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

[jocanon]

This is a good project. Some different "stuff". It will be interesting to see just how well the water cooling works.

I know, I hooked up the water just to be sure there were no leaks. The pipe felt pretty cold, let's see how it feels with 1200 watts though. One thing that has me a little bit worried is I just noticed when I was hooking up the vinyl tubing that it is only rated for up to 21c. Our design will shut off when the pipe gets up to a lot hotter than that, but I am not sure how the heat of the pipe will translate to the heat of the water flowing through the vinyl. I may have to upgrade at least the drain side of the vinyl if it doesn't hold up to the heat.

Edit: the max operating temp of the vinyl tube is 79.4C not 21. Maybe it will be OK.

 

[()blivion]

OK, I think it should be pointed out that the name plate amperage on most computer powersupply's is the max instantaneous current, not continuous. It's the current that will shut down the PSU generally. Though this is usually marketing propaganda and only found on consumer grade PSU's. It's more likely that your server grade units are more honest, though 55 Amps is quite a lot. I though that via other data we knew these things can source a certain level of constant current safely? What was that source and amount again, and does it match the name plate current?

This is need to know info most likely. This is because the current the power supply is made to survive, and the current it is made to shutdown at may have large negative gap between them. So even a properly built system is capable of operating in a way that could destroy the PSU's.

 

[jocanon]

There was a youtube video which I cannot seem to find now where a guy had it hooked up to a load bank and when he turned the current up to 55 amps the over current protection would kick in at that point. I know that does not address the constant current point you made. I have read in various posts on RCGroups of people that have used them at 50+ amps continuously without problems, here is one link (look toward the end of the first post):

https://www.rcgroups.com/forums/showthread.php?t=1581061

Short of that, there is not anything to guarantee that it won't ruin the PSU if it go up to 50+amps, however, that is one of the reasons I want to make the load bank is to test this out. I can sacrifice one or two PSUs in the name of experimentation and learning. I want to be able to push it to its limit and beyond to see just what it can really handle so that I am knowledgeable about it and can provide better customer support when I start selling these things. The question I have is, is there any risk in ruining the load bank if I push the PSU too high, that would be bad. I would hate to have to re-make the load bank, though it would not be the end of the world, just a royal pain in the you know what.

 

[()blivion]

The question I have is, is there any risk in ruining the load bank if I push the PSU too high, that would be bad. I would hate to have to re-make the load bank, though it would not be the end of the world, just a royal pain in the you know what.

Generally no, there is not much of a risk. It's physically possible for anything to happen when a supply goes south though. For example. The main transformer wiring could short circuit to the low side, dumping high voltage mains directly into your dummy load. If this happened, It would destroy the FET's instantly, And most likely anything else hooked up to them. This is unlikely though as they usually have strict safety standards and the SMPS has been well understood for ages.

What will more than likely happen is you will smoke a few supply's that are not quite capable of handling all the juice. Operate them near an open window with a fan in it so the smell doesn't fill your work area. And make sure to have a smoke detector or two in the room. You really should not leave them unobserved for any given amount of time, but 6 plus hours times your number of power supply's is a long time to wait and watch for the magic smoke to show it's self.

None of the fuses blew in the supply's I destroyed with this circuit BTW.

 

[jocanon]

Yes, those are good points. Also, speaking of fuses, it seems like a while back someone mentioned something about fuses, but then that conversation kind of died out...would you recommend putting fuses in the dummy load circuit at strategic points anywhere?

 

[()blivion]

A more than 5 Amp but less than 10 Amp slow blow fuse on every single MOSFET's ground would be ideal. It would be cheep to do too, use slot style automotive fuses. (The colored plastic ones.) They are great for this kind of stuff. You can even solder to them. You want Red colored ones, as that's the code for 10 Amps. Red will be less safe than orange but less likely to blow prematurely on you. 10 Amps is still within the safe zone for our FET's.

If you try and gang everything together, to make a 55~75 Amp single point fuse, and you're running a low load, then just one single FET goes short, the total current will have to exceed 55~75 Amps before it blows that one main fuse. Not hard to imagine what is going to happen if you are running a low load and a FET shorts and dissipates 30~45 Amps through it for more than a few milliseconds.

Edit: When they are fully on, FET's can dump this many Amps no problem, but that's because they are not dissipating Watts when fully on. We can't expect a circuit failure to play nice though.

If you wanted to get a finer grain amp rating for your fusing or a bit more headroom, you can get 15 Amp fuses instead and gang up one for every two FET's. This gives you ~7.5 Amps effective protection for each FET when running a modest current. So if one goes short,his buddy is already dumping some amps to the fuse, then the head room is smaller for the short one. With this way you will only have to buy half as many fuses also. The only bad thing is that if your running low load and one decides to go more short circuit, that will be ~15 Amps X 24 Volts or 360 Watts. The transistors will only survive this for a very short time. Our cooling can only help so much before the thermal impedance of the package reaches it's limit. Though I suspect even this will be safe enough since we are not necessarily dumping all the watts in our transistor.

No matter what, 10 Amps or less per FET at 24 volts supply will not damage them as that's within there safe operating area.

 

[jocanon]

Sounds good, I'll probably go with 10 amp fuse per FET.

 

[jocanon]

A more than 5 Amp but less than 10 Amp slow blow fuse on every single MOSFET's ground would be ideal.

So, that would be between the source (3rd pin) on the FET and the .13 ohm current sense resistor? Or would it be between the ground and the current sense resistor, or does it matter?

 

[()blivion]

Yeah, since we are protecting our circuits parts, it may be worth while to get 10 of these as well.

https://www.digikey.com/product-detail/en/V47ZA05P/F3048-ND/1009560

These are to protect against over voltage should something go wrong, and it will only cost you $5:00 to do.

 

[ronv]

Rule #47. A transistor protected by a fuse will always short to make sure the fuse blows.

 

[()blivion]

So, that would be between the source (3rd pin) on the FET and the .13 ohm current sense resistor? Or would it be between the ground and the current sense resistor, or does it matter?

Not sure. There are three ways you could do it.

(1) Between the resistor and ground
(2) Between the FET and the resistor, with the current sense line on the FET's side of the fuse.
(3) Between the FET and the resistor, with the current sense line on the resistors side of the fuse.

Config (2) is the most hard to figure out. If a FET shorts, and the current goes through the OP amps input to ground, then the thin current sense wire will ablate just as well as any fuse I would imagine. But with the fuse in between the sense loop and ground you may mess with the effective resistance and possibly compromise precision in setting current sense loop. However, fuses increase in resistance sharply as they heat up from over current, this will put a large voltage on the current sense causing it to try to shut the transistor off. This may or may not have a useful effect for us depending if the FET is still responding to signals.

I would say (1) put them outside the resistors, IE... between the resistors and ground. Though there may be a compelling reason to do (3), as it theoretically protects the OP-Amp from over current too. Though, if you dump anywhere near 10 Amps into an Op-Amp it's done for.... no matter what.

 

[()blivion]

Rule #47. A transistor protected by a fuse will always short to make sure the fuse blows.

I thought it was something like

"A 10 dollar transistor will always blowup to protect the 10 cent fuse"

 

[jocanon]

Yeah, since we are protecting our circuits parts, it may be worth while to get 10 of these as well.

https://www.digikey.com/product-detail/en/V47ZA05P/F3048-ND/1009560

These are to protect against over voltage should something go wrong, and it will only cost you $5:00 to do.

These are in addition to the 10 fuses, the fuses to protect against over current (too many amps) and the above to protect against over voltage...is that right? If so, would these go in the same place as the fuses, before or after the fuses?

 

[jocanon]

btw...I am all about added protection features, especially if it's only an extra $5. It would be much easier to replace a fuse or varistor than the whole thing.

 

[()blivion]

The above to protect against over voltage...is that right?

A+ deduction Tax man. You are 100% correct.

would these go in the same place as the fuses, before or after the fuses?

They go from the positive supply to ground. They look like an infinite resistor and conduct no current, right until the voltage exceeds a preset limit.... then they snap closed and divert all power through them, and NOT through your circuit. I have NO idea how many would be needed to protect against any given scenario. But at 100A each and 10 of them.... That should protect against even the worst of over voltage situations.

Edit: Note that they are good for 100Amps, MOMENTARILY. It's really complicated math (for me) to try and figure out exactly what they will save you from as far as constant over voltages. Much much lower I know that. Their physical size should be an indicator of their power. IE... not much.

 

[jocanon]

Okay, hopefully I can get another A+ with my drawing attached...or at least a B . I wanted to draw out my understanding of the circuit as we have modified it now to make sure we are all on the same page. I put two possible locations for the MOV, I am assuming you mean for it to be the lower option since the first would only require 1 MOV and you said I would need to buy 10. Let me know if I got the fuse and MOV in the right spots (for the fuse I used option 1 from ()blivions 3 options above).

 

[()blivion]

(1) Both the MOV options you have written down equate to the same thing electrically. Supply>>>ground x 10.

(2) The fuse is in the right place, though you need to put one under R1 also (hidden, it's left just a bit more).

(3) You drew all those symbols PERFECTLY, A+.

(4) I TOTALLY forgot we had added over current shutdown to our circuit, which is basically the same thing as a fuse. Even so, fuses will still have safety benefits. Edit: TOTALLY did not pay attention to the schematic, it is NOT over current, it is over temperature.

(5) I made you have 10 because we need 10x the level of protection that one of those can provide, and you get price drops for buying >9. You don't need to have one on each FET though.

On second though, may want to get just one fat MOV, if they make them (cheap) for this level of power. I wanted to try to make it handle a continuous over voltage, and this seemed like the way to do it. But It may not work out like I had thought.

 

[jocanon]

So as far as how I physically attach everything, will I put 10 MOVs right next to each other essentially in the same spot, i.e. running them in parallel from + to ground?

 

[()blivion]

So as far as how I physically attach everything, will I put 10 MOVs right next to each other essentially in the same spot, i.e. running them in parallel from + to ground?

That's what I was going for, yes. They are different than the current protection. Or at least I was thinking they would be. But now I'm not entirely sure. May need to call in an expert for this one. I *THINK* what they *MAY* do is, one will have a slightly lower trigger voltage than the others, it will trip first, slamming closed circuit, voltage will drop so the others won't trip, the current through it will climb through the roof, it will heat up massively and explode, and the others will follow suit until (Edit the protection becomes open and the high voltage is free to destroy the load.

 

[jocanon]

I TOTALLY forgot we had added over current shutdown to our circuit, which is basically the same thing as a fuse. Even so, fuses will still have safety benefits.

Did we add over current protection? I thought we just added protection against over temperature. We added a feature to read the current, but I don't think it shuts down if it goes over a certain limit...I could be wrong. In post 141 ronv talks about the over temperature cutoff.