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

[()blivion]

I can get power right at the posts inside the PSU, there are mounting screws I can unscrew and put connectors on, but I took the voltage readings right at that point where I had the bullet connectors connected to the end of the wire and it was consistent, results posted above, if your theory is right, shouldn't the voltage have varied here?...anyway, I will try connecting to the posts tonight and see if it makes a difference.

Odd... Then it would seem that changing the point you tap your power makes little sense if the voltage didn't change.

Still... there is almost no mistaking it, the power supply output going up and down is whats causing the fan speed to change. Your video makes it pretty obvious. We just need to know why this is happening. And based on the fact that your tapping into the big main lead suggests that it's certainly NOT being current limited to the point that the fan slows down... considering that lead is already carrying ~50 Amps. With out voltage changes, and current limits... I really can't see the power supply effecting the circuit. That *IS* really the only two things that could change...

Edit: Current limiting would change the voltage of course, which we are not seeing here.

:thinking cap on:

Hummm.... here's a thought. It's a long shot but "electromagnetism". With ~50 Amps running through the main wire, and a really sensitive circuit, you may be getting some induced charge that is high enough to cause the circuit to freak out. I also noticed that a lot of your wires are really long and running all over the place. I think you should maybe try to rebuild the fan circuit from scratch, but this time make all the connections as short as possible. This will also weed out other small common mistakes like shorts and such. I often just rebuild from the beginning rather than trouble shoot a suborn circuit, it's some what contradictory but it's often easier and faster.

Note:
Don't forget to heatsink the sensor to something large and room temp like I told you. It is really important that you do this, self heating *IS* a problem with these circuits sometimes.

 

[jocanon]

OK, will do tonight when I get home...I did have the thermistor heatsinked to the PSU's heatsink but I took it off to troubleshoot.

 

[ronv]

It might just be noise. FETs have large voltage gains (some more than others) so as little as 200mv can be the difference between off and full on. See below. That is why there is not much range - a few degrees between off and on.

 

[jocanon]

Some interesting observations, first, it help but did not cure the problem by moving the connections for the fan down to the power posts inside the PSU. It still speeds up indepenent of temperature when I increase the load, but only by about half as much as it did before. Second, this is very interesting, I tried just for kicks removing the capacitor I have in that makes the fan come on briefly when first staring the PSU and it dramatically improved the problem, again, the problem was still there, but this time only about 1/4 as bad. Problem is I want the capacitor in there because I decided I did not want to let the fans in the PSU stall because they make weird noises when they are slowly starting up and the two fans don't start at the same time so I am thinking that people might think there is something wrong. I think it will just be cleaner overall to tune the circuit so that the fans stay on all the time. So, the reason I need the capacitor is that the low speed I want the fans set to when cold is too slow for them to start up on their own, they need that initial boost from the capacitor.

Edit:
Maybe I can just try a smaller capacitor to minimize the issue. The capacitor goes from +12v to the gate on the FET, bypassing the thermistor for a moment at first start. ronv, is that a new circuit or is it the same as ()blivion's?

 

[ronv]

Ahh, a cap from the supply to the gate.

I think what is happening is there is some ripple on the power supply (noise) at the switching frequency. The cap couples the noise directly to the gate trying to make it turn on.
The schematic is just to show how sensitive the circuit can be. 100 mv of noise would turn it on. You can move the fan to the source side of the fet but then you have the range problem.

 

[jocanon]

Well, maybe I should use the amplified thermistor circuit you designed for the dummy load for the PSU as well? I imagine there is a suitable op-amp that is a single op-amp that would work instead of the dual LM-358?

 

[ronv]

You can use the same circuit. I have to go out for a couple of hours. Let me think about it. You might try a different fet if you have one. It might be less sensitive.

 

[jocanon]

I have two different types of FETs right and am using the one that is less sensitive already, tomorrow I can go to radio shack and get some new parts to experiment around with, especially if you can come up with some more ideas .

 

[ronv]

I need to search out a model to simulate with to get it just right, but you might try lifting the source from ground and adding a 27 ohm resistor (not critical) between the source and ground. Then remove your startup cap from the gate and just put it across the thermistor. What pot are you using and what size cap??

 

[jocanon]

I was experimenting with different sized caps, 100uf and 220uf 35v, depending on where I have the pot set too, seem to work the best. I was using a 10k pot.

 

[ronv]

Try # 529 with a 24 ohm resistor (If you have one) in the source.

 

[jocanon]

Post #529?

 

[ronv]

This one here.

https://www.electro-tech-online.com...fan-load-bank-pwm-circuit.128753/#post1085405

I think this might work,

 

[jocanon]

...Then remove your startup cap from the gate and just put it across the thermistor.

But that's where it already is right now because the thermistor is connected to the gate.

 

[ronv]

The center of the pot should be connected to the gate. One side of the pot to the thermistor.?

 

[jocanon]

I see, I have things hooked up differently, I will try to fix...gotta go now but will be back at it later today. Thanks for your help.

Edit:
Never mind, just bought myself a little more time.

 

[jocanon]

I have the center of the pot connected to ground one side to the gate and thermistor, will try your way...

 

[ronv]

Not to much difference but some. Better with the center to the gate.

 

[jocanon]

So I have one side of pot on thermistor, center on gate, should any part of the part still connect to ground as it did before?

 

[ronv]

The last terminal on the pot goes to ground.



Therm Gate Gnd
o o o

If it goes backwards reverse the outside ones.

PSS. It won't be as sensitive now so you may not see much change by touching it.