Help with PSU (Temp control fan, load bank, & PWM circuit)

[()blivion]

I put [the heat sensor] on the current sense resistors (because those get HOT)

The current sense resistors should get somewhat hot, but not TOO hot. Are ALL of them getting hot evenly? If something isn't working right then more load could possibly be getting dumped into each resistor than we planned. Or more into some resistors and less into others. But yeah, my sense resistors did get pretty hot too, and that was at only 100 Watts or so. In the end. they should be able to survive 6 Watts of heat, which at 0.130 Ohms that would be almost ~6.5 Amps per resistor, or ~65 Amps for 10 sections. At full load, we are pushing that limit.

How hot can those safely get?

105c on paper. In real life... they are pretty tough devices. If it will burn the crap out of you or boil a drop of water, Tis too hot. You could try your hand at some kind of cooling. Attaching a heatsink like device to the resistor bodies would help cool them. Gota be careful how you do it though so you don't end up shorting something out.

Anyway, in summary, awesome job guys! This thing is great!!! I will post up some pictures tomorrow, it is too late tonight and I have to go to work in the morning. Thanks!!!!!

No problem, glad you like it. In hind sight I would have made the sense resistors a lower value probably. Somewhere between 0.05 Ohms and 0.1 Ohms would be better I think. We do want some power to burn up in the resistors though, this lets them act like fuses in an emergency.


Edit: We do want the heat sensor on the FET's, it is mainly going to protect us from accidentally powering the load when there is no water going through it. If the FET's get destroyed from any kind of over power they probably won't even get remotely hot with the water running.

 

[jocanon]

They seemed to be all getting hot evenly. I did not get them up to 105c, I didn't know they could get that hot, I only let them go to however hot they could until they tripped the over temperature sensor (I temporarily put the temp sensor on the current sense resistors because I was most worried about the heat there). None of them were too hot to touch and they all seemed about the same temperature by feel. If they can get all the way to 105C then I think I might be OK, plus I will have fans on them once I get everything buttoned up.

 

[jocanon]

Question: How could I test each MOSFET individually to see if they are all bearing the load equally? Could I touch the leads of a voltmeter somewhere to test this out?

 

[jocanon]

Edit: We do want the heat sensor on the FET's, it is mainly going to protect us from accidentally powering the load when there is no water going through it. If the FET's get destroyed from any kind of over power they probably won't even get remotely hot with the water running.

Yes, I agree...I just put it on the current sense resistors temporarily because I wanted to test out the over temp protection and see how hot the current sense resistors were getting. When I get everything completely assembled I will put it right next to one of the FETs.

 

[jocanon]

I just noticed on Mouser they say the operating temperature is 275C for the current sense resistors...that's hot! If they can go that hot, I really think I will be fine. In retrospect I wish I have mounted them a little off the PCB, i.e. space between them and the PCB...oh well...I don't think the heat should bother the PCB much, right?

reference for 275C
https://www.mouser.com/Search/Produ...00virtualkey59420000virtualkey594-AC05W0R130J

edit:
sorry for the quadrouple post !

 

[()blivion]

Question: How could I test each MOSFET individually to see if they are all bearing the load equally? Could I touch the leads of a voltmeter somewhere to test this out?

If you touch the multimeter probe from the joint between the MOSFET's and the sense resistor, and measure that referenced to ground. Then you can figure how equally current is getting shared by how close the voltages are to each other. The only problem with this theory is if the sense resistors have a significant difference in resistance. They are 5% tolerance though so the worst any two could be off by is 10%, or 0.013 Ohms. If the circuit is working at all, then it's 99% certain that power is getting shared reasonably evenly. The only thing to watch for would be if you had one of the feed back loops wired wrong and it held that FET at a lower resistance than any of the others. And if it did, it would 9 times out of 10 be a full short, which would be blatantly obvious because it would trip the PSUs over current protection.

It would be very odd for it to be anything else.

I just noticed on Mouser they say the operating temperature is 275C for the current sense resistors...that's hot!

WOW!!! that must be military temprature grade then. That's not completely unexpected, they are power parts after all. The main standard temps are commercial, which is 70c-80c, and industrial, which is 80c-105c. Then there is military, buy they are usually rare and pricey. But these are just resistors, so it's not like they are made with rare materials or tricky chemistry or anything. Actually, they are built quite simply. They have a ceramic core, prolly NiCr resistance wire around that, the copper leads are welded to the wire, then a cement dipped coating over the whole thing.

So it's basically rock, metal, then rock again. I have never overheated a rock before.

 

[jocanon]

Here are some pictures of what I have so far. I am not done. I still need to glue the temperature sensor onto the copper pipe next to the FETs, put the whole thing in the CPU case, attach the volt meters, ect...I got it assembled this far last night so I could test it out and be sure it was working and it works like a beauty . Btw, go easy on me...this is my first project of this kind...so I am sure I did some things wrong in the way I soldered and what not, but at least it is working so I couldn't have done everything wrong . But do point out ways I could have done it differently/better because I want to learn. Thanks!

Edit:
I was just noticing that the one current sense resistor is very close to the wood and since they can get so hot I am going to cut some wood away from it so it is not a fire hazard.

 

[jocanon]

With wood carved away a bit. I think the polyurethane coat will help it not be as likely to catch fire too.

 

[ronv]

AMAZING!!!

One thing the dummy load did unexpected was when powering down the alarm beeps a couple times, not a big deal as long as it is not hurting anything, I doubt it is.

Missed that one.. I don't think it hurts anything. The temperature sensor output is good from 4 to 20 volts. So when the voltage divider on the comparator drops as the 12 volts falls the comparator thinks the temperature is to high when in fact the referance is to low.

Do the volt meters work ok?

Looking good!

You picked this up really fast. Good job.

 

[()blivion]

Well... It's not exactly bringing sexy back, but I bet it can load up a power supply just right. Well done.

You still guna mount it in a case and send us pictures of that right?

 

[jocanon]

Thanks guys! Yes, I will send pictures once I get it all done. I haven't hooked up the volt meters yet...hopefully I can do that soon. I got some readings from the FETs source to ground voltage:
On 20 amps (I will list the FET number then the voltage like 1-.29 means the first FET had a reading of .29 volts), so 20 amps = 1-.29, 2-.29, 3-.29, 4-.31, 5-.31, 6-.31, 7-.32, 8-.32, 9-.33, 10-.33

Then with 47 amps 1-.69, 2-.69, 3-.70, 4-.70, 5-.71, 6-.71, 7-.72, 8-.72, 9-.75, 10-.75

Note both the 20 amp test and the 47 amp test were done on 24 volt PSU as opposed to a 12 volt PSU.

I think that is pretty close. What do you guys think? I wonder why voltage increases steadily as you go from FET 1 to 10? Is that amount of variance anything to get excited about or is it small enough of a variance to not be concerned with?

 

[()blivion]

Very odd results. Your ground must not be great. Those results look like you are getting extra resistance added in as you get physically farther and farther from where your ground probe is. Also, if the voltages are to small, your meter may cry about it. (By cry, I of course mean give inaccurate results)

How about this instead. Put the positive probe on the pipe, right next to each FET. And put the negative probe on the source of the same FET. This should, (A) Make any stray resistance marginal, and (B) Provide a large enough voltage that there is little chance we are running the meter in a non accurate voltage range.

 

[jocanon]

Very odd, I did what you said (above) and at 30 amps 24 volt DUT they were all exactly 23.9 volts, with no load they were all exactly 25volts. But I tried the old way again just for kicks and I still get the same results as before with the variance starting from 1 increasing linearly to 10.

 

[ronv]

Not sure what is going on. It may be a bit of noise or something. The problem is that the reading is also to high. How are you measuring the current? The readings would put the current at 55 amps for your 47 amp test. You could try a couple of things to narrow it down:
1- Measure the voltage at the + and - terminal of each op amp. Don't move the ground on the probe, only the positive lead. These voltages should be the same and since one of them should be the same wire it should always be the same. Be careful, a slip could be bad.
2- It will be interesting to see how the output of the current measurement op amp compares to the measurement since it is amplified.

 

[jocanon]

I think you are right that the readings are higher than they should be because it is fully loaded (47 amps) when I turn the pot to only just past half way. I think if I turned the pot all the way on it would go up to about 70 amps instead of 60. I say that because I tested it out when I only had 2 FETs hooked up and with the pot turned all the way up it pulled a load of about 14 amps according to my watt meter, which should be fairly accurate, so divide that by 2 and get 7 amps per FET, multiply by 10 and get 70 amps. I don't know if this makes a difference, but the DUT is actually a little more than 24 volts, it's about 25.

 

[jocanon]

I did what you said. The voltage was so small I had to turn the volt meter to the 2000m setting. I got readings all over the board here. Of course these are very small voltages so I imagine they could be very sensitive. Some of the pos and neg terminals of the op-amps read about 14, others were 4 and 5, while a few others were actually negative 4 or 5??? I double and triple checked that I didn't have the probes flip flopped, I didn't the readings were actually inverse on a few, what does this mean? It seems to be working fine, I mean I am pulling a steady constant current. Do these strange reading really matter if it is working over all? Or does it mean that if I let it run for an extended period things will blow up eventually?

Edit:
I wonder if it has anything to do with my placement of the logic PSU. I have it soldered to the far end on the ground and the 12 volt wire. The end where it is soldered is where I was getting the higher readings on the first test. Also, I checked the voltage from ground to the wire that I soldered all the positive op-amps and the pot to and on one end of the ground wire I had 507 milli volts while on the other end i got 510 milli volts. I had the pot set to 40 amps on the DUT.

 

[ronv]

I think it is working ok, just not sure about how accurate the reading is going to be. Lets do one more and hook the meter ground to the center of the big ground then measure each + & - on the op amps.

 

[jocanon]

I think it is working ok, just not sure about how accurate the reading is going to be.

If you mean the reading on the temp and current volt meters, could we fine tune that by adjusting the values of R14,15,5,10,or 11 (just not sure which ones). I wonder if I could put a little trim pot in there to calibrate them exactly.

 

[jocanon]

Lets do one more and hook the meter ground to the center of the big ground then measure each + & - on the op amps.

OK, here's the results: I did the test at 40 amps and 24 volts DUT all were .51 volts except both (-)s on the last op-amp set (since they are in pairs) were .57. But even on this one the (+)s were still .51 volts. If you go back and look at my picture where I showed the under side, the one op-amp pair that is getting the off reading is the only one that is connected to ground on the one side of the DUT ground closest to where the 12v logic PSU is connected. All the other op-amps connect on the other side of this...don't know if that is of any significance.

Edit:
I just thought of one other thing that is different about that one set of op-amps, that is the one that is connected to the reading circuitry, i.e. the volt meters to read amps and temperature (I still haven't hooked up the actual volt meters yet, but I have all the resistors and everything wired up).

 

[ronv]

Sounds like it is just a little ground shift. We may need to calibrate the current meter when we are done - or maybe not. It's probably time to hook up the meters and see how close it is to your meter.