Finishing the active load

[Circlotron]

Back in 2004 I stated making an active load for testing power supplies. Never did get it finished. Dragged it out tonight for a look-see. The mosfets are ST Y32NB50. 500V 34A 0.11 ohm TO-247 max. They were a bit of a dog for SMPS PFC usage so I ended up getting a bunch of them free. Each source lead has a pair of paralleled 4R7 2W resistors. 15K gate resistors to make it very docile. I'll run a 1K resistor from each of the sources and join them together and use that common point to measure the current and send it back to an opamp. Each heatsink should be good for 300W continuous, lots more for short bursts. Each mosfet has a rating of 450W @ 25 deg C, derate 3.6 W per deg C.   

[ZeTeX]

Where did you got those heatsinks?

[uncle_bob]

Where did you got those heatsinks?

Hi

They look a lot like some I have from Heatsink USA.

http://www.heatsinkusa.com

There are an almost infinite number of web based outfits that now will sell you short chunks of pretty good extrusion at reasonable prices. For one off type projects they beat the old "buy 50 feet minimum and machine it" approach.

Bob

[ZeTeX]

Where did you got those heatsinks?

Hi

They look a lot like some I have from Heatsink USA.

http://www.heatsinkusa.com

There are an almost infinite number of web based outfits that now will sell you short chunks of pretty good extrusion at reasonable prices. For one off type projects they beat the old "buy 50 feet minimum and machine it" approach.

Bob

42$ shipping for 5$ heatsink.. come on

[uncle_bob]

Where did you got those heatsinks?

Hi

They look a lot like some I have from Heatsink USA.

http://www.heatsinkusa.com

There are an almost infinite number of web based outfits that now will sell you short chunks of pretty good extrusion at reasonable prices. For one off type projects they beat the old "buy 50 feet minimum and machine it" approach.

Bob

42$ shipping for 5$ heatsink.. come on

Hi

$6 shipping for $5 heatsink. There are lots of similar places all over the world. Pick one that is close to you.

Bob

[Circlotron]

Your heatsinks are made for natural convection, they would do much better in forced convection/fan cooling with 2-3 times this fin density.

You don't really benefit using such a quantity of paralelled mosfets, your bottleneck is your heatsink. Quick calculation : With guessed Rth(heatsink) = 0.5°C/W and Rth(mosfet, junction to case) = 0.5°C/W, the average Rth(junction to ambient) is 0.5 + 0.5/32 = 0.515°C/W. Using twice less mosfet, Rth(junction to ambient) = 0.5 + 0.5/16 = 0.531°C/W.

I built the entire thing out of unwanted samples and production leftovers at work. Yeah, the mosfet count is a bit over the top, but nothing exceeds like excess.   
Mosfets are great as a current sink because once you are above a certain current the characteristic curve is almost horizontal so changes in drain voltage mean almost no change in drain current. Of course, having so many mosfets means that it is just that much harder to get up on the horizontal part of the curve...

[uncle_bob]

Your heatsinks are made for natural convection, they would do much better in forced convection/fan cooling with 2-3 times this fin density.

You don't really benefit using such a quantity of paralelled mosfets, your bottleneck is your heatsink. Quick calculation : With guessed Rth(heatsink) = 0.5°C/W and Rth(mosfet, junction to case) = 0.5°C/W, the average Rth(junction to ambient) is 0.5 + 0.5/32 = 0.515°C/W. Using twice less mosfet, Rth(junction to ambient) = 0.5 + 0.5/16 = 0.531°C/W.

I built the entire thing out of unwanted samples and production leftovers at work. Yeah, the mosfet count is a bit over the top, but nothing exceeds like excess.   
Mosfets are great as a current sink because once you are above a certain current the characteristic curve is almost horizontal so changes in drain voltage mean almost no change in drain current. Of course, having so many mosfets means that it is just that much harder to get up on the horizontal part of the curve...

Hi

For most supply testing, having a load that behaves like a resistor is a handy thing. It's also a nice thing to be able to test with a current sink as a load. The each may / will do different things to supply stability.

Bob

[diyaudio]

Back in 2004 I stated making an active load for testing power supplies. Never did get it finished. Dragged it out tonight for a look-see. The mosfets are ST Y32NB50. 500V 34A 0.11 ohm TO-247 max. They were a bit of a dog for SMPS PFC usage so I ended up getting a bunch of them free. Each source lead has a pair of paralleled 4R7 2W resistors. 15K gate resistors to make it very docile. I'll run a 1K resistor from each of the sources and join them together and use that common point to measure the current and send it back to an opamp. Each heatsink should be good for 300W continuous, lots more for short bursts. Each mosfet has a rating of 450W @ 25 deg C, derate 3.6 W per deg C.

Nice work on the mechanical, hate to be the one to repair it, could have used a PCB with those current bus bars.

450W @ 25 deg C, derate 3.6 W per deg C.  However it says nothing when using the Mosfet in its linear region.. whats your SOA calculations per FET for a total of 300Watts  ?

[TiN]

I have some doubts regarding those heatsinks and fans depicted to handle 300W continuous, based on previous experience with similar DC loads. Should be more like CPU/GFX fansink type or you will need really fast and noisy fans. There is just barely enough surface area.

[Circlotron]

^^ The 300 watt estimate was based on an experiment with a single fan and a heat sink section slightly larger than the fan. Also a single mosfet in the centre. It would do 100 watts continuous but it was fairly working, and 150 watts was pretty rude. Got hot enough to smell it. Put the same mosfet on a 150x70x10mm aluminium slab with no fan and put 150 watts into it and eventually the slab got hot enough to melt solder. Ran it for 1 hour and it made a big scorch mark on the bench top but again it survived. Not the sort of thing you would do if you wanted a reliable commercial product though.

[Circlotron]

Okay, here it is running. I didn't end up using the opamp feedback circuit, just stuck the output of a 0-20V psu that has a 20 turn pot into the gates. About 5-7 volts is the approximate working range; the transconductance is mostly set by the paralleled source resistors. From cold to hot it drifts from 39 to 40 amps. Good enough for my application. A bit over 600 watts in the picture. After about 3 minutes at 15 deg C ambient the heatsink got to about 60 deg C.