Help with dynamic test load

[Roff]

How do you switch in the .01 ohms? I hope you crank the pot down to zero before switching.

 

[imix500]

Yes, I turn the pot all the way down. I have a 50A automotive type relay, since the max this will ever see is 24v.

 

[Roff]

What is the range of voltages of the supplies you will be testing?

 

[imix500]

+/- 20v and 48v single ended.

 

[Roff]

+/- 20v and 48v single ended.

How will you test a negative supply?
This line of questioning actually has a purpose, believe it or not.

 

[imix500]

Two options. Either going rail-to-rail or just testing one side at a time, swapping leads. These supplies are completely independent and tied together at the zero point so that should be ok.

 

[Roff]

What voltage were you testing when the power module died at 27A?

 

[imix500]

It's output was 24v

 

[Roff]

OK. I guess I didn't ask the right question previously. When I said range, I meant low to high voltage range.

 

[imix500]

No prob. Low end of the range would be 5v. High end I'm going change my answer and say 50v. It would be nice to test higher voltages for amp psu's, but that doesn't have to be for this unit.

 

[Roff]

The reason for these questions is I was thinking you could put a BF resistor in series with the collector to soak up some of the power. It's difficult to kill a properly-sized resistor. With such a wide voltage range, you would have to have another relay(s) to switch a resistor(s) in and out.

 

[imix500]

Hmm, that's not a bad idea. It would definitely increase the size of the unit to mount some big resistors, but if it helps...

So do you think the module just can't dissipate that much power? Looking at the data sheet it clearly should be derated at operating temp with less than ideal cooling etc, and it looks like I'm near the edge of dc SOA at 24v 27A.

I could go with a few smaller modules I've seen surplus if that would be a better route.

Thanks!

 

[Roff]

Hmm, that's not a bad idea. It would definitely increase the size of the unit to mount some big resistors, but if it helps...

So do you think the module just can't dissipate that much power? Looking at the data sheet it clearly should be derated at operating temp with less than ideal cooling etc, and it looks like I'm near the edge of dc SOA at 24v 27A.

I could go with a few smaller modules I've seen surplus if that would be a better route.

Thanks!

Yeah, and the thermal resistance from junction to case is around 0.1 deg/W. With a perfect heat sink, 24V, and 27A, your junction will be 67 deg above ambient. Do you know what the thermal resistance of your heat sink is?

 

[imix500]

I think I calculated it a few years ago.. I don't remember. I can do it again within a couple days.

 

[power58]

Load

I like to build electronic loads in 10 amp stages. One OP AMP that has its common mode voltage including ground. LM324's work nice. You can build 4
10 amp Cells using Darlinton's or FET's. As explainded before your junction temperature can never be exceeded on your shunt device. It is much easier to cool 4 heatsinks than one large one. I like to keep the sense resistor at 0.1
ohm. This gives a larger feedback signal with out excessive gain that could cause instability. Keep up the good work.

 

[power58]

Load

Should explain better. One Op Amp section per Load Cell. So 1 LM324 would run 4 10 amp Load cells. The inputs of each OP Amp cell are programmed by your load (voltage) Source.

 

[Roff]

I like to build electronic loads in 10 amp stages. One OP AMP that has its common mode voltage including ground. LM324's work nice. You can build 4
10 amp Cells using Darlinton's or FET's. As explainded before your junction temperature can never be exceeded on your shunt device. It is much easier to cool 4 heatsinks than one large one. I like to keep the sense resistor at 0.1
ohm. This gives a larger feedback signal with out excessive gain that could cause instability. Keep up the good work.

That sounds like an excellent idea.

 

[imix500]

The more I think about it the more I like it. Much easier to cascade, and if one section goes I just loose part of the loading. It shouldn't fail in a runaway fashion. Also, 324's are a tenth the cost of the Burr Brown chips.

The smaller two units have the same heatsink. I'm pretty sure one is big enough for the medium size unit, but I'm going to measure it tomorrow night to be sure. The largest one has a 12x12 lytron double pass coldplate, and I haven't been able to max out it's heat exchanger with residential power.

So, I need an opinion on whether to go with fewer higher power mosfet devices, or more conventional TO247 types. I still need to match them right? If so, I'll just go with the far cheaper 200W devices and inexpensive cement emitter resistors. Thanks!

 

[Roff]

The more I think about it the more I like it. Much easier to cascade, and if one section goes I just loose part of the loading. It shouldn't fail in a runaway fashion. Also, 324's are a tenth the cost of the Burr Brown chips.

The smaller two units have the same heatsink. I'm pretty sure one is big enough for the medium size unit, but I'm going to measure it tomorrow night to be sure. The largest one has a 12x12 lytron double pass coldplate, and I haven't been able to max out it's heat exchanger with residential power.

So, I need an opinion on whether to go with fewer higher power mosfet devices, or more conventional TO247 types. I still need to match them right? If so, I'll just go with the far cheaper 200W devices and inexpensive cement emitter resistors. Thanks!

You don't have to match MOSFETs if you only have one per op amp. You do need one sense resistor per stage also. I guess with 0.1Ω @ 10 Amps, you would use 20W resistors. (?)

 

[imix500]

Ah, so the op amps can't share a sense resistor? Ah, because they are driving the mosfets separately.
So with a .1 ohm sense, in order to get the loading of a .01 sense I'll need to adjust the gain of the amp to drive the mosfet harder?
I'll start looking for cheap ~500W devices.

Hey, can I use IGBT's?