James I just watched this video. I am using NC proximity swiches also, but in serial. Here is what I did. All switches blue wire to negative. 1st switch brown to 24v, and 1st switch black to 2nd switch brown. 2nd switch black to 3rd switch brown, and 3rd switch black to Acorn input. Have worked perfect for 4 years. Thanks for developing electronic leadscrew. Wonderful stuff.
I read your description of the wiring , makes sense to me . Did you need to connect the auto squaring enable jumper wire between the MASTRE and SLAVED axis?
as an engineer, i can't tell you how many times I have spent time designing solutions to be robust against issues that I predicted, only to find out later that I was optimizing the wrong thing or worried about a problem that, in hindsight, wasn't the most important thing to be worried about. ESPECIALLY in software... premature optimization is something I have slowly learned over the course of my career to avoid. One of my favorite quotes from John Carmack is "You can prematurely optimize maintainability, flexibility, security, and robustness just like you can performance." But good on you for sharing the entire part of that journey here... this is so great for helping people learn! Not just the final result, but the process!
I've repaired industrial CNC machines for while. Here are a few suggestions after fixing a couple thousands of broken things, some tiny things, some huge kabooms. Don't rely on the inductive proximity sensors for homing, they are off by a +-1 mm when activated that way, mount them sideways so the metal comes in from the side. As an oh-sh*t limit switch they are totally fine mounted the current way. On industry grade machine, they indicate that they are in the right ball park, the exact position is determined by a rotational sensors on the axis, but the good one are expensive. But you probably don't need micrometer precision anyway :) Also add a protection enclosure around the sensor, a metal chip flying across there might trigger a false alarm. Your board is fine from an electrical point, clean design and all, but personally I'd move the opto couplers near the limit switches (and have individual tiny boards) in the wiring. You only need +24V, GND and the two "safety" wires, so cutting the existing 4 wire cable close to the sensor and mounting the tiny boards should be easy. This also makes adding extra switches easier, since it is modular. But in the end it's your machine and I'm just a guy on a keyboard across the big pond :)
Not so sure about mounting the optoisolators near the switches. If they're to protect against hazardous voltages getting into the wiring somehow, they need to be as close as possible to the controller.
@@gcewing you are right when it comes to the current setup with only driving the opto-isolator on the controller side. Industrial machines tend to use a physical relay to fight this, not like the avid controller. The relay uses more than a few mA of current and by this is, it's more resilient against stray inductive currents and the like. Also a "just hanging on a thread" connection is too high in resistance, so the relay will open. But that is all a matter of environment. A few thousand $ of hobby/light commercial machine vs hundred thousands (or millions) for the heavy 24/7/365 room sized industrial stuff I used to work with.
I would use micro switches instead of inductive probes. They are much more precise. We use them on Voron printers and get around 0.005mm accuracy with them. And no false triggers because of metal chips 😅
Yeah I was thinking that inductive sensors don’t seem best for this application due to there being some uncertainty in the measurements and why not just use normal limit switches for homing that will be the same and repeatable every time.
17:02 What a beautiful old Radio Shack multimeter. One like that took me most of the way through college, but I went for the standard Simpson 260 as I started to build out my kit. Then came the Fluke addiction and the Beckman/Fluke wars...
James, I have inductive sensors on my BP conversion. I found they work much more accurately if your sensing metal target crosses the axial axis of the sensor at about 1 to 2 mm rather than approach it parallel to the axis. This has the advantage that a failure overrun doesn't smash the plastic end of the sensor. I use them on my driveway gate for limit switches as well. The gardener kept hitting the sensing leaf with his leaf blower knocking it out of alignment. Took a while to figure out what was going on since it only happened every couple of weeks.
This is what I've always been told as well; the "tip" of the sensing field isn't super reliable, so the target should move in perpendicular to the sensor, in the strong area of the field close to the sensor body.
Smashing into the sensors is a big thing. Especially on a larger machine I want my homing speeds as fast as possible but that does mean that the ramp down sequence takes more distance than the detection distance of the sensor. Also, I use standard normal open type sensors with the sensor detecting a hole in the frame instead of a tab at the end. I think this is some extra security since the sensor only triggers when working properly (I hope ...). Ehh... "BP conversion" ? I'd love to see a CNC'ed Bridge Port knee mill 😁
Oh yeah, I can relate to the over-design scenario. So often we engineers over think things. Great video as usual James. Anxious to see it all come together
You get extra credit for using that old Radio Shack multimeter. I think that's the same model my dad had when I was a kid! Your soldering looks fine to me. When I lift a pin like that, I like to insulate the pad underneath with some kapton tape. It probably isn't really necessary, but it helps me sleep at night. And it's even a nice excuse to overcomplicate things.
Yes, I did a double take on the Multimeter as well, I still have the kit version of it (Teal case). I can't say that I have used it any time in the last 30 years....
Just adding in yet another comment here to point out that you'll want to make sure the sensor is set up such that the triggering element enters in from the side, not in line, if used as a home switch. Having used them in both orientations on industrial equipment, when the two are aligned axially it's hard to get a repeatable activation distance.
Once being an industrial designer of heavy moving vehicles, You're totally right. Safety switches should be normally high. A broken cable, switch... should not open a highway.... You can use active low as a signal provided there's a pullup at the controller bringing the input high in case of cable failure. As a start up procedure verifying the signal switching would riase the security level. For analog controls on the vehicles the standard was: 0.5 to 4.5 volt were allowed. Outside that window the control was considered to be inactive. Machine directives is my translation of the set of rules. For optimal performance every analog input device was calibrated. As You say, things depend on the software. Suppose an end switch goes active. Knowing in what direction the movement was, it's easy to know in what end the alarm was triggered. You know what You're doing!
Stacking components is actually allowed under IPC 610. Possibly not advisable to put the resistor on the bottom though as the black surface is where it makes heat. A lot of modern pick and place machines allow for stacking, Europlacer does a very good job of it.
@@Graham_Wideman couple of options for this, you can either use a thicker stencil and hope that enough paste wicks to the top of the component or you can use a jet printer to add extra paste on top of the first component before placing the next one. Modern pick and place machines often have optional dispense heads (but these are quite slow and not particularly accurate). Alternative is that you put the jet printer between pick and place machines.
@@dquad Interesting, thanks! I work sometimes with a Pnp and have seen it once stack resistors in error. It didn't occur to me you might want to do it deliberately.
At 4:36 you discuss the open switches acting as antennas but in reality, N/O inductive proximity switches use solid state outputs which would be pulled high (or low depending on the configuration eliminating the possibility of noise.
Unless I need the sensitivity, I usually assume a forced beta of 50 and then in small signal application double the base drive requirement, which would make R12 more like 1 kilohm instead of 100 kilohms. Then R11 would be still about 10 kilohms to drive 1 milliamp through the base and 1 milliamp through the lower value of R12. But a capacitor across R12 is a good idea anyway, and what you eventually did removing Q6 is better yet. C1 through C3 seem a little large, but maybe the 1 millisecond time constant they provide is not. I would be concerned about the positional accuracy when using the limit switches to align the gantry motors.
100k was probably too weak of a pull-down for that transistor. Given, that the trace, connected to the base is floating and acting as an antenna, when at least one switch is activated, it could easily be picking up a stray EMI from the signal transition on the primary side. Also, there are small solid-state relays like CPC1017 that replicate a mechanical switch more closely, than a simple opto-coupler, since current can flow in any direction through them. Good thing that you removed the transistor because it would be a single point of failure. I'm jealous to your ability to walk in your garage without tripping on random pieces of machinery and tools :)
That multimeter! Museum piece :-) Enjoyed the over-design creating issues - I tend to do same, and transistor capacitances (cb, be and ce) have bit me in the past as well.
I suspect that the main function of the capacitor on the Acorn board inputs is for debouncing mechanical contacts, and probably not necessary for proximity sensors. It won't do any harm, tho', and will make the board more general-purpose.
Nice job ! When you showed us the schematics I taught "why doesn't he make an open drain directly with those optocouplers ?" Nice hack ! I would have "dremeled" the 24V track going to the optocoupler... more irreversible :D
Usually you want those switches on the high side of the input. In automotive there's the shutdown circuit concept, a series of NC switches when safe that are all on the high side of what you want to power. Same thing with any logic signal that regarding safety, high means ok/safe, low means not ok/unsafe. The reason is that it's way more likely to have a short to GND than to +24V.
Good job, simpler is better ... and safer. Just let Acorn do his part of the job. 😊 By the way: I heard from Uw Mattern that there is a new Acorn board.
Hi. You can connect NPN NC sensors in series ....signal of first one in N of second one....signal of second one in N of third one...and so on. Did you need also to define the IN1 HomeAll as NC instead of NO in Centroid Acorn software??
I definitely would be interested in buying one. Or the plans/instructions for assembly. You choose! Much easier to sell plans! Haha Though $60 just for connectors would price me out. So getting the Aliexpress connectors would be the way to go for me. Also, thank you again for taking us along with you in your Acorn CNC development! I am working on an Acorn CNC Mill and these videos are extremely helpful and constructive.
Hey James, I just recently got one of these machine kits set up. i have used your videos as invaluable guides. thank you for putting so much effort, detail, and clarity into your videos. They have been very helpful. Having run a few programs thus far I have encountered rigidity issues, and have been able to deduce that there is a lot of flexibility in the z axis. I was curious if you have detected this as well. To my understanding it is simply a result of the linear stage design not being rigid enough on its own without the additional support from an aluminum extrusion frame like the x and y axis(es?) both have. I plan to reinforce the z axis in an attempt to eliminate some of the lacking rigidity, and have a vague plan as of right now. have you seen these issues? is this something you would consider doing to your machine?
Last I checked Fusion's electrical modules were rudimentary at best. They're better than Fritzing, but drawing the circuits out by hand is too so that doesn't take much. Autodesk Eagle wasn't a very good EDA system to begin with, a stripped-down version integrated into Fusion is even worse. It's been a few years though, maybe they've improved? I doubt they're anywhere near as good as KiCad has gotten though, so if James were going to switch EDA package I'd expect him to go to that instead of Fusion.
@@PeregrineBF it’s gotten a lot better. Main weakness now from what I can tell is component selection and having to manually import stuff from other sources. And it allows you to link your 3d models to the circuit board directly and build them as an integrated whole. Ps eagle is gone and only fusion is really getting updated so all of the improvements are in fusion.
The issue you discovered is indeed due to capacitance. Your deduction is well thought out. I have seen this same issue show up in prototypes more than once. Great content!
I do not doubt it but I am surprised the result was visible on the LED. Unfortunately the datasheet lacks an output capacitance specification, but the output transistors are pretty small, so I would expect less than 2 picofarads which in this circuit would yield a time constant of 100 nanoseconds. The external parasitic capacitances are greater.
When I'm installing large 5axis machines typically they pair up X with A, Y with B and X with C so in your case that would be a B axis since it's in the same plane or just call it Y2 which some machines will use when powering large rack and pinion drives.
Fun video! Would you be so kind as to give the link for the "alternative" M12 connectors. M12's were my favorites when I wired up sensors on drilling rigs.
Thanks for the video. It is interesting to see what you are up to. Normally closed is the only mostly guaranteed safe method for a trip switch. When I built my cnc plasma table I daisy chained all the trips because I was afraid I did not have enough inputs. I would have been happier with a circuit for each switch. It worked anyway.
I looked at the datasheet for the opto, and you should only need one input resistor to +24v. The Vf of the input diode is 1.35v typ and 1.6v max. If you use a 2.2k resistor, the current would be (24v - 1.35v) / 2200 or 10.3 mA for If, which is a good value for normal operation. Max If is 60mA. If Vf is at 1.6V the current would be a bit less, which is fine.
I think it's good. The delay could impact the zero?? Depends on the velocity. Putting the cap across the photo diode in the opt isolator is a little dodgy. The diode is very nonlinear as it turns on. Will that change with temperature? I'm probably worrying to much. Great video.
If you are going to hide the design files for this board behind a Patreon pay wall, it might be a good idea to remove that Open Source Hardware logo from the board, restricting access to the design is not really in the spirit of Open Source. Don't get me wrong, I'm all for making money with your inventions and or using them as perks for your members, but you cant exactly call them Open Source if you do. Nice board though. Cheers.
I agree with you in the spirit of open source but then there's the 'free as in freedom' vs the 'free as in beer' discussion. I believe it's up to him to decide if there's a price associated with access to the source. This discussion has been going on since the early Linux days, see Stallman's writings on this.
Yeah absolutely, no disagreement here on if he should be paid for his work. Its good work and the man should be paid for his work should he choose to be (Patreon included) and if i were interested in gaining those files i would be happy to pay for them. However, there are very clear guidelines around the use of that logo “Open source hardware is hardware whose design is made publicly available so that anyone can study, modify, distribute, make, and sell the design or hardware based on that design". My only issue is with the use of the Open Source logo, making the design available behind a pay wall goes against the guidelines to which the logo represents. I just found it a little contradictory.
@@minchy83 Publicly available does not mean unpaid. What it does mean is as soon as I have bought it I can use, change, update, etc. and then market the boards under my name without any repurcussions. (bar me making it closed source) That is what publicly available means, the public is allowed to use it as their own for their own however they seem fit. Nothing indicates it is bad or unwanted or not allowed to make money of ones designs or hadware. It even clearly states so in the definition you pasted in your comment.
@@VYR1985 That’s not really how Open Source works. Sure you can make money from the design, building and selling a product made from an open design or offering and selling support for the product, but the design files are normally distributed under a free license which makes them available for free for anyone who would like to check them out. This is useful for the designer as the design can be audited and improved by the community which is the reason it is distributed as Open Source in the first place. If you want to use the design itself for commercial purposes then you generally sell the design as a product, not Open Source it.
Do you have a link or name of the panel connectors you are using? The regular ones on ebay etc have the socket fastening nut on the 'wrong' side, so you have to de-wire everything to mount the socket (or create a custom PCB 😉 )
So is it correct to say that duplicating the opt isolator circuit that is already present at the ACORN INPUTS was just for an added layer of protection and is not STRICTLY necessary? Did you try it without and find that there was a problem before adding this element? Could a person design a PCB without all that to basically clean up the mess of wiring that would be required to install all the M12 connections without it? Thank you.
I purchased a 2x3 benchtop from avid, was planning on running a linuxcnc setup, but the 7i97 board has been out of stock since forever and my 7i96s board doesn't have encoder feedback except for a spindle. Anyway so I'm now probably picking up an acorn board, so I was wondering will you be selling your m12 breakout board soon? Would be a huge help.
If you are soldering and your iron is on sleep mode or cooled down, wait for it to heat up then clean it and then solder, don’t just take it straight from sleep and stick it on the board. As it cools down and in some cases sleep makes it stay at a lower but not insignificant temperature, it will oxidise the solder, then you are putting that straight in a board, you should wait for it to heat up and then wipe the old solder off, put some new solder on and then use it. Might not be a problem if using leaded solder which I have no experience with but with lead free solder you need to clean it.
I could have sworn there we some off the shelf NC proximity switches that allowed series connection if you have an optoisolated input on the breakout board.
I've been wiring them in series without issues, but have not testet mechanical timing vs signal output. But it works. You wire 24v to positive on all sensors, and connect ground to the first sensor input. Then you use the negative output to feed the next sensor, and so on.
To simplify further, why not use NC mechanical switches? Inductive switches have their advantages, but is this particular application making use of them? Mechanical switches have better repeatability, even down to
The mechanics of the machine are specifically set up for inductive sensors. Plungers could be installed in these locations, but I wasn't able to find plunger switches with enough overtravel to survive a full-speed crash at 400IPM. If they're out there, I'd look at that as an alternative.
@@Clough42 Point taken on difficulty of finding mounting-compatible switches. I did stumble on the V-15-1CQ plunger switch with 5mm overtravel on aliexpress. Not sure if any good cause I never used it before.
@@stevewalston7089 if that's how you're supposed to use them then you must be ok with a line width that's equal to the hypotenuse of the triangle that makes up the cone. When I'm tig welding I'm looking for perfect fit ups so I want a finer line than that. Also have you used these metal marking pencils? They go blunt pretty quick under randomized rotation. Not like I'm going to track rotation when doing a few hundred cuts in a day. Might as well go back to soap stone, at least you can sharpen those with a quick swipe on the angle grinder, but they're not as nice for curves like tube notching.
The capacitor will probably not only help for noise immunity, but also with switch scattering. There's almost always a capacitor used for filtering some switch input. And that experience with cheap aliexpress alternatives is familiar. The producing tolerances are big. And often the QA is bad or nonexisting. Wouldn't want to use it for mass production.
Hi Do you sell breakup boards. Do you know where can I buy one. I am working on my plasma cnc table & the breakup board will make it easier on me. Thank you.
What soldering looks like is only important for production work, and anything you want to be accountable for. On my home stuff, I spend the time and I make it look as good as if someone was paying me to do it. Because I am paying me to do it, with the use of my free time. But sometimes I also need to make it work NOW and not in 1 minute. I need it NOW so I don't care what it looks like, so long as electrons flow then GOOD! People complaining about other soldering work sound too much like inspectors at a job, and last I checked this was not a job!
There are many reasons why i don't like inductive probes (had tons of problems with them on 3d printer) so for my cnc I've decided to just use good old mechanical ones. And they usually do have 3 pins that allows you to use either NO or NC.
The 4.7 uF is probably too much capacitance. It forms an RC time delay on the home sense. Along with the Acorn capacitor you not have an even larger home sense delay. Any difference in home speed could change home position.
James - once again you are showing that you are (almost) as a__l as I am: only you would plan ahead for the need of a tool to eject the assembled board from the connector fixture. Well done!
The principle with the transistor is alright, but the calculations need to consider that opto-couplers and transistors are analogue devices working with current rather than nice simple voltage controlled digital devices - they both have gain as well. The values of 10k and 100k appear as if they were plucked from the air. If we assume the controller inputs conform to EN61131 type 2 inputs, then this defines a logic 'HIGH' as being over 15V and logic 'LOW' below 5V. From 5V to 15V is undefined, so should be treated as potentially not the state you want! [the standard also defines the valid range for a 24V supply as 19V-30V, which is relevant if the controller is driven from a user supplied "24V", as calculations need to take into account the extremes - for here I'll just use 24V as a rough guide). To bring the input of the controller down to 15V and thus potentially get it to register as "LOW" we'll have 15V across the transistor, call it 1V across the opto LED in the controller leaving 8V over the 1.2k resistor. Therefore the current to bring the controller input into the undefined region is 8V/1.2k = 6.7mA. (to reach the defined "LOW" at 5V would be 18V/1.2k = 15mA) The MMBT3904 datasheet shows the minimum gain with an Ic of 1mA is 70, so potentially to get the controller input to register as low could need as little as 6.7mA / 70 = 96uA. In practice the gain is likely to be higher than 70, so base current could be somewhat less than this. To be sure of registering as "LOW" requires 15mA/70 = 214uA (but with higher gain possibly less). With the 10k & 100k base resistors as shown, to get the transistor Vbe into saturation just needs 0.6V across the 100k resistor: If we have 6uA through the 100k and 96uA into the base then we're looking at 102uA through the upper, so 1.02V over that. This means just 1.6V at the output of the three series optos is giving an undefined state at the controller. To reach a certain low then we're looking at 220uA base current and a voltage nearer 2.8V. This is a very small proportion of a 24V logic signal and in terms of the IEC spec this would still be in the 'LOW' state (
When you solder, what temperature do you set your iron at? I always set mine high up (750-850) so that I can solder faster, but some people suggest to only go slightly higher than the solder melt point, which I found is slow. Wonder if you had any opinions on this
Honestly I don't think the thoughts behind the original design were "way off base" or whatever. I wouldn't even call it an over complication. Your thoughts are pretty much on point and you want to reduce the risk of transients affecting the circuit. I've been down this road on my own machine and noise reduction is a topic that many people will discover, whether they want to or not. I think you might even find that the answer could have more simple that piggybacking another cap atop the resistor. Remove the 100k dropper and use a 1k - 2k resistor instead. This reduces the resistance in the path to ground for transients and I'm at least 95% certain that it would've worked exactly as you wanted it to. But, your original design probably should have included the dropper and a low value capacitor for noise filtering. Hell I would've over complicated this much more than you did lol.... A logic gate and 2 optos... Pretty sure the idea bouncing around in my head is not a "better" approach but I do enjoy a challenge. Logic gates can be simpler and harder at the same time. All the same design challenges apply though.
That was my thought as well. the 10K and 100K across the base didn't make sense. You would like the transistor to forward bias at about 12V (1/2 the 24) in his circuit. Or even a small zener in series with the base would have helped.
Hi, is it possible for us to chat by email? Ironically I came across your channel while doing my research. I had the same idea as you. I wanna buy a Avid CNC and use the Acorn and Clearpath servos. I’m interested in your breakout board.
The only problem with slugging the switches with a C is the delay it is intended to create. Yes it cures the spurius triggers but limit switches are ideally zero time to opperate. Nobody wants the hardware at some velocity and mass, travelling past the limit switches because the electronics took too long to see it. I dont know the time delay but any delay is potential tool tip gone or worse. And trouble with using a processor rather than pure logic is the communalative clock pulses time for sequential reading of the inpjts. I would do the safety limit switching in TTL or Cmos directly to the master STOP and feed that back to the processor for display or recording purposes. If the delta linear distance is only a micron or 2 ill conceed! Have fun.
We are the R&D manufacturer of stepper motor drives in China. Our products include open-loop drives, closed-loop drives, and fieldbus drives. We welcome bulk orders.
If you ever decide to sell these, I'd say don't bother with the cheap ass connectors. They will give you more trouble than the real deal and could break you're company's reputation
Hi , i’m building a CNC router with dual ball screws on the Y axis. I’m a professional, pool cue builder . Can u to build me a board . Pretty please. Thanks
Everyone's a sales man today I for one like content but am tired of the links below and the pay to play for everything...sorry i am unsubscribing every you tube video is looking to make money and when times are tough this is one of the first things to go..
The video is free... the schematic is shown in the video for free... the tutorial explaining the schematic is free... what more do you want? You can take that and play all you want without paying. The pay part is for the files that let you get the board made without yourself having to do the PCB layout, saving you some work. And for the PCB itself and components you're going to have to pay anyway, and that's hardly an unreasonable aspect of the situation.
Just like that you went political it fell for the b******* of the woke culture. sorry but I now have to cancel your channel too, worst part was I enjoyed listening to you.
James I just watched this video. I am using NC proximity swiches also, but in serial.
Here is what I did.
All switches blue wire to negative.
1st switch brown to 24v, and 1st switch black to 2nd switch brown.
2nd switch black to 3rd switch brown, and 3rd switch black to Acorn input.
Have worked perfect for 4 years.
Thanks for developing electronic leadscrew. Wonderful stuff.
I read your description of the wiring , makes sense to me . Did you need to connect the auto squaring enable jumper wire between the MASTRE and SLAVED axis?
as an engineer, i can't tell you how many times I have spent time designing solutions to be robust against issues that I predicted, only to find out later that I was optimizing the wrong thing or worried about a problem that, in hindsight, wasn't the most important thing to be worried about. ESPECIALLY in software... premature optimization is something I have slowly learned over the course of my career to avoid. One of my favorite quotes from John Carmack is "You can prematurely optimize maintainability, flexibility, security, and robustness just like you can performance." But good on you for sharing the entire part of that journey here... this is so great for helping people learn! Not just the final result, but the process!
I've repaired industrial CNC machines for while. Here are a few suggestions after fixing a couple thousands of broken things, some tiny things, some huge kabooms. Don't rely on the inductive proximity sensors for homing, they are off by a +-1 mm when activated that way, mount them sideways so the metal comes in from the side. As an oh-sh*t limit switch they are totally fine mounted the current way. On industry grade machine, they indicate that they are in the right ball park, the exact position is determined by a rotational sensors on the axis, but the good one are expensive. But you probably don't need micrometer precision anyway :) Also add a protection enclosure around the sensor, a metal chip flying across there might trigger a false alarm. Your board is fine from an electrical point, clean design and all, but personally I'd move the opto couplers near the limit switches (and have individual tiny boards) in the wiring. You only need +24V, GND and the two "safety" wires, so cutting the existing 4 wire cable close to the sensor and mounting the tiny boards should be easy. This also makes adding extra switches easier, since it is modular. But in the end it's your machine and I'm just a guy on a keyboard across the big pond :)
Not so sure about mounting the optoisolators near the switches. If they're to protect against hazardous voltages getting into the wiring somehow, they need to be as close as possible to the controller.
@@gcewing you are right when it comes to the current setup with only driving the opto-isolator on the controller side. Industrial machines tend to use a physical relay to fight this, not like the avid controller. The relay uses more than a few mA of current and by this is, it's more resilient against stray inductive currents and the like. Also a "just hanging on a thread" connection is too high in resistance, so the relay will open. But that is all a matter of environment. A few thousand $ of hobby/light commercial machine vs hundred thousands (or millions) for the heavy 24/7/365 room sized industrial stuff I used to work with.
I would use micro switches instead of inductive probes. They are much more precise.
We use them on Voron printers and get around 0.005mm accuracy with them.
And no false triggers because of metal chips 😅
Yeah I was thinking that inductive sensors don’t seem best for this application due to there being some uncertainty in the measurements and why not just use normal limit switches for homing that will be the same and repeatable every time.
@@conorstewart2214 Micro switches add another point of mechanical failure. Cheaper hobby CNC still use them widely
17:02 What a beautiful old Radio Shack multimeter. One like that took me most of the way through college, but I went for the standard Simpson 260 as I started to build out my kit. Then came the Fluke addiction and the Beckman/Fluke wars...
I still have the Simpson 260 that my dad had, and I started using, ugh, 50 years ago.
James,
I have inductive sensors on my BP conversion. I found they work much more accurately if your sensing metal target crosses the axial axis of the sensor at about 1 to 2 mm rather than approach it parallel to the axis. This has the advantage that a failure overrun doesn't smash the plastic end of the sensor.
I use them on my driveway gate for limit switches as well. The gardener kept hitting the sensing leaf with his leaf blower knocking it out of alignment. Took a while to figure out what was going on since it only happened every couple of weeks.
This is what I've always been told as well; the "tip" of the sensing field isn't super reliable, so the target should move in perpendicular to the sensor, in the strong area of the field close to the sensor body.
Smashing into the sensors is a big thing. Especially on a larger machine I want my homing speeds as fast as possible but that does mean that the ramp down sequence takes more distance than the detection distance of the sensor. Also, I use standard normal open type sensors with the sensor detecting a hole in the frame instead of a tab at the end. I think this is some extra security since the sensor only triggers when working properly (I hope ...).
Ehh... "BP conversion" ? I'd love to see a CNC'ed Bridge Port knee mill 😁
I had that multimeter when I was a kid. Didn't expect to see that today!
Way nicer than mine when I did this for the first time, just a blobs of solder and some SMD connectors lol. Nice work
The intro…. Epic! 😂 one afterthought, some led on the board indicating the switch, line state would be easy in case of fault finding.
Oh yeah, I can relate to the over-design scenario. So often we engineers over think things. Great video as usual James. Anxious to see it all come together
As for the price on the TE connectors being $6, that's a very fair price for them, especially from TE. I would have expected them to be $35!
They are very nice. No question about that. It makes a board like this pretty expensive.
You get extra credit for using that old Radio Shack multimeter. I think that's the same model my dad had when I was a kid! Your soldering looks fine to me. When I lift a pin like that, I like to insulate the pad underneath with some kapton tape. It probably isn't really necessary, but it helps me sleep at night. And it's even a nice excuse to overcomplicate things.
Yes, I did a double take on the Multimeter as well, I still have the kit version of it (Teal case). I can't say that I have used it any time in the last 30 years....
Just adding in yet another comment here to point out that you'll want to make sure the sensor is set up such that the triggering element enters in from the side, not in line, if used as a home switch. Having used them in both orientations on industrial equipment, when the two are aligned axially it's hard to get a repeatable activation distance.
I loved the Radio Shack volt meter.
Once being an industrial designer of heavy moving vehicles, You're totally right. Safety switches should be normally high. A broken cable, switch... should not open a highway.... You can use active low as a signal provided there's a pullup at the controller bringing the input high in case of cable failure. As a start up procedure verifying the signal switching would riase the security level.
For analog controls on the vehicles the standard was: 0.5 to 4.5 volt were allowed. Outside that window the control was considered to be inactive. Machine directives is my translation of the set of rules. For optimal performance every analog input device was calibrated.
As You say, things depend on the software. Suppose an end switch goes active. Knowing in what direction the movement was, it's easy to know in what end the alarm was triggered.
You know what You're doing!
Can’t wait to see your new machine cutting aluminum and steel! Considering a similar build.
Stacking components is actually allowed under IPC 610. Possibly not advisable to put the resistor on the bottom though as the black surface is where it makes heat.
A lot of modern pick and place machines allow for stacking, Europlacer does a very good job of it.
Out of curiosity, how does solder paste get on top of the lower component to solder on the upper component?
@@Graham_Wideman couple of options for this, you can either use a thicker stencil and hope that enough paste wicks to the top of the component or you can use a jet printer to add extra paste on top of the first component before placing the next one. Modern pick and place machines often have optional dispense heads (but these are quite slow and not particularly accurate). Alternative is that you put the jet printer between pick and place machines.
@@dquad Interesting, thanks! I work sometimes with a Pnp and have seen it once stack resistors in error. It didn't occur to me you might want to do it deliberately.
At 4:36 you discuss the open switches acting as antennas but in reality, N/O inductive proximity switches use solid state outputs which would be pulled high (or low depending on the configuration eliminating the possibility of noise.
It's entirely possible they are open collect or open emitter.
Unless I need the sensitivity, I usually assume a forced beta of 50 and then in small signal application double the base drive requirement, which would make R12 more like 1 kilohm instead of 100 kilohms. Then R11 would be still about 10 kilohms to drive 1 milliamp through the base and 1 milliamp through the lower value of R12. But a capacitor across R12 is a good idea anyway, and what you eventually did removing Q6 is better yet.
C1 through C3 seem a little large, but maybe the 1 millisecond time constant they provide is not. I would be concerned about the positional accuracy when using the limit switches to align the gantry motors.
Totally agree.. With the 24V R12 between 1K and 10K would be better suited. And C1-C3 seem to be excessively large to me too.
100k was probably too weak of a pull-down for that transistor. Given, that the trace, connected to the base is floating and acting as an antenna, when at least one switch is activated, it could easily be picking up a stray EMI from the signal transition on the primary side. Also, there are small solid-state relays like CPC1017 that replicate a mechanical switch more closely, than a simple opto-coupler, since current can flow in any direction through them. Good thing that you removed the transistor because it would be a single point of failure. I'm jealous to your ability to walk in your garage without tripping on random pieces of machinery and tools :)
That multimeter! Museum piece :-) Enjoyed the over-design creating issues - I tend to do same, and transistor capacitances (cb, be and ce) have bit me in the past as well.
17:02 One of the worlds best meters ever built.
it's always a pleasure to watch, thank you for you work
I suspect that the main function of the capacitor on the Acorn board inputs is for debouncing mechanical contacts, and probably not necessary for proximity sensors. It won't do any harm, tho', and will make the board more general-purpose.
If you could tell me I would really be interested in knowing who you use to make your boards cause it look like they do good work.
Opticouplers are something I stock, just for this kind of problem. I love the capacitor additions, thanks for pointing that out!
Nice job !
When you showed us the schematics I taught "why doesn't he make an open drain directly with those optocouplers ?"
Nice hack ! I would have "dremeled" the 24V track going to the optocoupler... more irreversible :D
Usually you want those switches on the high side of the input. In automotive there's the shutdown circuit concept, a series of NC switches when safe that are all on the high side of what you want to power. Same thing with any logic signal that regarding safety, high means ok/safe, low means not ok/unsafe. The reason is that it's way more likely to have a short to GND than to +24V.
Hi James...Another excellent video and expert commentary. Your hobbies fit you well.
Good job, simpler is better ... and safer. Just let Acorn do his part of the job. 😊 By the way: I heard from Uw Mattern that there is a new Acorn board.
Yes it is acorn 6. Full 5 simultaneously machining
Hi. You can connect NPN NC sensors in series ....signal of first one in N of second one....signal of second one in N of third one...and so on. Did you need also to define the IN1 HomeAll as NC instead of NO in Centroid Acorn software??
What is an "NPN sensor". Are you referring to a sensor with "Open Collector" output perhaps?
As has been posted, mechanical switches are simple and not bothered by transients.
I hope centroid sees this and makes appropriate changes!
26:05 What did you use to remove the flux?
99% IPA.
@@Clough42 Thanks! Makes the soldering look really good.
Thanks for the video!
The circuit board look great, really clean with the white silkscreen on the black mask.
I definitely would be interested in buying one. Or the plans/instructions for assembly. You choose! Much easier to sell plans! Haha
Though $60 just for connectors would price me out. So getting the Aliexpress connectors would be the way to go for me.
Also, thank you again for taking us along with you in your Acorn CNC development! I am working on an Acorn CNC Mill and these videos are extremely helpful and constructive.
Hey James, I just recently got one of these machine kits set up. i have used your videos as invaluable guides. thank you for putting so much effort, detail, and clarity into your videos. They have been very helpful.
Having run a few programs thus far I have encountered rigidity issues, and have been able to deduce that there is a lot of flexibility in the z axis. I was curious if you have detected this as well. To my understanding it is simply a result of the linear stage design not being rigid enough on its own without the additional support from an aluminum extrusion frame like the x and y axis(es?) both have. I plan to reinforce the z axis in an attempt to eliminate some of the lacking rigidity, and have a vague plan as of right now. have you seen these issues? is this something you would consider doing to your machine?
Any reason you don’t use the electrical stuff in fusion? Really trying to understand it better.
Last I checked Fusion's electrical modules were rudimentary at best. They're better than Fritzing, but drawing the circuits out by hand is too so that doesn't take much. Autodesk Eagle wasn't a very good EDA system to begin with, a stripped-down version integrated into Fusion is even worse. It's been a few years though, maybe they've improved? I doubt they're anywhere near as good as KiCad has gotten though, so if James were going to switch EDA package I'd expect him to go to that instead of Fusion.
@@PeregrineBF it’s gotten a lot better. Main weakness now from what I can tell is component selection and having to manually import stuff from other sources.
And it allows you to link your 3d models to the circuit board directly and build them as an integrated whole.
Ps eagle is gone and only fusion is really getting updated so all of the improvements are in fusion.
Just realized I haven't been getting my clogh42 notifications. Stupid algorithm
The issue you discovered is indeed due to capacitance. Your deduction is well thought out. I have seen this same issue show up in prototypes more than once. Great content!
I do not doubt it but I am surprised the result was visible on the LED. Unfortunately the datasheet lacks an output capacitance specification, but the output transistors are pretty small, so I would expect less than 2 picofarads which in this circuit would yield a time constant of 100 nanoseconds. The external parasitic capacitances are greater.
I lost points on an exam for saying normally closed was safer
Professors are just the best
When I'm installing large 5axis machines typically they pair up X with A, Y with B and X with C so in your case that would be a B axis since it's in the same plane or just call it Y2 which some machines will use when powering large rack and pinion drives.
Fun video! Would you be so kind as to give the link for the "alternative" M12 connectors. M12's were my favorites when I wired up sensors on drilling rigs.
Thanks for the video. It is interesting to see what you are up to. Normally closed is the only mostly guaranteed safe method for a trip switch. When I built my cnc plasma table I daisy chained all the trips because I was afraid I did not have enough inputs. I would have been happier with a circuit for each switch. It worked anyway.
I looked at the datasheet for the opto, and you should only need one input resistor to +24v. The Vf of the input diode is 1.35v typ and 1.6v max. If you use a 2.2k resistor, the current would be (24v - 1.35v) / 2200 or 10.3 mA for If, which is a good value for normal operation. Max If is 60mA. If Vf is at 1.6V the current would be a bit less, which is fine.
I think it's good. The delay could impact the zero?? Depends on the velocity. Putting the cap across the photo diode in the opt isolator is a little dodgy. The diode is very nonlinear as it turns on. Will that change with temperature? I'm probably worrying to much. Great video.
If you are going to hide the design files for this board behind a Patreon pay wall, it might be a good idea to remove that Open Source Hardware logo from the board, restricting access to the design is not really in the spirit of Open Source. Don't get me wrong, I'm all for making money with your inventions and or using them as perks for your members, but you cant exactly call them Open Source if you do. Nice board though. Cheers.
I’m sorry @@WheatMillington, I don’t quite understand your comment. Would you mind elaborating?
I agree with you in the spirit of open source but then there's the 'free as in freedom' vs the 'free as in beer' discussion. I believe it's up to him to decide if there's a price associated with access to the source. This discussion has been going on since the early Linux days, see Stallman's writings on this.
Yeah absolutely, no disagreement here on if he should be paid for his work. Its good work and the man should be paid for his work should he choose to be (Patreon included) and if i were interested in gaining those files i would be happy to pay for them. However, there are very clear guidelines around the use of that logo “Open source hardware is hardware whose design is made publicly available so that anyone can study, modify, distribute, make, and sell the design or hardware based on that design". My only issue is with the use of the Open Source logo, making the design available behind a pay wall goes against the guidelines to which the logo represents. I just found it a little contradictory.
@@minchy83 Publicly available does not mean unpaid. What it does mean is as soon as I have bought it I can use, change, update, etc. and then market the boards under my name without any repurcussions.
(bar me making it closed source)
That is what publicly available means, the public is allowed to use it as their own for their own however they seem fit.
Nothing indicates it is bad or unwanted or not allowed to make money of ones designs or hadware. It even clearly states so in the definition you pasted in your comment.
@@VYR1985 That’s not really how Open Source works. Sure you can make money from the design, building and selling a product made from an open design or offering and selling support for the product, but the design files are normally distributed under a free license which makes them available for free for anyone who would like to check them out. This is useful for the designer as the design can be audited and improved by the community which is the reason it is distributed as Open Source in the first place. If you want to use the design itself for commercial purposes then you generally sell the design as a product, not Open Source it.
When will the M12 board be ready for sale I am in need of one Please!!
Do you have a link or name of the panel connectors you are using? The regular ones on ebay etc have the socket fastening nut on the 'wrong' side, so you have to de-wire everything to mount the socket (or create a custom PCB 😉 )
So is it correct to say that duplicating the opt isolator circuit that is already present at the ACORN INPUTS was just for an added layer of protection and is not STRICTLY necessary? Did you try it without and find that there was a problem before adding this element? Could a person design a PCB without all that to basically clean up the mess of wiring that would be required to install all the M12 connections without it? Thank you.
I purchased a 2x3 benchtop from avid, was planning on running a linuxcnc setup, but the 7i97 board has been out of stock since forever and my 7i96s board doesn't have encoder feedback except for a spindle. Anyway so I'm now probably picking up an acorn board, so I was wondering will you be selling your m12 breakout board soon? Would be a huge help.
If you are soldering and your iron is on sleep mode or cooled down, wait for it to heat up then clean it and then solder, don’t just take it straight from sleep and stick it on the board. As it cools down and in some cases sleep makes it stay at a lower but not insignificant temperature, it will oxidise the solder, then you are putting that straight in a board, you should wait for it to heat up and then wipe the old solder off, put some new solder on and then use it.
Might not be a problem if using leaded solder which I have no experience with but with lead free solder you need to clean it.
I could have sworn there we some off the shelf NC proximity switches that allowed series connection if you have an optoisolated input on the breakout board.
I've been wiring them in series without issues, but have not testet mechanical timing vs signal output. But it works.
You wire 24v to positive on all sensors, and connect ground to the first sensor input.
Then you use the negative output to feed the next sensor, and so on.
Do these proximity sensors have any sort of bounce like a mechanical switch's signal would have? And what about hysteresis?
Well explained
To simplify further, why not use NC mechanical switches? Inductive switches have their advantages, but is this particular application making use of them? Mechanical switches have better repeatability, even down to
The mechanics of the machine are specifically set up for inductive sensors. Plungers could be installed in these locations, but I wasn't able to find plunger switches with enough overtravel to survive a full-speed crash at 400IPM. If they're out there, I'd look at that as an alternative.
@@Clough42 Point taken on difficulty of finding mounting-compatible switches. I did stumble on the V-15-1CQ plunger switch with 5mm overtravel on aliexpress. Not sure if any good cause I never used it before.
I hate when I out engineer myself. Thanks for sharing that!
Unrelated question: I saw your metal mechanical pencil. I have one too, how do you sharpen yours so perfectly?
Probably OCD school, but more likely he just rotates it often during use like you're suppose to.
@@stevewalston7089 if that's how you're supposed to use them then you must be ok with a line width that's equal to the hypotenuse of the triangle that makes up the cone. When I'm tig welding I'm looking for perfect fit ups so I want a finer line than that. Also have you used these metal marking pencils? They go blunt pretty quick under randomized rotation. Not like I'm going to track rotation when doing a few hundred cuts in a day. Might as well go back to soap stone, at least you can sharpen those with a quick swipe on the angle grinder, but they're not as nice for curves like tube notching.
The capacitor will probably not only help for noise immunity, but also with switch scattering. There's almost always a capacitor used for filtering some switch input.
And that experience with cheap aliexpress alternatives is familiar. The producing tolerances are big. And often the QA is bad or nonexisting. Wouldn't want to use it for mass production.
00:07 ... so any old Tuesday?
The time constant on your RC filter seems excessively high. How did you arrive at those values?
did you have to connect the -AUTO-SQUARING ENABLE JUMPER - ?
That’s called design evolution
Hi
Do you sell breakup boards. Do you know where can I buy one. I am working on my plasma cnc table & the breakup board will make it easier on me. Thank you.
That is why usually using pull up, or pull down resistors, when its NO ))))
When all else fails, follow the directions!
why not use an dc SSR instead ? direct from the 3 sensors
What soldering looks like is only important for production work, and anything you want to be accountable for. On my home stuff, I spend the time and I make it look as good as if someone was paying me to do it. Because I am paying me to do it, with the use of my free time. But sometimes I also need to make it work NOW and not in 1 minute. I need it NOW so I don't care what it looks like, so long as electrons flow then GOOD!
People complaining about other soldering work sound too much like inspectors at a job, and last I checked this was not a job!
I prefer hall effect sensors and mounting small magnets to the thing that moves.
There are many reasons why i don't like inductive probes (had tons of problems with them on 3d printer) so for my cnc I've decided to just use good old mechanical ones. And they usually do have 3 pins that allows you to use either NO or NC.
The 4.7 uF is probably too much capacitance. It forms an RC time delay on the home sense. Along with the Acorn capacitor you not have an even larger home sense delay. Any difference in home speed could change home position.
very interesting, when I wake up I think I can leave a positive comment : )
Who is your pcb board manufacturer?
James - once again you are showing that you are (almost) as a__l as I am: only you would plan ahead for the need of a tool to eject the assembled board from the connector fixture. Well done!
Thanks for sharing.
The principle with the transistor is alright, but the calculations need to consider that opto-couplers and transistors are analogue devices working with current rather than nice simple voltage controlled digital devices - they both have gain as well. The values of 10k and 100k appear as if they were plucked from the air.
If we assume the controller inputs conform to EN61131 type 2 inputs, then this defines a logic 'HIGH' as being over 15V and logic 'LOW' below 5V. From 5V to 15V is undefined, so should be treated as potentially not the state you want! [the standard also defines the valid range for a 24V supply as 19V-30V, which is relevant if the controller is driven from a user supplied "24V", as calculations need to take into account the extremes - for here I'll just use 24V as a rough guide).
To bring the input of the controller down to 15V and thus potentially get it to register as "LOW" we'll have 15V across the transistor, call it 1V across the opto LED in the controller leaving 8V over the 1.2k resistor. Therefore the current to bring the controller input into the undefined region is 8V/1.2k = 6.7mA. (to reach the defined "LOW" at 5V would be 18V/1.2k = 15mA)
The MMBT3904 datasheet shows the minimum gain with an Ic of 1mA is 70, so potentially to get the controller input to register as low could need as little as 6.7mA / 70 = 96uA. In practice the gain is likely to be higher than 70, so base current could be somewhat less than this. To be sure of registering as "LOW" requires 15mA/70 = 214uA (but with higher gain possibly less).
With the 10k & 100k base resistors as shown, to get the transistor Vbe into saturation just needs 0.6V across the 100k resistor: If we have 6uA through the 100k and 96uA into the base then we're looking at 102uA through the upper, so 1.02V over that. This means just 1.6V at the output of the three series optos is giving an undefined state at the controller. To reach a certain low then we're looking at 220uA base current and a voltage nearer 2.8V. This is a very small proportion of a 24V logic signal and in terms of the IEC spec this would still be in the 'LOW' state (
i would put some kapton tape under the lifted pin :)
I'm all about technology and solid state stuff, but knowing how said technology is, I vote for NC mechanical switches in this situation.
When you solder, what temperature do you set your iron at? I always set mine high up (750-850) so that I can solder faster, but some people suggest to only go slightly higher than the solder melt point, which I found is slow. Wonder if you had any opinions on this
Honestly I don't think the thoughts behind the original design were "way off base" or whatever. I wouldn't even call it an over complication. Your thoughts are pretty much on point and you want to reduce the risk of transients affecting the circuit. I've been down this road on my own machine and noise reduction is a topic that many people will discover, whether they want to or not.
I think you might even find that the answer could have more simple that piggybacking another cap atop the resistor. Remove the 100k dropper and use a 1k - 2k resistor instead. This reduces the resistance in the path to ground for transients and I'm at least 95% certain that it would've worked exactly as you wanted it to. But, your original design probably should have included the dropper and a low value capacitor for noise filtering.
Hell I would've over complicated this much more than you did lol.... A logic gate and 2 optos... Pretty sure the idea bouncing around in my head is not a "better" approach but I do enjoy a challenge. Logic gates can be simpler and harder at the same time. All the same design challenges apply though.
You probably could have just reduced the resistor values and kept the circuit. In particular the 100k could be 1k and the 10k 3 or 4.7k
That was my thought as well. the 10K and 100K across the base didn't make sense. You would like the transistor to forward bias at about 12V (1/2 the 24) in his circuit. Or even a small zener in series with the base would have helped.
HAVE YOU DECIDED TO SUPPLY THIS BOARD
Hi, is it possible for us to chat by email? Ironically I came across your channel while doing my research. I had the same idea as you. I wanna buy a Avid CNC and use the Acorn and Clearpath servos. I’m interested in your breakout board.
The only problem with slugging the switches with a C is the delay it is intended to create.
Yes it cures the spurius triggers but limit switches are ideally zero time to opperate.
Nobody wants the hardware at some velocity and mass, travelling past the limit switches because the electronics took too long to see it.
I dont know the time delay but any delay is potential tool tip gone or worse.
And trouble with using a processor rather than pure logic is the communalative clock pulses time for sequential reading of the inpjts.
I would do the safety limit switching in TTL or Cmos directly to the master STOP and feed that back to the processor for display or recording purposes.
If the delta linear distance is only a micron or 2 ill conceed!
Have fun.
Safety Controls our always NC we always wire separately and setup alarm for each sensor
We are the R&D manufacturer of stepper motor drives in China. Our products include open-loop drives, closed-loop drives, and fieldbus drives. We welcome bulk orders.
For future reference, you can get 2 wire prox's that can be wired in series.
Never use NO, always use NC and a pull up/down circuit.
China: "Is it good quality?" "Yes" "How far is it to the airport?" "Yes"
Hmm good idea I thought.
So,
I asked China "are all these parts free"
China replied "No"
I really hate a smart arse.
👍👏👏👏
Dyesssss!
If you ever decide to sell these, I'd say don't bother with the cheap ass connectors. They will give you more trouble than the real deal and could break you're company's reputation
Hi , i’m building a CNC router with dual ball screws on the Y axis. I’m a professional, pool cue builder . Can u to build me a board . Pretty please. Thanks
Am I the only one who thinks it's wierd to put the limit switches on the same input?
Acorn CNC MachinEs: ACME
Heh heh heh
Hi i like your videos but your measures in inches are killing me ( Europeans do everything in mm)
They don't quite plug in ... that'll teach you not to be a cheapskate!!!
Elon's rule: The best part is NO part
Everyone's a sales man today I for one like content but am tired of the links below and the pay to play for everything...sorry i am unsubscribing every you tube video is looking to make money and when times are tough this is one of the first things to go..
The video is free... the schematic is shown in the video for free... the tutorial explaining the schematic is free... what more do you want? You can take that and play all you want without paying. The pay part is for the files that let you get the board made without yourself having to do the PCB layout, saving you some work. And for the PCB itself and components you're going to have to pay anyway, and that's hardly an unreasonable aspect of the situation.
Just like that you went political it fell for the b******* of the woke culture. sorry but I now have to cancel your channel too, worst part was I enjoyed listening to you.
What are you talking about?
If you care that much your just as sensitive as the "woke people. May want to look in the mirror.