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

Here is one set up for about 6.5 to 22 volts for temperatures of 25 to 60C.

Since the stall voltage is so low lets just let it stall - pretty low current.

I'm suprised it runs down to such a low voltage. Most don't. Must be good fans.

I measured the temperature of the current sense resistors after running the PSU on full, 1128 watts, for about 10 minutes, the tempurature seemed to settle in at about 178C. I think the circuit above will still work even though it maxes out at 60C because I can work with where I position the thermistor (i.e. I don't need to put it right next to the current sense resistors but can put it a little ways off from them).

I have a 2 new issue:

1. I put the two TVSes on, one going one direction and the other going the other direction between +24v and ground on the dummy load side, but now when I put the PSU on the dummy load the PSU sparks and turns off...the fans are still turning on the PSU (this is what they do when they are plugged in but think they are not connected to a computer so they do not turn power on to the power rails), so it still has power, but it does not put out any voltage out of the leads. If I plug it in to a regular wall socket the PSU still powers back up and works fine, I think it thinks there was a surge and a saftey feature cuts the power or something. But I never had this problem until I put the TVS on and now I have this problem everytime. So my solution, unless you guys object, is to simply take the TVS out and take my chances. The only real risk would be if I were to unplug the DUT PSU with current turned on because the current in the lines would not have anywhere to bleed off to, right? I mean I should be fine as long as I always turn the pot down or power the whole thing off before I unplug the DUT, is that correct? The TVS, I think, is more of a safety feature to protect it from me making a dumb mistake of unplugging the DUT while power on the pot is turned up.

2. Secondly, this might not even be an issue, but the blue wires (if you go back and look at my pictures at post 387, the blue wires are the ones going from source on the FETs to the side of the current sense resistors opposite ground, they are 18 AWG) get very hot when I am pumping 1128 watts through the load bank. I took a measurement and I think they are about 200C (I say I think because I used a tempurature gun and since the wires are so skinny, I don't know if I was only getting the reading from the wires). I can barely touch them for a second, if I held onto them for very long, I would definately get burned. It's not so hot that the silicone insulation is melting off of them, but I do get a little bit of a smell of burning/or maybe just heating up plasitc type smell when on 1128 watts, it's not a very strong smell but definately there. The longest I have run it is probably about 10-15 minutes at this current level. Do you think I should have used a larger wire (a little late now...guess I could go back and re-do them all, but don't really want to)? Do you think I need to be concerned about the level of heat on these, based on my description? If the wires got so hot that they melted through the insulation, it might short the gates to the source because those wires are laying on top of each other...I tried to phisically separate them, but it might still be a problem if that happened.

My bad on the TVS diodes. Use one with the band going to +24 and the other side to ground. (I think)
I think the wires are big enough. It's probably heat from the resistors and FETs conducting up the wire. The fans will help a lot. If they were 200C you would have been burned but I'm not sure what is that hot - maybe the resistors.
Yep, I wouldn't put the thermistor in contact with the resistors they are to hot for it.

I noticed on the TVS there is a grey band on one side, is this what you are saying should go closest to +24 volts and the other side to ground? And then just take the second one off completely so I only have one TVS diode now?

Yes, I think so.

About issue #1
Some TVS's are unidirectional. That is, they work like diodes and conduct in one direction, block in the other. And some others are bi directional, so they will block in both directions. If yours have a band, they are likely the unidirectional type. And you should tie the stripe/banded side to the positive voltage. Or if you want, you can take two and tie them together with the bands pointing each other, like this...

====██▐===▌██=====

This will make a bidirectional unit. Then you can install this any way you want with no problems.

Your symptoms also sound like the TVS could be tripping prematurely. If this is the case, you can tie two together with the bands all pointing the same direction, like this...

====██▐===██▐=====

Then tie the stripped/banded side to positive. The parts will offer less protection, but also be less of a nuisance to you.

About issue #2
With 10 FET's and 24Volts at 1128 watts, we are only putting 4.7 amps through those wires. That shouldn't be anywhere near the current required to make them get very hot. As ronv has already said, those wires are more likely getting hot because the heat from the resistors is carrying over to them. Rather than them heating up because of power being wasted in them. I recommend that you cool the resistors with your fans as best as you can, that should help with the heat up I'm pretty sure.

In hindsight, it may have been better to not have burned up so much power in the sense resistors.

I took one TVS off and left only the one with the band on the +24v side and everything is working as normal now.

Regarding issue #2, now that I think about it, what you guys are saying makes a lot if sense, I am sure it is the heat from the sense resistors running up the wires. Hopefully they are still ok and not too hot, the fans will definitely help.

Is it OK to leave just the one TVS with the band on the 24v side, does that off less protection than setting it up with two like in your first example? It's a real pain soldering onto that huge copper pipe heat sink, so I hope you say it's fine with just the one .

Is it OK to leave just the one TVS with the band on the 24v side, does that off less protection than setting it up with two like in your first example? It's a real pain soldering onto that huge copper pipe heat sink, so I hope you say it's fine with just the one .

Click to expand...

It's 100% fine to do it how you have it now, so long as everything turns on and works still.

Actually, your way is safer in a way. It's a little difficult to explain, but a TVS diode breaks down (becomes a short circuit) when the reverse voltage across it raises beyond a certain set point. But it also already conducts (is a short) the other way, so it automatically protect against over voltages in the forward direction by natural design. Adding two together (as shown in my first picture) would actually marginally lower your protection as it would somewhat raise the forward directions trip voltage. You would only do as I have above if your supply were somehow putting out current in both directions, such as with an AC electrical cord. Your supply is DC, so you shouldn't need to do this at all.

Note
This way would be "forward" to most every electrical engineer...
(Ground)===▌██===(Positive)

And this would be backwards...
(Ground)===██▐===(Positive)

You hook them up "backwards" to protect normally. Forward would look like just a piece of wire to your power supply and cause you problems, which appears to have been what was happening to you.

Thanks for the explanation. I am somewhat surprised I didn't break anything permenantly when I had the diode hooked up in the forward direction shorting +24v to ground. I gues the PSU's built in protections saved me, but still, I am surprised it didn't fry anything in the dummy load itself.

ronv said:

Here is one set up for about 6.5 to 22 volts for temperatures of 25 to 60C.

Since the stall voltage is so low lets just let it stall - pretty low current.

I'm suprised it runs down to such a low voltage. Most don't. Must be good fans.

Click to expand...

Thanks ronv...just 2 questions:
1. Can I use this MOSFET, IRFZ24N, instead of D44H11 (since I already have a IRFZ24N):
**broken link removed**

2. I'm a little confused where the thermistor goes, is R3 the thermistor so it would have to be a 4k thermistor instead of the 10k that I already have, or is R3 just a normal fixed resistor and the thermistor is the bold black line between the fan and the bottom of R4 and U2?

You can use the MOSFET but you will loose a couple of volts on the high voltage. The transistor only looses a few tenths at the high side. I thought since we were going first class might as well go all the way.

The 4K is the termistor. I left it at 4K cause that is it's resistance at 60C - 10K at 25C.

Ah, I see, so the 4k thermistor IS my 10k thermistor. So all I need to buy is the MOSFET (I think I have all the other resistor values and I still have the op-amp left over).

I hate to see you have to pay shipping for just 1 transistor or do you need to get the heatsinks as well. If not you might be able to find something at radio shack.

I think Radio Shack might be my best bet. What are key features I need to look for?

If you can find an NPN transistor in the TO-220 package it will probably work. Voltage over 30 volts and current as high as you can find.

I couldn't find any transistors in the TO-220 package at the local radio shack. I need to order some parts off of mouser anyway so I can combine to save on shipping, I was just hoping to finish tonight, but it will have to wait.

If I had *ANY* money for shipping I would happily send you as many TO-220 NPN transistors as you want. But I'm flat broke ATM (More like constantly broke), sorry.

Many many junk electronics out there have big power NPN's. You may try your hand at salvage if you can find something that's worth tearing apart and feel like maybe doing the leg work. Would save you money and time, *IF* you find something right away. Any CRT screen is almost certain to have one near the power supply section. As will lower end or older computer power supply's usually.

It's just a matter of reading the numbers on the part into Google followed by the words "datasheet" to figure out what you have.

Buying them is far easier of course.

Before I got into all this electronic stuff, I threw away my old computer that didn't work anymore. Hard to think about now how many parts I could have salvage our if there!

OK, so I was testing the teperature controlled fan circuit on the PSU (not the the one on the dummy load). Everything worked great, almost. I glued the thermistor to the heat sink with thermal epoxy. I had to glue the MOSFET to the case with the same thermal epoxy or I get run away self heating issues when the fan is slowing down, with it glued to the case I don't have this problem. The only problem I am having right now is if I run the PSU on 47 amps, its max, for about 4-5 min or so, then when I turn the current down the fan behaves very strangly...as I go from 47 down to about 20 amps, the fan slows down as it should, but then from about 20 amps down to 0 amps the fan speeds back up again, going full on at 0 amps. If I turn it back up to 20 amps the fan slows back down to medium speed but then still speeds up again as I turn it back down to 0, it continues to do this until things cool off inside the PSU. I don't know what's going on, I wonder if the heat sink that the thermistor is glued too is getting too hot and causing the thermistor to behave strangly. Any ideas? Do I need to get a different thermistor, or is is something else entirely, the MOSFET??? Thanks!