EEVblog
ฝัง
- เผยแพร่เมื่อ 20 พ.ค. 2024
- Previous Video HERE: • EEVblog #258 - PSU Hou...
Rev C Schematic HERE: www.eevblog.com/projects/usupply/
Part 12 in the PSU design series.
A thorough look at the Rev C schematic and all the changes.
18650 Lithium Ion battery charger, software controlled switching pre-regulator, replacement of the uCurrent capability, return of the PWM DAC, accommodating the Wiznet Ethernet module, and a whole lot more. - วิทยาศาสตร์และเทคโนโลยี
You did it again Dave! Every time I decide to discard my own project you post something inspirational and pump me up to get going again.
IT'S AWESOME! I love he concept of the pre-regulator, geat effciciency and low noise. Now you are using the full potential of a microcontroller. I will use many of these building blocks in my Li-Ion battery powered projects.
When are we going to see the uSupply come back to life? I want one but years later and the project as stalled. Please bring back the micro supply!!! I really would like to see this completed. This is a treasure for us beginner makers and electronic hobbyists. Thank you very much :D
Thomas van den Dries made an updated version: www.instructables.com/id/Digital-Battery-Operated-Powersupply/
What happened, did Dave fail on this project and just hope over time this would disappear?
He's binned it, but don't despair he may find it again when he doe's another dumpster dive. LOL,
@kingoslo1 I know, I'm a bit disappointed at losing the Micro USB charging myself. Still thinking about it. That would basically require changing to a single cell battery solution, and less capacity, and no low noise linear mode. Or have another boost converter. or another switching battery charger chip perhaps.
i found it amazing how one little change leads to an whole new design, i like that mentality
@YeukyCV No, because the Vcontrol pin is connected to Vin. About 1.6V dropout max at full load.
@sifterjoe It's an extra step, and extra cost. The front panel has to be done anyway, so it's free to put it there. And what if you want to stack units?
Since you dropped the through-hole requirement, it seems like rather than measuring the current in two different ways, it would be cleaner to use the old MAX4080 and switch the Atmega out for an Xmega, which has 12-bit ADC/DAC built-in. A shame XmegaDuino is still missing analogReference and the Ethernet library (and it was no doubt much further behind three years ago when this video was posted).
Dave, It looks great as a field work power supply. Did you make a kit out of this design? would love to get one. Alex
@lovroore The whole thing will be SMD, except the processor (socketed) and regulator.
Dave, It's great watching all this I am learning so much from it and the other videos.. I am just a hobbyist to this and am working on my first power supply from coil winding a multi tap toroidal transformer to microcontrolled PSU.
Your talking about going CORDLESS but want to hook into a Ethernet cable.. Found a WIFI I2C on digikey free up 3 pins I think and cut that cord. Why not ditch those round batteries as well and go with Li-Polymer 3.7V 10000mAh or two maybe three for more power.
Hope the surprise secret project is you've finished this project because I think your last video rant was you changed the AVR to PIC. Cheers
@egressvk It is creative commons attribution-share-a-like. So you can use it (or parts of it), but you have to give the design back to the community under the same terms, and give credit.
Is there any reason why you didn't turn your tracking pre-reg into a proper tracking pre-regulator by referencing the switchers feedback to the output voltage of the LT3080? You would essentially eliminate the need for the EPOT
Also, your boost/sepic/flyback can be made to have a secondary, low-current & poorly regulated output for the LT3080 control supply, so you can decrease the LT3080 headroom & decrease thermal dissipation further...
@exscape It's optional, use it for whatever you want, or not at all. Yes, it's been mentioned before.
Thanks Dave, great video.
Is there a reason you have gone with 2 lipo cells rather than going with one?
Hi Dave, I noticed you you have a 5 pin header for the Sparkfun FTDI basic board. I think these require a six pin header. Pin 5 is CS and Pin 6 is GND. Both pin 5 and 6 are connected to ground.
As mentioned the CN4 I assume is wrong, one side is GND, the other side is the battery pack positive?
Can't you use a DC adapter that is 9V (8.5V?) in and use the switch to bypass battery.
You can also run the MCP73213 STAT output to the AVR to look for a flashing LED for temp fault, or bad battery.
Also you need an LED on 3.3V, the battery charger chip STAT LED goes off when the charge is complete (HiZ on STAT)
Thanks for sharing this.
Dave you're unstoppable!
I saw a few errors: The battery connector has 2 plus connector, but no negative.
The reset on the serialconnector is connected in series with the reset line, not in parallell.
Its awesome...but i saw your schematics and you have used a digital pot with MIC2253 boost regulator. Just wanted to know whether that works out well for you . why did you connect 820ohms resistor in series with it? also the entire digital pot resistance is in parallel with 2k2ohms resistor which reduces it even further.why did you need to do that? Also the inductor you chose must have been selected for a particular output voltage how does it work well with all the other voltages(dont you face any problems regarding higher ripple)? ....can similar methods be used with a Buck converter like LM2596 or the fav MC34063?
You still could use four 18650 batteries. You would charge them each with a TP4056 controller. The output of the batteries could just go to a boost converter to whatever voltage.
Also, if you're going SMD anyway, you can save major bucks and board space on especially your passives, and get much better ones as well.
It's not a clash. It can be both mains and battery powered, and the Ethernet module is entirely options, it costs nothing to add it.
Hi Dave, many thanks for this content. What program did you use to draw the schematic and test? Much appreciated.
@zzzzbest So beginners playing around with it can change the chip if they brick it, or use another micro.
I'm coming from the (embedded) software side.. but I'd go for a STM32F100 MCU instead of the AVR. It has two 12-bit DAC's and ADC's, plenty of I/O's... It might cost a little bit more but you'll save on external chips (no more IO extenders, get rid of the external DAC an ADC chips, .. ). Do I make sense? After, all, soldering SMD devices is easy, didn't you say so? BTW; you can directly connect the quadrature encoders and get their position directly through device registers. Amazing!
High Dave, loved the video. I see someone asked for the code. Is it proprietary? Thanks, Brian
Can't you simply charge the cells in parallel? The 5v from a USB port would then be enough
Really cool, since it's a portable unit I can't wait for the drop test.
That series battery connection looks funny. I'm sure, the battery is connected between VBAT and ground.
Revisit on 2016, have the same idea for bloody bench power supply is too bulky. Design a portable power supply, switch model regulator is better in size and power lost and step down is better for battery supply!
Halfway through the video, I thought you'd sacrificed the farm to save the animals. By the end, you'd won me over. I think you've nailed it.
good and informative video Dave ! "I love it!" :)
Great disign and very well worked out, not a good use of money for the voltage out or current but .....
Maybe this was discussed in a subsequent video that I didn't see or on the blog, but my concern about the tracking pre-regulator design is that the thermal dissipation of the linear regulator was based on a 2V drop, but this would only apply to output voltages from 2V below the battery voltage and up. If the output voltage is 1.8V for example and the batteries are at 7.8V, you would still have a ~6V drop, and 6W dissipation. I wonder if the heatsink you picked can handle that.
Didn't understand this either
He is varying the output voltage of the switching regulator to keep it 2 volts above the output voltage. Therefore, the linear regulator will only ever see Output +2V on its input. The power dissipation of the switching regulator is much less (maybe even negligible.)
@@jbarker2160 it is a boost converter so can't lower the voltage, it can only boost. If the battery voltage is 8v an the set output voltage is 2v, the linear regulator still needs to be able to dissipate 6W at 1A (minus the voltage drop at that shunt resistors)
@YeukyCV You have to factor in the loss at 1 A (1 V) over the current shunt, in addition to the LT3080 drop.
How does U12A work as a voltage follower? U12 pin 8 is tied to V+ and when there is no current or very little current through the 1 ohm sense resistors pin 3 will also be V+. The LM358 is not a rail to rail op amp so pin 1 will only go to Voh. I understand why the voltage follower is there but I don't see how this current sense circuit could possibly work.
You could move the back-light color to a dip switch or even jumpers if you end up needing those pins. Like you say, being able to infinitely vary your back-light color is cool and all, but not a game changer. I also like the comment about a back-light disable switch should you want that little extra bit of battery life...
@EEVBlog: Hi Dave, always love your video's! Will there be a part 13 for this project? I Really would like if you share the gerbers files and Arduino code for this project.
On further thought, you should use a Zeta converter. The Zeta converter, like a buck converter, has continuous current at its output and therefore produces less noise than a boost, flyback, or sepic. Like a buck-boost, its output can be either greater or less than its input.
I'm confused - what's the Ethernet for? I don't think it was ever mentioned. Some sort of web interface for stats or such?
Dave, you don't have the pins for this, but you could change the charging circuit Rprog to be two series resistor and a small FET to short out one of them so the AVR could control a low charing mode and a high charging mode and in the software you could based this on supply output current and battery voltage level. (So Rprog is 2k or 1.2k)
Hi Dave, Is there ever going to be part 13?
I do like where you have gone with these changes, and I was not a fan of the previous design because of its low volt/amp. But making it portable and higher volt turns it into a device that i want.. Just having it portable would have been enough.. But the higher volts is a cherry on top.
If you use ENC28J60 module (10 Mbps only) you can save another 7-10 $! I got one from Ebay for 11 $ shipped, it has 3.3V suply, SPI interface and 0.1" 10 pin connector.
This is great stuff Dave :-)
On the diagram in this revision, you have 2 lo-pass filters for the PWM DAC input.
Assuming you are still running this off a 10KHZ clock, I believe you've got the first filter cap wrong, as it looks into a resultant resistance of R25 & R36 in parallel, which equals 3.2K
The cap should then be approx 3 times greater in val, e.g. 330nF ... ?
Thanks for your video series, I really enjoy them :-)
how is the psu coming along right now and do you have automatic shutoff and what is the price going to be
I liked the power supply scheme and the narration in the video is very nice.
I think it would be better if this power source could be made with discrete components
Where in Jupiter's name is part 13? All this and we are yet to see the final version, such disappointment :(
About the bat. capacity: those '3000mAh' cells with built-in protection rarely if ever deliver. Even currently the best cells in the industry typically give you 2600-2850mAh from 3.7 to 4.15V (which is what you'd want for this kind of application). The only way to get to really high capacities is to use overchargeable (4.25-4.35V types) and overdischargeable (2.9-3.3V) cells, which you just can't get. Normal cells would die after 50 cycles or so. So for ebay/DX cells, expect 2200-2400mAh.
@kkpdk I agree 12V DC adapter is too high and the limit on the DC-DC converter chip is 10V anyways. With a 9V external DC, that is still +70degC charging a dead battery at 550mA. And +25degC for charging a mostly full battery. (Plus you have the battery self heating). The charging chip will shutdown at 150degC so worst case he adjusts the charging current resistor to be smaller (but that would mean 10 hr recharge time) Wonder if he can put chip at back and tie to aluminum L bracket.
Seeing the evolution of the design, I wonder why to use the linear regulator and the switched converter? Can't you get rid of the linear regulator and improve the filtering for the switched converter? What are the considerations?
I think the issue would be quality of the regulation at the output after a buttload of LC filtering. The output voltage can sag temporarily with heavy load (the series inductance), and maybe overshoot, when the load suddenly vanishes or is disconnected. Although I'm not sure. I think LC filtering is fine if the load impedance remains constant.
What about if the output voltage is 0V at 1A, with a 9V boost voltage you're dissipating 9W...
@kingoslo1 About the backlight, since it's connected to the microcontroller (via the I2C-expansionchip) you'll probably be able to switch it on and off in software, and since you can change the software like you want to you could even dedicate a button on the front for that, if it's that important.
Check the power dissipation in the charger. It looks borderline.
The mcp73213 is a linear regulator. With 8V on the cells, 12V 1A in, you are dissipating 3.4W in the charger ic. 45K/W, +143 deg.C temperature rise (>abs max, poof). At 0.5A the rise is 70 deg.C, you get around 54 deg.C max air temperatur inside the box. Maybe ok since the batteries have to live there, but take care.
Hi, Dave. Did you know Atmega can go up to 16-bit PWM on TIMER1 so you don't have to limit to 8-bit pwm
Dave, make sure you give it the ability to run off a 9V battery instead of only 8.4V because you could possibly make a cheaper 9V, lower power, microamp resolution supply option. Also, what about a DPDT switch to disconnect the battery charger and battery and connect straight to the DC wall adapter jack.
I just picked up at Part 11- I thought you might be incorporating POE charging... or even operation.
Nice. I didin't catch mention of the money saved at parts, but I think 10bit AD conversion is plenty for power suppluy. At some time I'd expected you'll to move to another Atmel chips with more IO:s (thus eliminating I/O expander), but not sure witch one would be Arduino compatible
All in all, I like this one a lot more. It has a lot less unnecessary Fluff and puff .. .
I’m trying to do something similar with an lm317 just because that is what I have, but I cannot get the lm358 to work as a differential amplifier like shown, anyone else have any luck with using the lm358 as a differential amp?
Hey Dave, this video series is great! would you consider using the atmega32u4 instead of the atmega168?. the 32u4 is what the newer Arduino Leonardo uses, and it has builtin usb support, so you would get serial-over-usb support for free. would probably increase the cost slightly, but might be worth it. keep em coming :)
Why not put the LCD soldered to the backside of the PCB and have it coming out top of case? You could laser cut the plastic case or put a window in.
@maksym2k Wasn't me, I'm still pulling the WebM file; but it has been a few revisions, and whether your design wonks are Spec1, Rapid Devs, field engineers who need it to squawk and fail to hi-rel (120 degrees C instead of 90-100) ultracaps if it's warming up much, their metrics for utility:trouble ratio could have failed low. (Or maybe it's not entertaining to toddlers at 430x320; or their membership in the USB Consortium noted >2.1A >5microseconds charging *no log* and had to vote it down.)
Well this was quite unexpected.. I had an idea to make precise, atmel controlled, battery powered, linear power supply for quite a while in my head. Now I will probably stop sketching schematics of my supply and just wait until yours will hit the market..
One question though - how it is with you and copyright? Say if I finally got myself to do my supply, but borrowed current sensing from your uCurrent solution. Unfortunately, not for home use, the plan was to sell it commercially.
Thumbs up!
With all the updates to the schematic, have you considered going with micro-controller to some thing like a PIC or ARM, that have higher resolution ADC's and PWM, on chip DAC, and Ethernet, USB, and more IO. I know the cost of the micro-controller is higher, it would let you to have a smaller parts count. I under stand using the Arduino framework, however there are almost as simple frameworks for PIC32 (chipkit) and some ARM's, mapleor mbed. One thing that would be great for me, a Xbee socket.
What is the Ethernet going to be used for? Why choose that over a regular USB serial?
@Mrkirill578 Unless you want to log the power consumption of a device to a computer. Or control the output of a supply to mimic something like a solar cell over a time period. There are a couple uses I could see to connections to a more easily programmable interface.
You say it dissipates only 2W? But the boost can only go down to VBAT-0.6V, or 8V. At 1Vout, that's 7 watts at 1A out... Might want to change your boost to a sepic or flyback...
Dave, do not forget to put some extra I2C memory in. With an ethernet connection you probably want to show some sort of web-page. The ATMega probably has not got enough memory (left) to store all data for it. And maybe you could use the memory for logging currents or charge programs (e.g. desulfation, fast charge, trickle charge, internal resistance measurement). With the boost converter in it, it could probably charge its own batteries. Perpetuum mobile style. ;-)
@eevblog, at 0:13:18 CN4 seems to be connected wrong, Should not CN4 have one terminal at GND and another at VBATT. You have there 1 terminal at VBATT and another at the U2-3 (the VBATT terminal on the chip. This is having the batteries floating with respect to GND not what I think you were intending.
@hjohnson7801 Yes, it should have been 6 pins of course. Fixed.
@martyfriedman666 Bench supply is just the generic term used for such things. It was always going to be battery powered.
Because note that you can't get legit cells from anywhere but the original manufacturers (golden power, panasonic, etc.). Outside of legitimate brokers, all resellers use 'fell off the back of a truck'-cells which can come from QA-rejected batches, stolen batches, used or prototype stuff, etc. etc. They are rarely actual to-spec performing, legitimate cells. And that protection is usually dodgy, not really something that would stand up to regulation scrutiny.
when it will continue?
are you programming this in arduino software?
Dave thanks for the video's ,I watch you regularly .Since you have got a dc to dc converter and micro on your board couldn't you have used that to charge your batteries from any input voltage under software controll, (just not at the same time as you where using the supply).
I was going to buy one of your supplies and mod it, to be able to supply 3 Amp.Your changes have made that much more difficult to do I think :( .
battery connector ground?
What's the latest on this Dave? Will we ever be able to buy the PCBs ?
Why Ethernet? What use could this possibly serve?
@barcelona08018 a) Extra capacity b) They fit.
where can I download the firmware of your project
Very nice development of the PSU Dave!
Now I really hate 'how about' comments, (this isn't) but I was thinking of a PC type 'blowhole' setup where the charger IC/regulation/batteries are on another board under the case top and a small fan extracts hot air upward. I know, a separate PCB :-(
Very interesting, and thanks again.
Project finished by Thomas van den Dries: www.instructables.com/id/Digital-Battery-Operated-Powersupply/ (thanks to Luc De Meester for the link).
I would also love to see a really high power variant of this, as yes there are those you can buy, but not an open source design you can modify and build upon.. Say 0-15V, 0-200 Amps. I loved this series as it really helped understand the design of it, design choices etc. But delivering 100Amps at 5V and still have good control would most likely hit other problems...
0-200A on 15v lol
Well, I fiddle with retro computing, old mini/mainframes from the 60s and 70s.. a cpu easily chugs 100amps themselves.. :)
a 200 amp partly linear design? ........well you better bring one hell of a fan and heatsink.
First now I found out you wanted this to be a batt powered PSU, genius is the main thing coming to my head, I think definitely I will play with your design and probably tweak it to my own needs. Sitting here just after watching this video I have two main thoughts, 1. A batt PSU is what I always have been missing, and 2. Why did I not think of this myself... For me, I would be okay with spending a bit more money, so I am thinking of actually building 2 in one to have the capability to run higher voltage, or higher current, or even +- voltage, of course it will maybe take more space, and more battery, but still, this is what I would like, that is the beautiful thing about open source, that people can make their own versions.
The idea is you could have several to series or parallel to get higher voltages or currents.
+EEVblog I'll buy 3 right now!
@barcelona08018 twice the capacity.. 2700mAh wouldn't last very long. Also more efficient running on 8.4V. Three (or four) would be better but he couldn't find a cheap charging solution chip and the DC-DC options are more limited if you have 12.6V input and a wide range of outputs (same for 4V input). If the input is 6-8.4V input you get a lot more options of efficient cheap DC-DC chips.
could you post arduino code
Will you ever do a series on designing a full on lab power supply? Something like 0-40V 0-10A?
***** 400W is not that much power. 10A is a decent current. I would actually aim for more than that if you plan on designing motor drivers.
Cristi Neagu I'm working on such a PSU (30V - 3A but you can crank it up to 10 no worries, just costs a bit extra for a fatter toroidal transformer). I think I have it all sorted out. But I'm kinda slow with KiCad. It's going to have as much as 3 outputs (only because of cheap 128x64 display), depending on how many you need (and afford for us Romanians :D ) and hopefully as much built-in software niftiness as possible and of course it''s all open-source. I'd say I'm one month away from pushing to Github or Bitbucket or something.
TechnocraticBushman That sounds great. Can you post a link here when it's done, please?
Cristi Neagu Sure. Better still, I'll subscribe or add you to a circles thing.
TechnocraticBushman Thanks.
Neat project but I don't see the point in having Ethernet at all, not sure why it’s even an option. I cannot think of a case where you would want to be able to change power settings remotely. I suppose it could be for logging current use or something but it would be easier to use serial for that. Its a clash with you original idea of having a portable power supply.
it's an IoT thing... you know Dave loves them ;)
I LIKE IT DAVE.
PRICE IT RIGHT AND YOU GOT ANOTHER BUYER !!!
@sifterjoe OR, he can design his own case or plastic LCD support and make it with Makerbot!!!
this rev.C schematic...i cant figure it out! sh*t just got real! keep it up Dave
Dave, I think you want the MCP73213 with the factory preset Timer Disabled. I was thinking what happens if you run the supply plugged into DC adapter and using say 1mA from the supply. Without the Timer Disabled version, the charging chip might time out and go into shutdown mode and then the batteries would just drain.
I think the 8.2V version is more robust for more cells you might get. If you get good 4.2 cells you are losing some capacity, but also extending the life a bit.
I am a bit confused, maybe someone can help. The "Sense" is for sensing the output voltage, but as far as I see, it runs directly to the Controller, which would fry it, wouldn't it?
Sense voltage is lowered with voltage divider to reasonable voltage for the ADC analog pin on the microcontroller, and those ADC pins have very high impedance, hence very low current going in.
OK thanks. seems like i haven't seen the voltage divider.
@ssj3gohan456 Yes, you get what you get...
if it's battery powered, why not replace the Ethernet with WiFi? :P
I like thath Idea!
Adding ethernet capability to a power supply is the same as adding an internal combustion engine to a microwave :)
Oh, Dave, wait, we have a problem.
You want to charge two batteries in series using only two(main) connections.
Why there is no balance lead for middle connection of batteries?
If we get this way, one of the batteries CAN EXPLODE!
Because if one battery will have bigger IR than second, it get discharged faster, eg. when we discharge batteries, we will get at no load 3,3V and 3V(this weaker) its 300mV difference.
When we get charging those two batteries in series, to 8,4V, we will get on each battery 4,5V and 3,9V!!!
One of the batteries can be very overcharged!!!!
It can explode!!!!
Yeah you want to get protected batteries, but someone can use unprotected one!
Anyway, if we will get protected batteries, one of them will be not charged to 100%
+Arek R. he is using protected bateries, he said that the are a bit longer becose they have a protection pcb
Still bad, charging will be not finished on one of the batteries.
He finished that PSU anyway?