EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: paulca on November 24, 2017, 02:14:54 pm
-
So I'm thinking my next project would be a DC load.
I love the look of this:
https://www.instructables.com/id/Arduino-Programmable-Constant-Current-Power-Resist/ (https://www.instructables.com/id/Arduino-Programmable-Constant-Current-Power-Resist/)
However I think it's a little beyond me. To build it I would need to order the PCB and it's all surface mount, some of the components are harder to find in the UK so I am likely to make many mistakes.
What I need is something a little lower level, but can be extended to have PWM control from an arduinno or PI later and ADCs for sense and display.
As I want to test LiPo charger circuits etc. I need something with a decent bit of load, something like 100W.
What about the one Dave outlined in his video, would I be better off trying that one out?
Should I just start at the business end with a chunky mosfet and sense resistor set and work back?
Any defacto threads on here for similar projects?
-
Look into the YouTube channel ScullCom. He does a bunch of wonderful videos and has a pretty good series on designing and building a DC load.
Sent from my Pixel XL using Tapatalk
-
That looks good, thanks. It looks like he does the development iteratively which means I can get something running fairly quickly and then follow his later videos to improve upon it.
Thanks.
-
Peter Oakes has a couple of Youtube videos showing similar:
The second video shows some performance improvements regarding response times.
There's also 2 Blog posts on Element 14 with comments and schematic for your viewing pleasure
https://www.element14.com/community/groups/test-and-measurement/blog/2015/06/02/electronic-dc-load--design-and-build-to-test-psu-project (https://www.element14.com/community/groups/test-and-measurement/blog/2015/06/02/electronic-dc-load--design-and-build-to-test-psu-project)
https://www.element14.com/community/groups/test-and-measurement/blog/2015/06/09/electronic-dc-load--performance-improvements (https://www.element14.com/community/groups/test-and-measurement/blog/2015/06/09/electronic-dc-load--performance-improvements)
https://www.youtube.com/watch?v=vd5IBFFjnOc&index=23&list=PL_atu5RtEPi4aNzoMtZ5_S6ruhFR98T_p (https://www.youtube.com/watch?v=vd5IBFFjnOc&index=23&list=PL_atu5RtEPi4aNzoMtZ5_S6ruhFR98T_p)
https://www.youtube.com/watch?v=rh32ylmlz-A&list=PL_atu5RtEPi4aNzoMtZ5_S6ruhFR98T_p&index=24 (https://www.youtube.com/watch?v=rh32ylmlz-A&list=PL_atu5RtEPi4aNzoMtZ5_S6ruhFR98T_p&index=24)
-
Here is my version of Peter's design. I choose components to limit the load to 2A.
EDIT: It definitely needs a fan for currents over 0.5A
-
Basic question time.
Load sensing. I get the gist. Ohms law. If a resistor of a given value has a given voltage drop across it then the current can be calculated. Put that voltage into an op amp of some form and provide it was a reference voltage and it will alter the control output (to the MosFet) effectively limiting the current to the set value (the reference voltage on the opamp).
However, does this not mean you are always putting the full current load through the sense resistor? So if you are trying to limit to something like 20A at, say, 15V you need a resistor capable of carrying 15*20=300W?
Is there not a way to decouple the sense circuit somehow, by, say, dividing off 1/100th of the current through a 1K or 10K resistor and then multiplying it with a gain of 100 on the opamp?
The other thing I get a little confused about is, does the current sense resistor not then alter the actual output load current/voltage? (Assuming for this it's not a dummy load, but limiting current to an actual load)
EDIT: I realise I'm confusing two different things. In a dummy load I *must* consume all the current, that's the point. In a current limit power supply I instead want to alter the load current and voltage as little as possible so I would sense differently.
-
In Peter Oaks circuit. Between the highside + and - there are only the mosfet and the sense resistor.
If I plug in some numbers at the sense resistor:
r=0.1(ohms)
i=5(amps)
v = i*r
v = .5
p = iv
p = 2.5w
But... if I connect a 12V supply (that I want to load) to V+ and V- there is 12V to drop, the sense resistor (if we have 5A) is only dropping 0.5V, so the mosfet must be dropping 11.5V.
The mosfet therefore is taking 11.5V * 5 Amps = 57.5W
Now I understand why the heat sink is on the mosfet. What I don't understand is how the mosfet voltage drop can be changed.
If I think of the MosFet as a resistance then at 5Amp, 12V it is presenting a resistance found by:
r=i/v
r=5/11.5
r=0.43Ohms
If I run those with 12V and 1Amp
Sense resistor drop:
r=0.1 Ohm
i=1 Amp
v = i*r
v = 0.1V
p = v*i
p = 0.1 Watts
Mosfet:
v = 11.9
i = 1
r = v/i
r = 11.9 Ohms
p = v*i
p = 11.9 watts
So if I wanted to run it at 20A I would see:
rSense = 0.1 Ohms
iSense = 20
v = i*r
v = 2V
p = i*v
p = 20*2
p = 40W
vMos = 10V
iMos = 20A
rMos = 0.5 Ohms
pMos = 200W
I think I see now. Still need to working on exactly how the voltage drop changes across the mosfet, expect the whole op amp thing that says "It must" for the maths to work.
-
The idea in Peter Oakes design is that he applies a 0-500mV voltage to the positive op-amp input. This corresponds to a 0 to max current through the load and sense resistor. The op-amp supplies the necessary voltage to the MOSFET in order to have 500mV across the sense resistor at max load.
In your example for a 0-20A load at 15V, the value for the sense resistor is 500mV / 20A = 0.025 ohms with a power rating of at least 10W. The power dissipated by the MOSFET is (15V - 500mV) * 20A = 290W.
-
Dave did a dummy load along the same design. A little different perspective my help.
https://www.youtube.com/watch?v=8xX2SVcItOA (https://www.youtube.com/watch?v=8xX2SVcItOA)
-
I think I can set something up to play with what I have now. Just at mA current.
Using a basic NPN transistor instead of a mosfet and aiming for 200mA (the limit of my transistor), I could in theory use 5x10R resistors in parallel to create a 2 Ohm sense resistor. The sense voltage would then be 400mV. I have an LM741 op amp.
The PSU will limit me to 200mA when (not if), when I mess it up, but I might have a play while I watch the rest of the videos and order some mosfets and sense resistors.
-
Dave did a dummy load along the same design. A little different perspective my help.
I watched this at least twice, but the bit I was always missing is my epiphany above. I was always doing the sums as if the full voltage drop was across the sense resistor and wondering why it wasn't on fire. When run Ohms law from the "known current" perspective on the sense resistor it starts to make more sense.
-
The sense resistor sees all of the current but only a fraction of the voltage which is just as well because heating of the sense resistor and its temperature coefficient of resistance is what ultimately limits accuracy.
-
Out of interest the attempt at using NPNs and a load of 10R resistors ended badly with one dead transistor and some funky smells :)
Using the circuit from Wikipedia.
https://en.wikipedia.org/wiki/Current_limiting
It's possibly because I didn't have an actual load and was just trying to use a 5k pot as the load. So the resolution around the "couple of ohms" load was too small and if it went to sub ohm the circuit stopped working and collapsed to a short circuit heating the transistors up.
Anyway it was incredible hard to balance things with such dicky little transistors and using the 0.6V base to emitter voltage across the sense resistor.
I did get it to limit a couple of times, but at around 250mA but then it quickly collapsed to a short as the NPNs over heated.
A 2n2222 bought the farm in my last attempt. I don't think it liked 1 amp. I wasn't paying attention, didn't notice the CC limit light on the PSU, forgot I had dialled it up to 1 amp, but I noticed the smell. Poor thing. It now identifies as a couple of resistors or a diode and a resistor in the component identifier gizmo.
-
So I got most of my bits in the post for the dummy load. No enclosure yet, but I want to breadboard it for a while, then decide on perf board and fan/heatsink or what-not and only then will I know it's size. Ordered to Peter Oakes spec. At some point I'll need to consider a meter on it too.
I also got a much smaller 12A mosfet and I thought I would try a baby 1A load first.
So I did my sums. I need 4V to turn the mosfet on. I tested that and it starts to conduct a bit lower than that, but is more or less fully on at 4V. So I need 4V out of my op amp max.
The 0.1Ohm resistor at 1A will only have a 0.1V drop and 0.1W of heat. Therefore I need a gain of 40 on the op amp.
I only have one of the nice opamps that Peter uses at they were a few quid, not pence, so I don't want to cook it accidentally until I have experimented. So if I want to use a cheapo LM741, it's minimum output for a 0-9V rail is about 1.3V and it's maximum around 8V. So it should be fine without splitting the rails. Assuming the mosfet doesn't start to conduct at 1.3V. I have a hard time trying to read the various log graphs on the mosfet datasheet.
Anyway. If I connect a 3S lipo to the mosfet the voltage (from storage charge) will be 11.3V (ish). Meaning a 11.2V ish drop across the Mosfet. 11.2*1A = 11.2Watts. I have the a small PC chipset heatsink I ripped out of an old server that I'm sure will take 12 watts if I bolt the mosfet to it.
Can anyone see why this won't work?
I'll test the opamp gain first for 0V to 0.1V input and make sure the LM741 will respond to 0.1V properly and give 4V out. Then I'll connect it to the gate of the mosfet. I'll keep the DMM measuring current out of the LiPo so I can pull the plug if it goes run away. The LiPo is question shouldn't bat an eye lid at delivering 50A if I let it.
Could I use a LM393 comparator instead if the LM741 doesn't work or would that make things more complicated?
-
I believe this makes more sense:
http://tinyurl.com/yd74fhye (http://tinyurl.com/yd74fhye)
Adjustable between 9mA and 2A.
-
Here is the circuit and actual parts I used for my version of Peters circuit:
- IRFP064PBF (https://www.mouser.com/Search/ProductDetail.aspx?R=IRFP064PBFvirtualkey61370000virtualkey844-IRFP064PBF)
- TLC272ACP (https://www.mouser.com/Search/ProductDetail.aspx?R=TLC272ACPvirtualkey59500000virtualkey595-TLC272ACP)
- REF02CP+ (https://www.mouser.com/Search/ProductDetail.aspx?R=REF02CP%2bvirtualkey66880000virtualkey700-REF02CP)
- TO-247 Heat Sink (https://www.mouser.com/Search/ProductDetail.aspx?R=RA-T2X-64Evirtualkey58810000virtualkey588-RA-T2X-64E)
- Precision Pot 10Kohms 10turn (https://www.mouser.com/Search/ProductDetail.aspx?R=3590P-2-103Lvirtualkey65210000virtualkey652-3590P-2-103L)
- Sense resistor 0.10 ohms => 5.0A max load (https://www.mouser.com/ProductDetail/Vishay/AC05000001007JAC00/?qs=%2fha2pyFaduggcyyn9%252boyOBkvVxQLACEEiKV8RXHs5bbpn55k0ZLzHw%3d%3d)
- Sense resistor 0.24 ohms => 2.5A max load (https://www.mouser.com/Search/ProductDetail.aspx?R=AC05000002407JAC00virtualkey59420000virtualkey594-AC05W0R240J)
- Sense resistor 0.50 ohms => 1.0A max load (https://www.mouser.com/ProductDetail/Vishay/AC05000005007JAC00/?qs=%2fha2pyFaduggcyyn9%252boyOHv5tbisgfxEsBleiOMuyYE7SjvqbdF3sQ%3d%3d)
- 40x40x10mm 12V Fan (https://www.amazon.com/gp/product/B01GPEQZ8A/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1)
I did not choose exact value and small tolerances for the resistors because I don't care about precision settings. My capacitor values are a little larger than Peters because I don't need a high frequency response.
With the heat sink I used and NO fan, I reached the maximum power of the MOSFET with a 1A 15V load for a short duration. I believe with the small fan listed above, you could reach a 2.5A load. Peter's 5A load will require a much larger heat sink.
Also, I believe Peter made a mistake with the 5V reference adjustment pot (R2). According to the datasheet, it should be tied to the output pin instead of the input pin.
(https://www.eevblog.com/forum/projects/check-out-my-electronics-load/?action=dlattach;attach=755808;image)
-
I am planing on using a CPU heatsink with it's fan for the mosfet. I'm not sure of the power rating. It was originally sinking an AthlonXP 2.4GHz a quick google suggests that CPU was designed for 60-70W dissipation.
Given a use case of a 3S LiPo capacity tester at 12.6V that gives me a maximum theoretical load of 70/12.6 = 5.5A (or 70/(12.6-0.5)=5.7A) So if I follow Peter's circuit directly (which I believe is for 5A) I should be fine. But will need to keep an eye on the mosfet temp.
I even ordered drill bits to drill a hole in the heat sink for the Mosfet bolt.
Next thing I want to research is thermal monitoring and possibly limiting. I have worked with Dallas 1-wire sensors and the raspberry PI and if I implement the control signal via it's DAC I could thermal limit based on a 1-wire, but I'm sure there are other options.
I doubt I will run it at 5A for very long. To test a 5Ah LiPo it would be "nice" to run it at 5A, but I would settle for half that.
-
Given a use case of a 3S LiPo capacity tester at 12.6V that gives me a maximum theoretical load of 70/12.6 = 5.5A (or 70/(12.6-0.5)=5.7A) So if I follow Peter's circuit directly (which I believe is for 5A) I should be fine. But will need to keep an eye on the mosfet temp.
If your target max load is 5A, that is fine.
If your target max is eventually something smaller, you will have finer control if you use a larger sense resistor.
Testing with the sense resistor value you will use in your final design will give a feel on how your load is going to respond in general use at varying currents and voltages.
-
In looking at the ref02C spec, it indicates that the min sink current is 8mA. The 10 turn adjust pot should be something like 500 ohms. However, my load is working okay with a 10K ohm pot. Maybe a smaller pot then Peter and I used would be better.
-
In looking at the ref02C spec, it indicates that the min sink current is 8mA. The 10 turn adjust pot should be something like 500 ohms. However, my load is working okay with a 10K ohm pot. Maybe a smaller pot then Peter and I used would be better.
I have to watch his videos again, but there are two pots aren't there? The main control pot and the fine adjustment on the vRef?
I bought a large wire wound pot from ebay for the main adjust pot (it's 50K) and I have a few different 10 turn trim pots, I think I have 1K and 10K.
I'll have to rewatch his video later to revise, I'm in work so it's awkward to watch youtube atm.
I hope to create the smaller <1A load this evening or tomorrow evening to get a feel for things.
EDIT: I can see in the video preview his schematic shows both trim and adjust as 50K pots. I can make the trim a 1k, 10k, or 50K is I need to.
-
I have left the trim pot unpopulated on my PCB. Depending on the precision you are trying to achieve, you may not need the trim pot. I just dial in the current I want by watching the current meter.
I would start with a 1K ohm 10 turn pot for the main adjustment.
Attached is Peter's first design and his new design that I haven't seen him finish. Remember, Peter burnt up his original design.
-
I let some smoke out. Only a little. My breadboard wires didn't like 4A.
I've been gotten by the b****rd LM741 again. The thing is so annoying. No matter how well you plan a circuit out, no matter how much maths and simulation you do, the LM741 has some form of limitation that makes it do the exact opposite of what you wanted.
After the 4 amp incident I powered the circuit from a 9V battery and used the current limited PSU to feed the hot side at 1A limit.
So I probed while the limiter kept things fairly sane.
Non-Inverting input: 65mV - Set voltage for 650mA.
Inverting input (from the sense resistor): 100mV (1A sensed)
LM741 output: 8.08V
WTF? Have I jumped into an alternative reality or did I completely miss something? The inverting input is higher than it's non-inverting it should be driving the output to the lower rail. Why it's pinned to the upper rail?
I put the lm741 into voltage follower mode, putting it's output directly back onto it's inverting input. 1.9V was the lowest it would go on the output. I tested by winding up the non-inverting input until it finally unpegged and started climbing once I passed 1.9V
So not only is it's output limited to something like 1/4 Vcc above the lower rail, it's inputs are also limited so something considerably higher than 100mV. In my tinkering with different sense resistors and cooking a few 10 Ohm resistors till they burnt my fingers, I could not get the op-amp to respond, even at 1.6V inputs, the two inputs stayed apart and the output stayed pinned at 8V
Do you think microwaving an LM741 will help? By that I mean, do you think it will make me feel better?
On the brighter side the LiPo should have been capable of 10A+, which might have cooked the 12A mosfet, my DMM fuse and melted my jumper leads and possibly smoked the LiPo itself. So there are positives that it only put out 4A (in fairness it's only at storage voltage and it's only a doddy wee 500mAh 3S pack.
Do I really want to start splitting the rails, coupling caps and that lark or finding a better op-amp?
-
The LM741 is NOT a single supply op-amp. Nor is it a rail-to-rail op-amp. Its output will not be able to get low enough to control the MOSFET. You will need to bodge together a dual supply for the LM741 or use a single supply op-amp. Use a TLC272 or Dave's LM324 or a LM358 for testing.
-
The LM741 is NOT a single supply op-amp. Nor is it a rail-to-rail op-amp. Its output will not be able to get low enough to control the MOSFET. You will need to bodge together a dual supply for the LM741 or use a single supply op-amp. Use a TLC272 or Dave's LM324 or a LM358 for testing.
Thanks.
I had realised after much frustration on previous experiments that the LM741 was not rail to rail on it's outputs, but it could go as low as 1.3V, so I figured it's under 2V it might do. What I did not expect is that it's inputs pretty are far off the rails too. Nor did I expect it would slam to the top rail when it's inputs where out of range.
I've ordered a few more wider range opamps, the LM324 and a few LM358s. I also have the TLE2142 from Peter's design which has rail to rail inputs and a Vcc- +0.1V minimum output, which is well under 2V obviously. I'll play with the 2142 tonight and see if I can keep the smoke in.
I also realise that I was lucky testing it with such a small weedy little LiPo, if I had of wired in a much more beefy LiPo the wires would probably have went red hot, let much more smoke out and probably set the building fire alarm off to my great embarrassment. I'll test with the current limited PSU until I'm comfortable it's tamed.
-
It works! :o
The baby load V1 works. Minimal circuit, just the opamp, sense resistor, mosfet and a voltage divider pot to set the input. The key was using the TLE2142 which has rail to rail inputs and close to rail to rail outputs.
Mosfet on a PC chipset heatsink (about 1.5"x1.5"x1.5") attached with one the those little plastic spring clips. The only high side stuff that went through the bread board was the sense resistor and the ground return.
I fired it up at 1A, 12V and it got too hot to touch the mosfet after about 30 seconds, a few minutes longer and the heatsink was too hot to hold onto for long. The sense resistor was cool enough to hold onto. The breadboard jumper wires groaned and smelt funny, but I could still hold them and they weren't melting.
I put a cooking probe on the heatsink and it got up to about 67*C. I used an IR temperature probe to see what the mosfet case was and the maximum reading I could get was 80*C.
However it burped and cut out a few times then stabilised again, I'm not sure if I tripped the mosfet thermal protection or breadboard shenanigans but I decided it was probably running too hard without a fan and backed it down to 0.5A and noted that the heatsink temp fell slowly to about 57*C and stayed there for another 5 minutes.
So I recon it can handle 0.5A with 1A short bursts. With a fan on the heatsink it might do 1A sustained.
This version won't give me an ability to cycle a LiPo in a sensible time, my Accucel6 can sustain 0.5A. I want something a bit more beefy. However this one might still be useful from time to time for loading electronics circuits up.
For the LiPo capacity tester I will need to build the full Peter Oakes 5A version with a CPU heatsink and fan with thermal grease and a proper bolt on the heatsink. Then I'll see how many amps it can sustain.
Thanks to everyone who helped and put up with my frustrations and stupid questions. Couldn't have done it without you.
-
I had no idea I was being talked about over here 8)
Glad the project is useful to you, I have another collaboration right now on Element14 for a full software controlled load, PCB designs and everything, you and others of Daves fans may find interesting
https://www.element14.com/community/docs/DOC-88209/l/programmable-electronic-load-adc-and-dac-boosterpack-test (https://www.element14.com/community/docs/DOC-88209/l/programmable-electronic-load-adc-and-dac-boosterpack-test)
-
DAC/ADC control and metering is on my radar, for further down the line, thanks Peter.
On a different note:
Would this work or be naive for input protection?
5A Self-reseting fuse
https://www.bitsbox.co.uk/index.php?main_page=product_info&cPath=214_215_222&products_id=1534 (https://www.bitsbox.co.uk/index.php?main_page=product_info&cPath=214_215_222&products_id=1534)
I am more interest in protecting the power source than the dummy load in this case. Say for example I am draining a LiPo battery and my mosfet fails to short circuit. I would like, if at all possible to avoid the resultant LiPo explosion. That might suck somewhat.
If not that, a traditional 5A plug top fuse or 5A cartridge fuse?
----------------
On a different topic any basic pointers in the direction of thermal limit circuitry? I gather for a hard limiter a thermistor used in a voltage divider feeding a comparator can result in a high output up until the temperature hits a limit and then the comparator drops to a low output. The question would be what to do with that high/low output. If I put it to the base of a NPN 3904 in between the drive op amp and the mosfet would it
a) be able to cut the control signal to the mosfet to below 2V ( I expect so )
b) interfere enough with the drive feedback to alter the performance of the device
Also, if I removed the comparitor and drove the 3904 with the termistor voltage divider could it be made to soft limit, ie increasingly backing out the mosfet gate voltage above a certain temperature.
Would applying a voltage from the low side supply rail into the feedback of the op amp via the 3904 work? So if the termistor divider rises over 0.6V the 3904 will start feeding extra voltage into the sense feedback line backing the opamp and thus mosfet off.
-
Would this work or be naive for input protection?
5A Self-reseting fuse
http://www.mouser.com/ds/2/240/Littelfuse_PTC_Rline_Catalog_Datasheet.pdf-1021735.pdf (http://www.mouser.com/ds/2/240/Littelfuse_PTC_Rline_Catalog_Datasheet.pdf-1021735.pdf)
So if I am reading the datasheet correctly this would not work for my needs.
A polyfuse rated for 5A hold would not actually trip until it gets to 10A or so and even then it could take 14 seconds to trip (at 25A!). 14 seconds at 25 amps, is probably fine for a LiPo, but it seems the device itself is only rated to 100A MAX, when I assume it detonates or otherwise fails.
They also have horrible temperature derates meaning the 5A hold fuse could actually begin to trip much lower if it gets hot near the Mosfet (or just by the current flowing through it). So the dummy load could cut out as low as 4A after a prolonged period trips the poly fuse.
The poly fuse could be used as a much lower sustained current limiter, to allow bursts of higher current but I expect for the upper short circuit protection I need an actual fuse.
-
Is this an anomaly with the Falstad sim or am I having a blonde moment?
I found the idea from a thermistor product site, they had a circuit which pulled the mosfet gate low using a thermistor if it reached a thermal limit.
When I attempt to simulate it, it doesn't matter how hard I open the NPN transistor the gate stays high and open.
http://tinyurl.com/yajtt862 (http://tinyurl.com/yajtt862)
Here is the product circuit:
http://www.amwei.com/news.asp?news_id=89 (http://www.amwei.com/news.asp?news_id=89)
What did work in Falstad was reversing the idea and adding voltage into the sense loop. However for that to work I had to put a diode on the sense line, but falstad did not simulate the forward drop across the diode. If the forward drop proportional to current flow? There is obviously close to zero current flowing on that line.
-
If the forward drop proportional to current flow? There is obviously close to zero current flowing on that line.
Yes, the forward drop is totally depending on the forward current. It starts at zero and can go up to over 1V. See datasheet from your diode. That does not go lineair. The current from the diodetest in a DMM will result somewhere between 0,6 and 0,8V depending on the type of diode and the current from your meter.
Also, a diode needs some current, so with no load and a very high resistance input DMM (>1Gohm) you will measure some fantasy value because there is no current flow, and no current is no voltdrop.
-
Yes, the forward drop is totally depending on the forward current.
So in theory this should work without altering the sense voltage?
http://tinyurl.com/ya37wn7l (http://tinyurl.com/ya37wn7l)
EDIT: I'm baffled, this shouldn't work. The 100mV sense voltage doesn't come close to the 800mV forward voltage on the diode.
-
Anyway, phase two was a success. Using the big mosfet and the CPU heatsink.
Gotta love AMD OEM Coolers. It's like a small block laptop cooler. Using my "Lighter Putter Outer 100" from my other tongue in cheek post to run the heatsink fan and everything worked the first time I plugged a load into it.
I even trusted it enough to connect my biggest LiPo to it. 5Ah 25C 12.6V (capable of an alleged 125A). SAFETY TIP!: Along with a LiPo analyser which gave me a loud alarm if the pack unbalanced or a cell dropped below 3V.
With the ExTech monitoring true current I trusted the Chinese battery analyser to watch the voltage and cell volts. It's current reading disagreed with the Extech by 400mA, so I ignored it.
Very impressed by the AMD cooler. With my kitchen temperature probe stuck in the heat grease beside the mosfet the highest temp I recorded was 37.4*C with the load pulling 3.0A It was able to consistently hold that temperature for half an hour at 3.0A while the lipo discharged and I could see the temp fall by about 0.1*C for each few 100mV the LiPo dropped by. Even at 3 amps I could comfortably keep my finger on the mosfet case.
The little thermistor lighter putter outer worked perfectly. The thermistor all insulated up and stuff into the heatsink core. When I put the load on the fan started slow after a few seconds, got to full speed fairly quickly, drawing 100mA. When the load was shut off the fan continued to run for 10 or 20 seconds before shutting off. When the load was running at only 0.5A the fan ran at about 50%. I have no idea how I managed to randomly set the thing perfectly without trying, but it pretty much is perfectly tuned. If anything the fan could be backed off a bit more.
I have days when things always fail and I get frustrated. Today, everything worked. Pretty much the first time I plugged it in and tried it, it worked.
There are bugs however. The PNP running the fan is sinking 4-5mA for no good reason when it should be off and the fan shouldn't be that variably when it's being driven from a LM393 comparator, it should be a hard on/off. I'm not sure I want to fix the latter, but I will probe the thing to see what's going on. Also when I touch the breadboard or even the heatsink I get another 200mA current, so I have open noise being introduced from my hand somehow. (I really should probe it!)
I couldn't test it to 5A as I'm still using just a voltage divider with a trimmer and 3A was all I could get it to go to. 3A actually required I power it from 15V which kinda upset the PNP driver for the fan. It was merely toasty.
Next plan is to implement the better control circuitry with the opamp and voltage reference.
Other things I learnt today.
1. Using overrated cables (12AWG) makes soldering connections difficult. My 25 watt Weller struggled and I have several burns on my finger tips. 14 or even 16AWG probably would have done!
2. Soldering single core wire to mosfet pins is a nightmare. They don't flex when the solder takes so any movement of your hand at all when you release the heat and the joint is bad. Later I found it is better to strip back much more single core and wind it around the pin first, then just fill the winds with solder. I might redo the gate joint this way.
Busy day:
(http://i.imgur.com/3IpnBWc.jpg) (https://imgur.com/3IpnBWc)
The circuit:
(http://i.imgur.com/ueNfVuL.jpg) (https://imgur.com/ueNfVuL)
The high current side:
(http://i.imgur.com/8wGbtYq.jpg) (https://imgur.com/8wGbtYq)
* I'm using the CPU cooler's spring bracket and a high strength carbon fibre strut bar (from an RC heli) to hold the mosfet in place. It is putting sizable (enough) force on the mosfet.
The Fan:
(http://i.imgur.com/oK2oRmu.jpg) (https://imgur.com/oK2oRmu)
-
Now I got to torture my bench supply charging that big LiPo. The Tenma 30V 5A single version. It's fan is very noisy when running at 5A but it worked for the 45 minutes or so it took to get the LiPo up to 12.6V and go to constant voltage.
Safety first, I had a cell balancer attached so I could see the individual cell voltages and let it attempt to balance them at the same time.
I had to pause the charge a few times to allow the cell balancer to catch up. Once one cell got to 4.21V I killed the charge current and waited on the balancer to do it's thing, then restarted the charge.
The cell balance is interesting to watch. With:
4.20V 4.15V 4.17V reading,
It would switch into balance mode using the the higher cell to charge the lower cells and read:
4.17V 4.16V 4.17V for about 30 seconds, then it would pause and sense the cells for a while.
4.20V 4.16V 4.17V reading,
Back and forward between cross charging and sensing. Until finally it read:
4.18 4.18 4.18 the "BALANCE END" buzzer went off and I restarted the charge current.
EDIT:
Now the downside. While the power supply can charge the pack and the analyser can balance it, the balancer struggles to keep up, even when the PSU is back to 0.8A. So I have to keep pausing the charge and I suspect as I get closer to full charge it will be more and more often.
The Accucel will come out to finalize the balance charge. It has smarts in it which will not only lower the charge current based on how out of balance the cells are, but it can also actively charge individual cells (I believe).
-
Got the full Peter Oakes design version running on the breadboard this morning.
5A sustained resulted in a heatsink temp of 48*C stable (kitchen probe) and a max measured surface temperature on the mosfet of 61.3*C (using an IR thermometer). This was after 5 minutes sustained.
I had/have an issue with the trimmer on the Vref. My trim pot maxs out and I can still only get 4.985A (though I'm not really complaining), I would have expected the trim on the Vref to allow adjustments below and above the ref.
I also 'fixed' my fan controller. Took out a load of motor noise with a 100uF cap across the motor feed wires and took out a load of bouncing/ringing/osculation on the comparator at transition with a 1uF bypass cap. I also swapped round from using a PNP and sinking to the comparator output and used a pull up and an NPN as most of the datasheet applications choose this approach. I could run it in hysteresis but I would need to carefully adjust my thermistor voltage divider and reference divider which I'm not sure I can be bothered for a fan controller. The downside is it no longer soft starts and no longer winds up and down at marginal temps, it's now a hard cut on/off. So it pulses every few seconds for a minute or so when cooling after a hot run. Not sure which I prefer. The bouncing was effectively adding PCM control to give it a softer switch on at lower temperatures.
Next.
Some form of metering for voltage and current. I know this might destabilise the sense side, but I'm not really aiming for that great a precision. Knowing the current and voltage of the load without having to use a DMM is more useful to me. The downside is the only ammeter I have claims to be 75mV per 50A. I think I might start a new thread to ask about this as it seems unreasonably low.
-
Heat. Did some sums. First figuring out what the sums where. Then trying to find values for them.
I couldn't find a thermal resistance for my AMD cooler, but given the "Thermal Design Power" of the chip it is meant to cool is 68W, I should be okay at 5A 12V.
I also determined that even if I bought a £100 PC water cooler with dual 12" radiator (thermal resistance of circa 0.1 C/W) I could still only push it to 120W or 10A 12V. Higher and higher priced units shaved another amp tops, but started to cost £200. I can also get a little bit more by going to the 100A Mosfet as it seems to have better junction to case and case to sink TR, but the downside is, it has a lower max Jt.
So, staying with the basic CPU coolers, if I want more amps I will have to go parallel mosfets each with a CPU heatsink. A task for way, way down the line. 5A is grand for what I need right now.
The 5 gallon heat sink bath of vegetable oil just sounds like a smelly mess waiting to happen.