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

Yea, there are some pretty high currents and voltages flying around in there and we may not be getting power and ground close to the circuits we are driving. If you want to play you could try one of your .1 ufd caps across the thermistor to see if it makes it better or worse.

Better than that, maybe from the output of the op amp to the - input

Dusey52 said:

The following is speculation based on observation so take it for what its worth.

- The power supplies have built in thermal protection which includes controlling the fan speed to suit the needs of the power supply. This has been observed and if you think about it, the server manufacturer would be crazy not to do it this way. These are field replaceable units, if the thermal control was in the server and it failed, then both (redundant) power supplies would be toast and so would their replacements.

- Pin 4 is there so the host server can call for additional cooling (for the server, not the PS).

- jocanon's fan test peaked at ~ 7v because the PS internal temp wasn't hot enough force it any higher (see below).

- The servers usually live in a comfortable 65-70 degree environment however they can withstand much higher temperatures. We've had computer room cooling failures where I work and the servers start shutting themselves down at about 95 degrees ambient.

If you're going to offer a temp 'assist' circuit then why not pick a temperature where the speedup starts and another temperature where the fan reaches full speed? Load by itself doesn't mean nearly as much as load AND ambient temperature. Just measure the temperature of the air comming out the back of the PS. I've measured several servers in operation and the temp comming out the back is 110-120 degrees and their fans are all running at their slowest speed.

Click to expand...

Thanks for the input Dean. So are you saying that you think the PSU fan would go higher than 7 volts if it needed to? I wonder though how these are set up to operate by the manufacturer, I mean, if you don't short pin 4 to ground the fan runs much faster from the start. Are they designed to be run with the pin 4 shorted to ground? I think that is the question. If they are then maybe you are right, but I was going under the assumption that pin 4 to ground was slowing the fan below what they were designed to be run at under load.

Also, can you explain what you mean in the last paragraph. I am not sure I follow. I thought by putting the thermistor on the heat sink it would essentially take into account ambient temperature as well since the PSU would get hotter if the ambient temperature was hotter.

Edit:
I wanted to add, if you don't short pin 4 to ground the fan still does the same thing as far as speeding up as it heats up, the min speed and max speed are just much higher.

All servers that I'm familiar with will spin the fan at full speed at power-up and then slow down to min speed during the BIOS boot-up process. The only other time I've observed them running at anything other than minimum speed is when there's a cooling issue and the ambient room temp is > than about 85.

The problem I see with using the heat sink (other than its energized) is we don't know the max temp it’s designed to run at other than guessing based on the maximum case temperature of the attached components. We can make some educated guesses about the intake and output air temps based on the environment these things are designed for and the real world experience we and others have had with them. The servers where I work run with the PS output air temp in the ~115-120 degrees 24x7x365 and the fans don’t speed up until it goes much higher (sorry I don’t have a number that). So I would pick a temp you’re comfortable with, say 125 for arguments sake, as a starting point to start spinning up the fan and then a delta, say +15 degrees at which point the fan is running at maximum. You can easily test this with your load & a thermometer. Just make sure the cases are closed so the air is moving as designed. You can artificially raise the ambient air temp with a hair dryer or heat gun. Or you could even use the output air of one PS and feed it into the intake of a second PS.

All servers that I'm familiar with will spin the fan at full speed at power-up and then slow down to min speed during the BIOS boot-up process. The only other time I've observed them running at anything other than minimum speed is when there's a cooling issue and the ambient room temp is > than about 85.

Click to expand...

I can confirm this and extend it to cover most all computers in general as well. This is a motherboard safety feature, I'm not exactly sure what it's for. Presumably to insure the fans are spinning, since we already know they will run with much lower power than what it takes to actually start them. Or possibly a precaution for the event the system was just power cycled or hard reset directly after having heavy load on it. But as far as I know this is a motherboard controlled thing only, and does not apply to the raw PSU.

In any case, it's worth investigating the circuit to see if there is real thermal fan control in these PSU's and not just secondary effects of heat messing with regulation or something similar. Useful information... Doing it by artificially increasing the PSU temp as indicated above may work for this as secondary effects should only change the speed so much. Consumer PSU's usually don't have thermally controlled fans, but they certainly do exist. And you can count on server PSU's having all the bells and whistles. Honestly neither fan system would surprise me.

I'm going to leave the over voltage crowbar on the dummy load for a later time. (hint: lazy)

So Dean, just to make sure we are on the same page, I want to list out the points I gather you are making to see if I understand correctly:

1. Are you are saying you think it would be better to put the thermistor at the back of the PSU to measure the temperature of the air blowing out the back because this would capture the ambient temperature and the temperature of the attached components better than attaching the thermistor to the heat sink? And that the heat sink is a bad idea because it's energized (I still don't see the problem with that though as long as the thermistor is fully insulated).

2.Secondly, I should make the temperature controlled circuit not raise the speed of the fan until the air is above say 125 degrees Fahrenheit and then make the speed of the fan increase linearly as the air gets hotter until it reaches max speed at 140 degrees.

3.Lastly, I should test to see if the fan speeds up to max on its own by artificially raising the ambient temperature with a heat gun, blow dryer, or another PSU, and if it does then a temperature controlled circuit might be a mote point.

Am I correct in all the above statements? If not, please correct me where I am not getting it. Thanks!

Edit:

All the above being said, I still wonder if it would be better to just stick with the temperature controlled circuit the way I have it right now, that is speeding up the fan linearly until it reaches max speed under full load at room temperature for the following reasons:

1. I don't see any draw backs to this except the minor issue that the fan will be a little bit louder when under heavy loads. I only see a potential upside, that being that it will keep things running cooler.

2. Leaving the fan mod with just shorting pin 4 to ground the case got up to 120 degrees under full load, which is hot to the touch, with ambient temperature at only about 75 degrees. People could almost burn themselves just by touching the case, whereas with the temperature controlled circuit the way I have it configured now, you can run the PSU at full load all day long and the case will only get up to about 80 degrees Fahrenheit so you can handle it without issues.

3. I am not sure comparing the usage of the PSU in a server is exactly the same as the environment we will be using them in. I could be wrong, but I don't think a server will max out the PSU continuously for any extended period of time, it probably has peaks and valleys of current draw, whereas for our purposes, we will be constantly drawing near full capacity of the PSU while charging lipos then it will taper down as the batteries reach their full charge. That being said, it might be safer to just let the fan run at full once the internal temperature of the PSU reaches an equalibrium point with the fan on full blast and full current draw. I think taking that approach would err on the side of caution as we don't know the safe operating temperture of the PSU. (I wish I could get a datasheet on one of these).

Any thoughts from anyone?

Here is my 2 cents worth, which is probably worth 1/2 of that.
From a reliability point of view cooler is better especially when running close to to the maximum spec like you are.
If you put the themistor in the air flow you can still make it turn on fully but there will be some time lag where the actual FET will be hotter than it would be with it attached to the heat sink. You could make up for this by having it go to full speed at a little lower temperature. The fact that the heat sink gets to 70 to 80 C says the circuit is pushing the limit. My bet is that there in not much margin at maximum load or they would have made it a 1500 watt supply.
Having said all that I like the idea of getting the thermistor off of the heat-sink where there is high voltage.
I don't know if the fans keep running after an over current event, but if they don't there is probably not much to be gained by running them at power on since the themistor on the heat sink reflects case temperature of the FET which is the part you are trying to cool not a CPU or something.

Well, I think it is worth a lot more than 2 cents ...the way I figure, what is there to loose by running the fan faster than needed? Nothing really, but if we don't run it fast enough we risk the life expectancy of the PSU since we are going to be pushing it to the max.

As far as putting the thermistor on the heat sink, is the only concern that the leads may short to the high voltage on the heat sink, or is there more to it than that? What if I used one like this (I already ordered it from digikey, I am looking at it right now and the insulation around it is very thick, as thick as 22AWG wire insulation so I don't think there is a risk of it shorting to the heatsink):

https://www.digikey.com/scripts/dksearch/dksus.dll?vendor=0&keywords=bc2648-nd

No, it's just having anything in proximity with the 170 volts that could possibly get into the low voltage stuff.
I'm just paranoid.

I agree with 'the cooler the better' from a technical point of view and anyone wanting maximum cooling (or close to it) doesn't need to do anything. The fans run at near max speed on an un-modified PS. The balancing act here is managing both heat and noise. The fans at full speed put out a fair amount of noise and there's a desire by many to keep the noise down as much as possible while still providing adequate cooling.

I also think gluing the thermistor to the heat sink is probably not the best place for it. First, it's energized, possibly with line voltage and second, there are other heat producers in there. Gluing the sensor to one of the heat sinks will have the effect of weighting the heat contribution of that heat sink well above the other contributors (like that transformer jocanon mentioned). If I remember correctly there's aleast one other heat sink in there as well.

Well, I have to admit, I don't understand because if it's fully insulated I don't see how it would get in, but I think it is two against one here and I will probably loose the debate Anyway, you and Dean have more experience than I, so I will trust you, at any rate, even if it wouldn't be a problem, I am sure 9 out of 10 people will have the same concern and not want to buy the circuit for that reason.

So, ronv, above where you said to try the cap from the output of the op-amp to the neg input, was that to hopefully make it so I could put the thermistor on the transformer without getting noise? If so, I can try that.

Dusey52 said:

I agree with 'the cooler the better' from a technical point of view and anyone wanting maximum cooling (or close to it) doesn't need to do anything. The fans run at near max speed on an un-modified PS. The balancing act here is managing both heat and noise. The fans at full speed put out a fair amount of noise and there's a desire by many to keep the noise down as much as possible while still providing adequate cooling.

I also think gluing the thermistor to the heat sink is probably not the best place for it. First, it's energized, possibly with line voltage and second, there are other heat producers in there. Gluing the sensor to one of the heat sinks will have the effect of weighting the heat contribution of that heat sink well above the other contributors (like that transformer jocanon mentioned). If I remember correctly there's aleast one other heat sink in there as well.

Click to expand...

Dusey, yes, there are two heat sinks, but the one never even heats up and it's not energized. So if I could fix the noise problem with the transformer, do you think that would be a viable option?

As far as noise, I just tested out the circuit the way I have it and in a real world setting it worked great. Here's the thing, the fans on the PSU sped up, but never even got louder than the fans on my PL8, since we can't slow down the fans on the PL8, as long as we are quieter than they are then I don't think there is any real world benefit to quieting them much more, but they were still much quieter than the PSU fan at max. I charged 6 4 cell lipos. It took about 11.5 minutes at 24v and 30amps.

Well, I tried Dean's suggestion, I put a heat gun into the fans intake with pin 4 shorted to ground without any temperature controlled circuit and he was right, it sped up to full speed. The minute I removed the heat gun it slowed back down. So does this mean that it already has thermal control and there is no need for the circuit we made? I think it does. Or doesn anyone still think it would be beneficial?

Edit:
I kind of hate to admit it after all the work we have done, I was really hoping to add some value to it with the temp controlled circuit, but I have to be honest, facts are facts, I don't think it needs it. At least I haven't spent any money yet on circuit boards to put it into production. Maybe the real value that has been added by all this is simply to prove the fact that it is safe to mod pin 4 to ground. I will post a YouTube video demonstrating the built in thermal protection as soon as my iPhone finishes preparing the video I just made.

OK, here is the video. ()blivion, ronv, and Dean. When you all have the chance, can you weigh in here whether you agree that there is no real value with adding a temperature controlled circuit based on the new info in this video. I want to post this up in RC Groups, but only after I see if there is a consensus on my conclusion that the circuit is not needed. Thanks for all your help! Now back to the presidential debate!

https://www.youtube.com/watch?v=vP3qDlh3p-U

That’s a pretty convincing video! The response to the application/removal of heat from the heat gun was quicker than I thought it would be. (Something that's hard to do if the sensor is tied to a large mass of metal.)

I guess you’ll need to decide if it’s worth pursuing further. As ronv said, ‘cooler is better’. I think there’s value there for those folks that live in hot climates or just want to stay ahead of the curve and spin the fans up more aggressively than the factory setup does.

I think you can trust the guys at HP. If they went to the trouble to do it I would think they probably did it right.

Most PSU's almost always have two heat sinks. One for the power switches, and one for the Schottky output rectifiers. They more or less do produce heat independently from each other at different load conditions. So it is wise to average the temps by reading the air temp near the exhaust. The heatsinked parts are not the only things that make heat either.

Edit: Oops... a whole other page of posts are here I guess :/

ronv said:

I think you can trust the guys at HP. If they went to the trouble to do it I would think they probably did it right.

Click to expand...

I agree, with the caveat that we're not exactly using these things as the HP engineers intended.

OK, here is the video. ()blivion, ronv, and Dean. When you all have the chance, can you weigh in here whether you agree that there is no real value with adding a temperature controlled circuit based on the new info in this video. I want to post this up in RC Groups, but only after I see if there is a consensus on my conclusion that the circuit is not needed. Thanks for all your help! Now back to the presidential debate!

https://www.youtube.com/watch?v=vP3qDlh3p-U

Click to expand...

Very conclusive video. Fan is most certainly thermally controlled. And yes, it makes our thermal control circuit redundant and unnecessary.

I think you can trust the guys at HP. If they went to the trouble to do it I would think they probably did it right.

Click to expand...

Also agree. Worst case scenario we could still identify the thermal circuit and mod it rather than make a whole different circuit addition.

You know what they say. If you can't fix it feature it. What you might try is a resistor from the fan control to +12. This would make it run at full speed sooner. Pick the size to max it out at say 35 to 40 amps at room temperature.

Whoops,, I guess it is a resistor to ground. I think you already ran a bunch of values.