile, i've seen people use aircon butterfly valves and an inline blower to get the oven to cool down quickly. toaster ovens are also terrible at keeping the heat inside them. i used rockwool which is use for fireplace insulation around mine, made a huuge difference.
When developing a product with mains, it helps to cover the high voltage section of your board with a insulator, taping a thick piece of acrylic for instance. I also recommend checking if you're plugged into a circuit with an RCD/GFCI. It prevents you from getting zapped when touching the phase (though not when touching phase and neutral)
Good idea, but acrylic burns like wood. A better solution is to use polycarbonate (Lexan), it is flame retardant and UL approves it.
*Yes reflow should be solved. And there should be great SMT protoboards also. And pick and place with a robot arm, a vac to pick up the components, and OCR to read the vals, and a shaker to shake the components upright should also be a community project. Post a video comparing that reflow method to the sand in a pan on a stove method. Thanks in advance.*
I give you an A+ on your hand soldering. Excellent.
There is a simple solution to even out the hot/cold areas in the T-962 oven. What you need is a fan to agitate the air inside. This oven is a bit cramped, but if you drill a hole from the top, run a shaft through it and attach a DC motor controlled by PWM, or a stepper, you can control the rotation speed. Inside the oven what you need it a propeller, something basic and made of a metal that can withstand the max temperature temperature the oven can reach. I like to use a minimum of 25% above the rated value as a safety factor. A twisted strip of metal with a hole in the center should do, you don't need something fancy. The bearings need to be able to handle the temperature as well, so make sure you get some that will be able to handle the heat. This won't require a high RPM, in fact, very high RPM's may generate so much draft that it could displace the components from their desired locations and it won't make your oven as homogeneous as it can be.
When you implement this the oven's characteristics will change, so for very delicate processes, it could be a challenge to get it just right.
Firmware changes won't do much. You can get the oven to be homogeneous, even if you don't make any mods, but for that you need to lower the climb rate so much that the heating curves will be dramatically different, and could render the process useless for your desired purposes. The reason why you can't get it to perform any better has to do with the heating elements and how they are placed in the oven. If you had 4 of them, one in each corner, and were able to control them individually, you would have much better results. This also means 4 thermocouples at the very least, ideally 6. One close to each heating element, one for the PCB and one for the air inside the oven. The air thermocouple will govern the process temperatura (process value), the PCB thermocouple will be your failsafe (it will cut off the heating if it overshoots the setpoint), and the 4 in the corners would determine which heating elements need to provide more or less heat to even out the temperatures in order to keep the oven under tight control.
In theory you can pass on the fan with this type of control. It will make the oven more efficient and quieter, and your process will benefit greatly. However, you will need to spend more money in heating elements, implement 4 PID's and do a bit of head scratching and possibly melt your retinas a bit learning how this works. It's up to you to decide which solution you want. One is messy, complex and expensive, but as a tighter control and is more gratifying, while the other is way simpler and it could just be enough for you.
If you think about it, it's probably better to just get an old microwave oven chassis and build it from scratch. That way you can implement everything, and it's a neat project. Plus, you will get a lot more clearance, the T962 is quite cramped. Not a bad desktop solution, just not good enough for SMD soldering. You get the PCB's charred in the center and the solder may not even melt at the edges.The control is limited to one thermocouple which tracks the process value, and the heating elements (they are 2 if memory doesn't fail) seem to be connected in parallel, meaning 1 output only. A bit of a caveman approach, if you ask me. Not hot enough? Club it a few more times, it will get there.
Another solution could be a multi-zone hotplate. You can get a steel plate, divide the plate in 9 zones like in a tic tac toe board, and control each of them individually. Whatever suits you best. The control can be done using MCU's like the arduino, raspberry pie or ESP32, they are not that difficult, and languages such as microPython are a good option for a slow process like temperature control.
If you want to go the extra mile and be able to provide rapid cooling, you can always run a copper tube welded to the metal plate and flowing from and to a tank. To avoid getting the tank too hot, you may require a mini cooling tower. It's visually fantastic and very efficient, but not a good choice for a lab as the resulting condensation can cause mold and it's also not good for electrical/electronic devices that cohabitate in the same room. You will also need to compensate for the lost water.
This is really over the top, I just thought that it would be interesting to add for the sake of providing a bit more info, and people can look it up if they so wish. Keep having fun kid, you are doing a nice work. Rolling up your sleeves and having a go at what you want is a great example for others to follow. As long as people play it safe, the worst thing that can happen is a failure. If they don't, well then they can burn down the house and the entire family will have to share the basement. Not my idea of fun.
I basically built what you're up to but for incubating eggs with a standard light bulb instead of reflowing solder paste. The right PID coefficients have been the most tricky part here. For obvious reasons I could not tolerate any overshoot whatsoever. I need however a target temperature as smooth as can be in a range of less than 0.2°C. To top things up the ramp up should be quite fast. So, 5 minutes max to 35.7°C, no overshoot, keep it there, little wiggle room of E~0.2°C. And, yes, make it robust against disturbances, no big over- or undershoots there either. Hard to hit those targets combined. Up to now, there's always been tradeoffs, haven't been able to meet all criteria at once, yet. Top prio is E and overshoot. Undershoot and ramp-up is least critical. With reflow ovens it's maybe less of an ordeal. Keep us in the loop, I'm curious about your reflow profile confidence level.
What temperature sensor arrangement are you using? Type of sensor, placement and so on? And does your chamber have a method to circulate the air (like a fan)?
@@Graham_Wideman K-type (for the PID loop) and an SI7021-type sensor (via esp32, tasmota) for verification. Both between and at the top of the eggs. Third, 7021-type, sensor at incubator ceiling. Fan to distribute heat/prevent hot spots. Incubator is isolated by 4cm styro. No window but an esp32-cam. Dimmable (PWM) LED strip (still flickers a bit at the set frequency). Fan does not do it's job as intended yet. Still very distinct heat strata (astonishingly so). WIP.
Very nice to see you doing reseach and development for a new product! BTW I'm guessing that Joule may be a provisional name, but be aware that there's a sous vide thing called Joule as well (Interesting how they got a trademark on a unit)
you can trademark things that aren't related to the actual thing (eg Apple), because this gives an assumed distinction
@@gedr7664 excellent example! thank you
Not only that, but getting a trademark and having that trademark hold up when contested are two _entirely_ different things. Trademarking a measurement unit name (especially one that's an established and clearly defined standard unit) is likely to not be enforceable in any practical sense, regardless of any assumptions of distinction from a marketing perspective. Nobody will confuse a fruit for a cellphone, but a glorified water heater and a unit of measurement for electrical power have too much overlap in too narrow a cognitive space.
(Disclaimer: IANAL but I've had to deal with IP rights, and there's a LOT of interesting shenanigans involving trademarks.)
@@OddlyIncredible Apple trademarjed Newton, then there's Tesla, quite a few companies named Henry, and no doubt more, considering many units are named after people, as is Joule. I don't think that having your would-be trade name already in use for a unit really presents much of an obstacle.
MOC3031 is a zero crossing capable driver, you were really close with the MOC3022, I might be mistaken, but part of the issue could be if you are PWM'ing anywhere in the middle of the sine waves, you are snapping through a large inefficient linear region to get saturated each time, just like with transistors it's not somewhere you want to be for long at all, pretty sure that's another reason zero crossing is often more preferred. I made a 555 drive a MOC3031 for a spot welder controller I designed. With the triacs, pay attention to the ratings and do the dissipation calculations(as you've already realized), also some are available in plastic isolated packages, others will possibly make a heatsink go live.
The way I implemented something similar for a soldering iron without an expensive SSR and ruining the power factor+EMC is to use back to back FETs with a photovoltaic gate driver. To convert the PID controller's output to pulses, I used a RNG and compared with the PID output every interrupt to decide whether to let the upcoming pulse through or not. On average this works quite well since you're doing it at 100-120Hz and the time constant of everything else is in the 10s of seconds
Also definitely make sure you're using an optocoupler to detect zero crossings, having a fully isolated controller is a really good idea for safety especially if you want a human pressing buttons on that board. Make sure you respect creepage and clearance requirements for your voltage level too across the isolation barrier. AC optocouplers are the simplest implementation for this imo
Stephen: My top suggestion is this: Assuming you want to do better than just wafting hot air over a board until the solder melts, your task is to get the oven to perform a series of heating steps to get the PCB, specifically the solder, pads and component leads, to follow a thermal profile, for example that published in component datasheets. Many oven mod kits attempt to get the temperature feedback from a thermocouple just floating in the air, perhaps above the board(s). But it turns out that's a pretty poor approximation to what the board is experiencing. So at some point you will need some thermocouples mounted to PCBs, and thermally coupled to pads (maybe with high-temp glue), that you can place in the oven to test actual performance, including in different locations in the oven. You can even do what I ended up doing -- use a small PCB with thermocouple attached to act as the feedback sensor, and it sits on the tray next to the board(s) being reflowed.
This is especially useful for lead-free solder where the margin between good soldering temperature and burning the board is relatively less than for tin-lead solder.
Several thermocouples-on-PCB is the approach used for profiling commercial conveyor reflow ovens. (These are usually connected to a data logger inside an insulated box that accompanies the test board through the oven.)
Anyhow, you're going to need a few thermocouples, and it's useful to get something like an inexpensive 4-channel handheld thermocouple "meter" with logging. I got a slightly upmarket PerfectPrime TC0520, which can also output data to PC for plotting. Pretty convenient for a variety of tasks. We even used it with 4 long-lead thermocouples attached to a PCB to do some basic profiling on an industrial conveyor reflow oven we work with.
Anyhow, bottom line is that rather than puzzle over vaguely better or worse soldered boards, you'll want some ground truth, which you will get closer to with thermocouples mounted to PCBs.
I've worked on similar circuits in the past (mains appliances controlled by a microcontroller for home automation, though at lower currents and without PWM), so I have a few notes/suggestions:
You explained phase angle control (/phase cutting), but also mentioned that full-wave switching might be better for EMI. As far as I know, there are also power factor specifications, at least for larger loads, which may require lots of filtering if using phase cutting. Generally, switching near zero-crossings is more efficient anyway (for resistive loads), so this seems like an easy choice, also given that the oven's response time is dozens of seconds at least, so you likely don't need the faster/more granular control of phase cutting.
Most solid-state relays I'm aware of also use triacs/thyristors for switching, and hence share the ~1V drop and high power dissipation. Their main advantage is ease of use and low part count, but since you're making a custom PCB anyway, it's not much extra effort to implement that functionality yourself (or get a zero-crossing controller and a separate triac). There are plenty of triacs available that are beefier than the one you used, as the manufacturers know they need lots of heatsinking, though you'll likely need to go THT (maybe TO-220 or TO-3) to actually dissipate dozens of watts.
Alternatively, you can use high-voltage MOSFETs for switching AC, either using two of them back-to-back or by shunting the DC outputs of a full-bridge rectifier. These offer generally lower voltage drops and hence power dissipation, but they are likely less reliable (due to transient overvoltages, see also below) and are harder to drive (no automatic shutoff on zero crossing, needs correct polarity of gate voltage), so many people seem to recommend against this. If the ovens can run on DC though, there also the option to rectify and zero-voltage switch, as used by a soldering station at th-cam.com/video/erKCA71q7cg/w-d-xo.html : you can rectify but not buffer mains, so you get unidirectional ~110V RMS, but with regular zero-crossings, so a single MOSFET (with low on-resistance and hence continuous losses) can be used. Then by switching near those crossings ("zero-voltage switching") at low voltage and current, you can get by with very low switching losses.
Finally, before this becomes an actual product, I'd recommend you carefully test its EMI *resistance*, as having mains on the same PCB as a microcontroller can be interference galore. I'm currently planning another revision of my project, because I've noticed that switching off fluorescent lights in an adjacent room sometimes causes signal glitches in the low-voltage section, and this will likely only get worse with the larger currents and frequent switching you're planning to implement.
(Disclaimer: While I'm studying EE, this is not my speciality, and I'm by no means an expert. Please take all the above with a grain of salt)
Indeed running a high wattage load at those kinds of power factors are quite "rude" to the utility company...
With regard to regular high voltage mosfets the i^2*r quickly (I has comparing @16A and IGBTs not triacs) became as bad as the IGBTs, good (expensive) GaN and SiC did however beat the IGBTs.
I'm inclined to believe that here a triac is probably just fine unless Stephen really wants an efficient controller.
Diode rectification comes with its own "horrible" voltage drop.
@@TalpaDK Right, my low-voltage habits are showing - of course IGBTs are an option at mains voltages, and I did forget the bridge rectifier voltage drop.
Regarding I^2*R losses, the video mentions 10A, so I estimated that
@@hadinossanosam4459 reflow ovens are often >= 1300W so I would expect a bit more than 10A for a US installation, but yes a EPC2304ENGRT with its Ron of ~3.1mOhm would be way easier to cool even @ 13A (523mW * 2). Is also not your garden variant Si mosfet but actually one of them shiny GaN ones (and a bit on the "expensive" side).
Stephen would probably have to spring for something even fancier, as he would probably also want to target 230V countries which would push the voltage requirement up to 325V+margins.
I also hope that he is considering adding a MOV and a fuse to his setup for some basic transient protection regardless of his choice of switching device.
@@TalpaDK..fuse YES ...mov...not really necessary with the device in series with such a resistive load .
Did very similar project as my bachelor thesis (using hot plate additionally cooled by air pump for air mattress).
I was surprised how small triac you used couse remember how much mine got really hot - few minutes later and now I see you also faced that problem :)
Good job though!
Actually, there is a product that satisfies all of your requirements - the BetaLayout Reflow Controller V3 Pro. I am working with it for years now without issues. Just a little fiddly to place the temperature sensor, but the results always turn out beautifully.
Using the S.S.R , both types are available...Phase control type and Z.C.D. type...with the zcd it's controlled as how you
explained .. that's known as
" burst firing mode"...
Phase control can give tight temp. control..with a good ckt. design...and input voltage fluctuations and " sine to linear ramp" linearisation circuit can be implemented to give a linear temp.rise over the entire temperature range....
for best results with phase control, one can initialize a burst mode followed with phase control for temperature tracking accuracy.
Concur on switching to an SSR. It's still a proportional control system where the power to the heaters is proportional to the multi-cycle on-off time. Actually, this makes more sense as the thermal reaction time of the heating elements is actually quite slow as you've shown by the full-on ramping time experiments at the beginning. In effect, you're modulating reduced heating rates where that test curve is the maximum achievable heating rate given the existing heating elements, oven insulation, oven thermal mass and thermal leakage through the door edges of the unmodified oven.
12:38 and here is where you probably should have covered the top side of the board around the AC section, because all that exposed area is ripe for finger touching while its live.
One more thought:
There are solutions to your problem out there. Someone else mentioned a PID controller, which is true (but I don't know if they handle the kind of curves you're trying to follow). What I thought of, is heat treating ovens for knifemaking and metal working. Different temp ranges, but they also handle fairly complex heating curves that are user-programmable.
Looking forward to seeing how this project progresses. Always interesting to see different approaches, though my gut tells me that the thermal lag of the heating elements is so long, compared to that of your control loop, that you are not going to see any better control than your traditional bang-bang controller that is typically used. (on and off for long periods of many cycles) Certainly sub-cycle control, with the triac won't gain you anything here. Good luck with it, interested to see how it goes.
hum...a few thought about the slow reaction curve
i assuming the heating elements are some kind of threded/induction elements
...when on a magnetic flow builds up (magnetic flux)
..when thay are rapidly turned off...the element(inductor) will react by reversing the flow of electrons.. aka puch back effect, and since thay cant flow back trough the turned off triac ther will be residual energy left when the next switch on comes...
solution:
...u add an inverter to the triac gate signal, connected to a shunt triac and a diode connected to a large capacitor with a discharge resistor to slowly drain it safely (no short)
...this should increase the responsivness of the owan itself...
...to test, u should check with a lage capacitor and no way of discharge it how what voltage u get after some time so u know of much excess puchback u need to dispose...
..u can checkup puchback or flyback effect to understand this better...
I suggest you take a look at driving soldering irons or other open source reflow oven projects, Soldering irons are driven by 24V (AC OR DC) typically but usually driven by 2 SSR's or solid-state relays. I would say this is likely to be the best approach.
Keep in mind resistive heaters don't care whether you are driving them with AC or DC so you could potentionally convert it to use DC and use MOSFET's. I would just do some research if the heating element used can actually work on DC.
I've done some of this in the past. Maybe not applicable in this instance but, when driving inductive loads, you should switch complete mains cycles (I divided the zero-crossing pulses by 2 and worked off one edge).
Also, rather than n cycles on and n cycles off you can provide small bursts of several cycles on to match the dynamics of the halogen heater elements.
You are making something really similar to the Beta Layout Home-Office SMD Kit that Dave Jones made a video about almost 10 years ago. We've had that in the office for several years now and its awesome ! Best way to get a cheap reflow over I've found so far.
As far as mains control on the heater goes - the best cheap way i know of controlling those currents is with a beefier TRIAC, in like TO-220 or bigger case plus heatsink coupled with a zero-cross triac driver. Something like the MOC3043SM, that is extremely popular. Its a way cheaper solution compared to a solid state relay. The way i like to use this combo with a regulator (PID or other) is by just turning the cycles in groups of 20. 1 cycle is 5% power, 2 is 10%, etc, etc. If i want finer control i can get up to groups of 40 cycles. Since heaters are VERY slow, the end result is the same as if I'd used per-wave chopping.
I picked 20 since I'm in the EU, we have 50Hz mains here so 20 cycles is 0.5S. 40 is 1S.
I'd like to note that the tab on many triacs is live. You must provide proper ground isolation because the heat sink can become line potential. You can get isolated tab triacs, so be observant. Be careful of unproven sources for SSR's. There are literally billions of fakes out there. Say they're rated for X amps, when investigated have internals rated for 1/3 of requirement.
An SSR will work great. We used one with an off the shelf PID controller to control a crock pot a few years ago. You still need a heat sink, and you need to wire the center screw post to ground (a lot of guides don't tell you that). But seriously, you don't need a micro controller or anything for this, unless you want a fancy ramp or a timer or something. If you just want to put something at an exact temperature, an off the shelf PID/SSR kit will work fine.
A fancy ramp is exactly what Stephen wants to do, as specified in component datasheets. The general principle is to ramp the board (and components) up to a plateau below solder-melting temperature, let it sit there briefly so that all of board and components get close to that temperature, then "sprint" up to a little over melting temperature, melt for a limited time, then a controlled descent back to ambient. The idea is for the time above solder melting temperature should be long enough but not too long. That much precision is not necessary for making a few hobby boards, but it is important for producing larger quantities where you're confident that the components have accumulated minimum damage from the soldering process.
@@Graham_Widemanto go that far , calls for a contoured temperature profile according to the process possible in this small oven.. but here everything is manually labour intensive..hence a bit of overkill.
@@Graham_Wideman then he can do it with an SSR, a mosfet, and an Arduino. You can prototype this with a RAMPS 1.x board, no soldering required. Or a solderless breadboard would work too. It's simple to do.
@@BrainSlugs83 Your suggestion lacks a graphical display, which of course could be added. But for Stephen, it's probably just as quick to knock out a PCB as it is to mess around with breadboards, or getting set up for Arduino and RAMPS etc. Instead, he gets a system with an MCU he's familiar with and already has a suite of libraries for. So sure, all sorts of things can be done with Arduino, doesn't mean that path is simplest for everyone.
Basically it’s a phase cut dimmer your making, I could suggest an FL5160 it is controlled by 0-5v signal, and uses 2 IGBT which you would need to mount to a heat sink.
I think an SSR is the best solution to this, just be sure to get a genuine one from a reputable manufacturer. There's lots of cheap clones out there if you don't buy from the usual suppliers. Also, keep in mind that SSR's tend to fail CLOSED if/when they do, so some form of thermal runaway protection should be implemented if this is to be a product.
Loving the progress so far though, this is an awesome project!
Triacs to control 15 amps come in TO3P or TO247 packages. Some TO220 triacs can take the load but you must have the heat sink to back it up. Also check the ratings on the IEC power connector and cords. Many of them are not rated for 15 amps. 14AWG (insert metric equivalent here) wire minimum. Elsewhere in the world, where ~240v power is standard, current is half.
You need to pick up an isolation transformer. This way if you ground yourself while grabbing high voltage you don't make a circuit. Also, I have built 1500W heaters with crydon SSR's that work great. I also recommend adding a thermal fuse to prevent the oven from cooking itself if something goes awry on the control board. I have taken the heater product through FCC and CSA ISO601010 testing if you need help or advice.
You can always run multiple Triacs in parallel, spreading out the current across multiple components. That doesn't help with your EMC problem but it would prevent the parts from overheating.
Equal current sharing in 2 or more triacs is difficult to achieve....a better option is two anti- parallel SCR's each handling one half of the waveform...control ,even from an opto coupled triac is easy with just a polarity steering diode to each scr gate...
SCR's are much more robust and have better & higher noise immunity than triacs...
This scr config. has been a standard in high current applications.
Years ago there was a product called the Reflowster. It was a controller that turned a toaster oven into a reflow oven with zero modification required. I still have one.
Came here to mention Reflowster and saw your comment. I haven’t used mine for several years, but it worked great!
@@doogulass yeh. shame they disappeared...i think they came a little too early... there is more and more demand for it now.
Perhaps they ran into issues with people using newer, more expensive toaster ovens that have electronic controls built-in. The Reflowster really needs a dead-simple toaster oven with physical thermostat that won't be screwed up by power cycling.
You can't go without Zero-crossing nowadays. In you case, I would go with MOC3083M + BTA16-800, through hole so you can find cheap heatsinks, and eventually a good old relay to switch the load on for extended time like when you are heating full power. Heat sink may not even be needed with the relay.
Definitely not an enginner but just a thought, if you just make sure the time it is on for or off for is a multiple of 1/60 of a second (or 1/50 here in the UK), you will always get a consistent amount of power. I feel like something like this is operating over the sort of timescale where being on for 5s then off for 5s would be fine (I think I saw someone in the comments already saying this is about the speed that microwaves run). I also think if the toaster ovens have anything else going on, 'dimming' the current might cause problems that switching for longer sections might not. Also I like the idea of putting an IEC connector on this. I think that is one of the more universal connectors common here in europe as well as the US. I love the idea though, I'd like to get an oven for the lab I work in but there is no way I'm getting a reflow master past the electrical inspections that we have to pass (as much as I know they are safe and solid). Whenever I have students that want to work with mains, I point them to those off the shelf smart plugs and have them write some network code to control it rather than doing anything with the wires... or glue a servo to the switch on a power strip! Anyhow just some thoughts hope something is useful here! Keep up the good work!
You should have still have pretty good granularity with the zero crossing SSR approach. Since you're in the US, you've got a zero crossing 120 times a second. If working over a period of 128 zero crossings, that's 7 bits of resolution with an update rate of once every 1.067 seconds. You just need to have a counter register increment by 1 every zero crossing and reset at 127, have a compare that increments by powers of two. And for each bit in your desired output power value, you either turn it on or of for a length of time that is equal to the value of that bit. So if you're trying to output a power 105 out of 127: 105 is 1101001, so you'd be on for 1 zero crossing, off for 2, off for 4, on for 8, off for 16, on for 32 and on for 64. If you wanted the full 8 bits of resolution, count up to 255 instead and your update rate becomes 2.13 seconds. With the right PID tuning, you could get pretty accurate temp control.
Just a suggestion from a UX standpoint, I would ditch the buttons and use an encoder for the final product. Just like printers do it, I bet because of 3d printers you could get really good deals on 16x2 lcds and encoders from china. They are also more accessible parts for anyone building the device.
This video reminded me of the Keith Wakeham's Maelstrom fan project. He made a device to switch the mains voltage of his cooling fan for indoor cycling so that it would blow harder as your heartrate went up or pushed more power. He has a series of videos on the development on TH-cam. Looking back he used a RobotDYN PWM AC light dimmer module with zero crossing detection just like you designed but the triac on that board has a fairly substantial heat sink. The prototype was very good but the project was cancelled in beta due to the difficulties and cost of getting the device certified. In his final update videos he talks about how the certification would cost tens of thousands of dollars. Good luck, I really hope this works out but I can see why most oven controllers are more diy.
Quartz and resistive elements have a curve, so it might be worth just expanding the duty cycle to a few seconds of ontime, instead of rapid pulsing. Marlin etc have been doing PID tuning for years, I wonder if that code might run better than rapid switching.
I have a background with theatrical dimmers for tungsten lights. It isn't uncommon to see triac based dimmers in the 1KW - 2KW range, but IGBT based dimmers are becoming more common. The main difference being which half of the wave is cut. IGBTs you want to turn on at zero crossing to limit the current flow at that point, but turn off in the middle of the wave. The triacs work as you described and turn off only at zero crossing. A good design incorporating a triac or IGBT shouldn't produce that much electrical noise when controlling a purely resistive load, and in the environments where theatrical dimmers are used, there are typically many thousands of watts of dimmed lights in use in the same building and powered from the same pad mount transformers (those big green deals the power company installs with underground 7200v running to your property). We never had any noise issues that could be traced back to our dimmers with nearly 100KW total load, even with over a hundred audio channels - more than a dozen of which were UHF wireless - and multiple guitars (which are great at picking up stray electrical noise).
From the diagram it looks like you're switching the neutral and passing the hot straight through? In DC I know it's standard to switch on the ground, but afaik with mains you really want to be switching the hot and keeping the neutral unbroken to minimize likelihood of someone getting shocked
Ok, that was hard to see until the end. so mutch stuff I wanted to throw in. first things first: to use a triac you need a snubber (cap in series) so the triac can't be hold open and a filter for the triac (so he switches smove). Than, like with solid state relays (they work with triacs internally) you need some kind of falt protection, they normally stay open when failing. Now to the emi... that's a real problem. I recommend the Würth Elektronik design kits for emi noise reduction (but I think there will be some american stuff to, I'm german) BUT you will need a spectrum analyzer to know where to go for.
after all triacs a great for mains control in high precision applications, I use them in the development environmental chambers to control heat and humidity.
PS: use another triac, one that does not have mains on its backplate :)
Also look into the practical heat output cycle on the quartz - there is a lower practical limit to how short of a time slice results in practical heat. I've just defaulted to 15 seconds to be perfectly safe and accurate enough. Maybe after a ramp up to heat soak you can go shorter.
Another approach I have used to control a heating element is the opposite of the leading edge triac control, trailing edge control. There's a really great trailing edge controller ic 'FL5150' that drives a pair of MOSFETs with just a plain DC control signal. Although I suspect this would have the same thermal problems as the triac at those high currents.
Trailing edge ..so one always turns the switch on at zero voltage..so no inrush/ turn on spikes..and phase controlled turn off can be a soft turn off type so the EMI/ EMC compliance can be met easily.
Interesting idea to control the temperature. Curious to see how it performs. I went a different route with mine and used an oven with multiple heating elements. I have a short video on my channel with a link to more details.
I can control the ramp up speed by only turning on the top two for example. I find this is good enough control, but I guess one down side of this approach is cold/hot spots. Mine is fully insulated, but this meant I wasn't able to cool the oven fast enough if I overshoot so I added an exhaust fan. From my understanding, this isn't recommended most professional reflow specs, but I'm not doing production work so these trade offs were okay for me. Anyway, looking forward to the next video.
Zero cross, run a pid loop that can only output at 5hz. Buy 2 toaster ovens and put both heating elements in one unit. Put a blowfan to counteract overshoot acne cool at the end.
I built one and had to do the above to make it perfect
Awesome design, but I saw that small triac frying a mile away, considering the current that oven draws. Definitely look at a SSR rated well above the current rating of your oven. If you build this for sale, and want a UL listing, that's very admirable. So over engineer it and put warnings everywhere. Also, put thermal fuses on your SSR to avoid fires. Your approach to pulse the AC sinewave coming in from the mains is quite interresting. I have been an electronics technician for over 40 years trained by the US Navy. We have 100 KW radio transmitters that autotune. Meaning, the tank circuits of the different stages of amplification tune automatically. They use the AC sine wave much like you did here, except we also look at the polarity of the AC sine wave coming in to drive a DC motor. Polarity sensing lets us move the DC motor in the correct direction to change either the inductance of a coil or a variable capacitor. I AM Very interested in following this project as I am doing the same thing you are... controlling the temperature of an oven. I am very good with hardware but I cannot code at all. I am using a 32 bit ESP32 with WiFi and BlueTooth to accurately control the temperature of a cheap, commercially available meat smoker. I have subscribed to your channel just to see how you progress with this project. You might consider making a Meat Smoker controller with this same circuit, but more robust and able to handle the heater element currents but please make sure you include PID. Also, a wifi or bluetooth connection to a mobile phone app. A meat smoker controller might be more commercially viable for you to make. Much more demand than a flow oven. Do both. I'll buy your meat smoker controller!
I have been talking with Seon of Unexpected Maker, who is reference in this video. I have just purchased his Reflow Oven Pro, or whatever it's called. He seriously warns that using his device to control my meat smoker oven is outside of it's use-case. I must insulate my meat smoker to allow that controller to have a chance to work. I will make a video of how I incase my $150 meat smoker with ceramic fiber "KAOWOOL" and an outside frame of some sort to insulate my smoker. Then use Seon's Reflow oven Pro controller to make it it work, if I can make it work. Not a fault on Seon's controller if it doesn't work. It's a reflow oven controller. NOT a smoker oven controller. But if it DOES work! Yum-yum! Great food with little baby-sitting of the temperatures in the future!
Inspiring work! A few words of caution and some suggestions: if your product touches the main lines (120v / 240v) it must be UL certified. Otherwise you are exposing yourself and your company can be held liable for any fire incidents or electric shock etc. Alternatively I suggest you source a high power smart dimmer that is already UL certified and only send control commands to it....
The problem is that UL certification is lengthy and expensive. A DIY kit without the power parts can probably avoid the headaches
I've thought about this too. Unfortunately you will still need plenty of certification for anything with mains, and every country is different. The US is a bit special in that you don't need to be licensed to fiddle with mains. A lot of other countries require tou to be a licensed electrician to work with LV.
The best way around this is to keep it ELV. Find a solid DC supply and use that. Then on the toaster you need to go simpler - look into RV/Caravan, Truck, or Marine (boat) electric ovens that run from 12v or 24v is my suggestion. Though with ELV being up to 50v AC or 120v DC, if you can find something that runs nearer those limits you can cut way down on current.
Sorta like 3D printer heatbed controls - those are usually 12VDC/24VDC at a few hundred watts and use simpler power-MOSFET PWM control and thermistor temp sensing. Unfortunately this would mean your reflow area would probably be a small hotplate and not a big enough thing to process a whole 300mm^2-ish PCB panel.
I'm building a transformer winding machine ATM, and instead of integrating mains PSUs I've stuck a barrel socket in the back and it runs on a 12V 5A wall wart because screw having to spend $125K and about six months of back-and-forth (as of 2018 when I last checked into this) getting RFI/EMI and mains safety certifications.
@@OddlyIncredible Yep, you can get off the shelf ovens that are up to about 500W from my quick googling. Unfortunately they are a lot more expensive than an ordinary toaster oven.
That said, you could also try modifying a toaster oven that runs on 110v AC to run on 110V DC. Since the elements are resistive, it should be able to handle it without too much issue.
The biggest problem is switching DC instead of AC with relays. Pwming that much power will be very inefficient, unless your duty cycle is very low - which is basically a variation on the classic bang-bang control.
I agree that for people like us used to buying most components for cents (or at most single-digit dollars) per part, it's definitely painful to consider buying something as expensive as an SSR for a project like this, but you're correct that a zero-crossing one dramatically simplifies your design in a whole bunch of ways. It's probably the way to go here.
You could use two MOSFETs to switch whole halfwaves with your zero crossing detection. But you would need pretty low Rdson to make a significant reduction in conduction losses. Like 10mΩ for 1W per FET losses. Basically designing an ZCD-SSR by yourself. This would be super neat and I should definetly at some point try it out, BUT for this project you will be far better of keeping the mains of the PCB and using SSRs designed and already certified(!) to handle the current required. My Power Electronic heart bleeds electrons typing this, but after some thought it seems the best compromise. Or you add more heat sinking capability, but thats a whole other can of worms with case isolation and so on.
Back-to-back power MOSFETs and a zero-crossing photodiode gate driver makes for a great and super-efficient SSR. Made a few of those for projects and they're solid performers.
@@OddlyIncredible
For 60 Hz or 120 zcd's and two back to back igbt's or mosfets...
(N ch.. g-d-s in " series " with another as s-d-g) ..its not necessary to have any zcd based triggering.
This is reduced complexity at the cost of slightly higher switching losses..if one considers a switched pwm of around just 240 to 500 Hz or thereabouts...this can provide sufficiently good regulation..
@@analoghardwaretops3976 Yep, that would give pretty robust control. However, that having been said, if he's using a conventional heater element or quarts IR tube oven, he doesn't even need that because the total thermal hysteresis of these toaster/convection ovens is _crazy_ high - for example, I have my oven controller set up with a _10 second_ dwell time.
All he'd really need to do is drive the heater at a 10s "on" time and 50% duty cycle to bring it to soak temp, hang there for a minute or so, go full power to reflow, then switch off and let the oven coast back down from there - no real need in this application to do waveform level control.
My T962A has been great. I did the mods and custom firmware. I soak it at 60 for a few minutes first, and try to avoid putting anything right at the edges of the tray.
it has no heating modules at the bottom and that's a problem.
@@mrechbreger no issues for me.
@@gregstinkertown you might know this, even if you rework outside a heat bed is recommended. If you do this inside this oven you stress the components unnecessarily. It might work but for sure it's not good practice.
Different solder has different melting points, this oven is definitely on the lower part of supporting various pastes.
I would never put TG140 into a single sided heating oven only..
ZC detecting SSR is def a way to go. We use Omrons to PID the beds in VORON printers. I’ve had one PID a 1.5kW heater with a heat sink attached. Pro-tip: Set your PID cycle to whatever your mains frequency is to avoid flickering lights 😊
what do you mean by "Set your PID cycle to whatever your mains frequency is to avoid flickering lights". If you have a ZC detecting SSR, it's already switching at some integral multiple of the mains period, ie: at a fraction of the mains frequency.
@@Graham_Wideman In Klipper, the default PID frequency is order of magnitude lower than 60 or 50 Hz, so you end up switching on every Nth zero cross, and that makes your light flicker. The proper way to fix it is not having lights on the same circuit as outlets but that requires fixing your house wiring.
You can do the 0 crossing chopping with the schematic that you have, just use a to220 triac, 1kW of resistive load shouldn't be too hard. For this aplicacion seems like the better alternative, it's the same as the ss relay, cheaper and smaller as you can control the heat sink for the application, and pretty much any small one would do on a to220 or similar size package.
Chopping mid wave is (or was) useful for light dimmers, as missing half cycle would make for bad flicker.
Other option would be to actually use pwm, if it's much faster than mains the problem you mentioned wouldn't affect the power calculations, and the higher frequency would make for smaller filtering components, much more reasonable than a huge inductor you'd need for the triac half wave. (for pwm a transistor shorting the DC side of a full wave rectifier is used, choose your poison, mosfet or IGBTs are usually the options)
Good luck with the project!!
10:24 Love how you're getting all excited about triacs... For us old folks that used to go out clubbing in the 70s and 80s, these are indelibly linked to lights flashing to the beat of the music.
We're gonna need a bigger triac. You can get some pretty beefy SCRs, silicone controlled rectifiers. I used them many decades ago instead of relays.
50Hz mains have 20mS period, you don't need to be quicker for heating. For lighting it do matter, but not for heater. There are some MOC optotiracs whit built-in zero crossing. I like tht and alu heatsinks for power stuff. You get better cooling with no electrical tab isolation, but then you have a chunk of metal at mains live :) Check for example BTA16 and BTB16 triacs
The real problem is not the controller but the oven it self. It's too larger for the task with too much thermal mass given the limited power input due to mains limitations. I found the quarts tubes tended to burn things before reflow was completed. What is needed is a smaller lower mass type oven cavity that can heat and cool fast, allowing the PID to work correctly. One thought I had was to control the temp with some sort of servo driven air blend door, that could select between a reserve of heated air and cooling air. The heated for the hot air source could be stored in some preheated thermal mass that gets charged prior to starting the the cycle. You could heat charge the thermal mass over a period of time even allowing the use of low voltage input power eliminating the need to deal with mains voltages. Think of a 3D printer with low voltage heated bed and hot end.
You can use the fact that you are not switching a lightbulb to your advantage - the switching is what is heating - but there is absolutely no reason to switch it 100 times a second - probably 10 times a second is more than enough probably even just switch it after measurable temperature change.
The first strategy I can think of is getting an off-the-shelf solid state relay and using delta sigma modulation to determine which half waves to let through and which to block. Maybe running the modulator at half frequency so it works on full waves would be better for EMC, that needs testing.
Using sigma delta is essentially picking particular constants for a PID algorithm. It's not clear that those constants are necessarily optimum, since they don't incorporate any knowledge of the thermal momentum of the oven.
Triacs are usually paired with a DIAC and capacitor. They are capable of handling fairly high current considering they're frequently used on power tools.
Love your face at 8:03 - such a look of joy at something working!
I hope you get this working. Would be really useful. -JC
PWMing is ... noisy from a stability POV and even more so at these power levels. Me, I'd approach it differently and focus on adjusting the AC voltage amplitude being delivered to the load to control power output. Thus, the AC output power waveform is stable and you're just modulating its amplitude.
Use triac and a zero crossing opto isolator and then slow your "pwm" down to a second or two. the power blowing your triac is due to you switching at full voltage rather than on the crossing. A heat sink might be a good idea too even with zero crossing.
IBM did mains switching like this 50 years ago on the IBM/370 to pull off lower voltages
IGBT and a photovoltaic isolator (works like a photocoupler, but instead of having a transistor output it provides voltage -- enough to power an IGBT)
Zero crossing you can do in software
I still have one of your "Atlas Labs" business cards and i'm so happy to see how far you've taken it with Lumen and now this. You're doing good stuff and I can't wait to see the next product line.
Solid state relays really is the way to go for these kinds of things 🙂
Resistive heaters like that have a response time of more than a second, a normal relay or solid state relay switching on or off once a second would be more than fast enough, no reason to overcomplicate stuff.
Yeah, work with those triac in the past and it can defenately do what you are asking of it (when you select the right one) but it will always need cooling, you can add a big ass heat sink on the back or you can use a silicon thermal pad and let the heat sink in the metal inclosure but unfortunately the toaster casing seems to thin for that.
So we actually have a device that does exactly what you are designing- a reflow controller that you plug a toaster oven into. One thing that differs between it and your design is the rate at which it turns on and off. The one we have switches once every second at most. At that frequency you don’t exactly need to worry about the zero crossing issue. I can find out the name of the controller if you are interested!
That info would be useful. Also check that it does indeed run an unmodified toaster oven, and also what does it use for temperature sensing?
@@Graham_Wideman the toaster we use has very little modification to it. We added a small layer of insulation and added a thermistor probe. The controller has an outlet you plug the toaster into. I'll ask what model the controller is
I'm not really sure how accurate you need the control to be but deciding to switch power every 0.25s or so would mean that you don't need to evaluate the phase of your ac. I don't really understand why you need the heater control to be at such a high frequency.
100% chopping at zero crossing is so much easier, less problematic and give you enough control for what you try to do.
But also would be interesting to play with refractive bricks and Nichrome wire (heating element) to build a fully automated p&p + reflow system on rails
Most of the SSR‘s just consists out of 2 back-to-back Mosfet‘s. You can try that as well.
I think you can use ramp/soak controllers for this, they are made by non-etsy style companies for about $100