50V 100A 500W Load Tester

Discussion in 'The Projects Forum' started by tom66, Dec 9, 2011.

  1. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    Plan: To build a load tester capable of testing computer power supplies, CPU VRMs and other various power circuits.

    Specifications:
    * 0 - 50V source voltage (55V surge) (Not reverse polarity protected)
    * 0 - 100A current (100A at >0.8V)
    * 0 - 500W dissipation (up to 1kW for 10 seconds)

    It will be based on an old amplifier case which I have recovered from an amplifier I found in a skip. The amplifier has a large heatsink. The unit will be air cooled, or maybe even water cooled, if I can work that out.

    Hardware plan:
    * 2 x 0.01 ohm (paralled) 100W current shunt resistors attached to the main heatsink. Both devices dissipate 25W each at 100A, so they are operated well within ratings.
    * 150A automotive fuse (rated to 63V.)
    * 18 x IRFZ44VZPBF mounted on the heatsink, evenly spaced, with mica pad and thermal paste. Each fet dissipates ~28W at full load.
    * Kelvin current sensing across each shunt resistor (individually summed.)
    * Kelvin voltage sensing optional; by default sensed near the main terminals.
    * 16x2 LCD.
    * Optical encoders for setting parameters.

    Software plan:
    * PIC controlled (dsPIC or PIC18.)
    * Can set constant current, constant power and constant resistance.
    * Can view voltage, current, power, simulated resistance and temperature.
    * Support for transient and varying loads (for testing transient response) as well as fixed configurations.

    TODOs:
    * Figure out what type of wire used to route 100A through the device. I have a lot of wire removed from computer power supplies. It's AWG18. This can carry about 14A per wire. So I was thinking of paralleling a lot of these wires (x10 each way) together to route the internal high power bus.
    * Figure out how to fit the fuse - for now it's going to be soldered in, but it would be better to have a fuse holder in case the fuse blows.
    * Figure out how well the amplifier's heatsink will work with air cooling, and if it will be good enough for 500W (I think it will be.)
    * Design the main control board.
    * Think about the best connector for the power terminals. Initially I will just loop a lot of AWG18 and maybe use some kind of adhesive or solder to keep it together... However maybe some kind of high current banana plugs would work better?

    I've already got the power shunt resistors and thermal paste, and more components are on their way.
     
    Last edited: Dec 9, 2011
  2. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    (Stash point.)
     
  3. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    (Stash point #2.)
     
  4. R!f@@

    AAC Fanatic!

    Apr 2, 2009
    8,754
    760
    You must be crazy..................
     
  5. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    I built one here (called Big Dummy);

    [​IMG]

    The two heatsinks came from a junkyard piece of equpment and were already populated with the big transistors, already wired as darlingtons. That was common in high power linear power supplies, 1960's 70's 80's etc.

    It uses a pot to set the current. The biasing is set up at the moment to work with about 5v-48v and 0A to 30A.
     
  6. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    Variable current sinks aren't cheap. He'll be able to build one for far less than one could be purchased. Though I think he needs to double the ability.

    1000+W Power supplies are more and more common, and I've seen several pass the 5 and 10 amp tests, but fail on higher loads. 12V Rails are rated over 100A these days. :eek:
     
  7. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    You only live once!
     
  8. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    The goal with this is to make it expandable, so I'll leave some space on the heatsink to fit another 18-36 MOSFETs, effectively allowing 1000W or 1500W dissipation. As it's a single rail only device, to test computer power supplies I'll have to build a smaller dual rail 150W unit, for loading the 5V and 3.3V rails, and then just use some resistors for the -12V and 5Vsb lines. But those are future plans.

    Looks very nice. How much can the array dissipate without active air cooling? Can you really put 30A through those banana plugs?
     
  9. bertus

    Administrator

    Apr 5, 2008
    15,652
    2,348
  10. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    Resistors put on the heatsink with a liberal amount of thermal paste. Next, I will drill through the metal, and mount them on using some bolts. Then, I will decide how best to place the MOSFETs along the heatsink.
     
  11. paulktreg

    Distinguished Member

    Jun 2, 2008
    612
    120
    I'll watch this thread with interest.

    I've tested ATX power supplies for a while now using resistor networks. I can fully load a 1000W power supply but I need something a little more flexible.

    [​IMG] Uploaded with ImageShack.us
     
  12. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    Looks nice. This is only a single rail load tester, though. You'll need some resistors for other rails, or lower power load testers. Would be good for loading the 12V rail. In its current incarnation, it would be okay up to 41A on the 12V rail, but future upgrades could up that to 100A (which is 1.2kW.) 1.2kW is very difficult to dissipate effectively.

    --

    Meanwhile, I soldered the bridge between the two resistors. 8x AWG18 wires from a computer PSU are bundled together and soldered into one large high current wire, capable of up to 128A. (The wires need only carry about half of the full current, or about 50A, because the current is split between the two resistors.)

    The small wires coming off the resistors are the kelvin sense wires. To maximise accuracy (eliminating resistive losses), they are soldered extremely close to the resistor body. The current accuracy will be initially ±5% (as that is the tolerance of the resistors) but should be easy to calibrate using a multimeter to around ±1 to ±2%.

    The kelvin wires produce a differential voltage. As each resistor has a ±5% tolerance, two differential measurements are taken and summed/averaged together. Each side of the resistors is measured.

    I found that removing the resistors to solder them doesn't do the paste any good; it looks like I will have to re-paste them. Thermal paste is expensive, so I've learned my lesson... :rolleyes:
     
  13. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    Current shunt array complete.

    Have also drilled through the heatsink (steel). Need to get some M3 bolts to mount them to the heatsink. Am also thinking of fitting a clamp to push them onto the heatsink made from some bent metal probably.

    Yellow is the "positive" end of the resistor.

    I tested the resistors under load. With an uncalibrated 3A current, they dropped 14.2mV. The predicted drop is 15mV, if they were truly 0.01R. That's an error of about 5.3%; considering the uncalibrated current source (a benchtop power supply), that seems about right. What was interesting is when I first tested it, I just probed at the power supply and was getting about 45mV! - a substantial error introduced by the wires.

    I was thinking of adding a precision 1000mA current source on the board to test the resistor array with. It would be inserted into the circuit by a relay. It could be a calibrate option, so the unit can reduce the error introduced by the resistors. Calibration when the unit is hot is also possible.
     
    Last edited: Dec 10, 2011
  14. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    I'm not sure of total dissipation, I never bothered characterising the heatsinks. In use if it gets warm I have a small 8" plastic desk fan I sit next to the thing. There are 16 big TO3 transistors that do the bulk of the dissipating, they are easily good for 50W each depending on derating for heatsink size of course.

    The banana plugs seem to be fine for 30A, they are the large older style ones, with a solid brass stud (tin plated) and on the back has a round lug and soldered leads. The front has options for banana jack, or for higher currents a round lug or twist a wire around the stud and screw it down.

    Are your resistors rated for 50 amps each? Or peaks since you mentioned peak tests? The resistance wire inside will fuse over a certain current as it will be much hotter than the metal body of the resistor. Generally I would have used a commercial 100A or 200A shunt for that job.

    What are you using for the main dissipation? I've seen some reliable high power dummy loads using jug elements, or elements from stovetops or hot water systems.
     
  15. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    This is something I will only find out from testing it under load. 100A is the maximum current and will never be exceeded. 150A will blow the fuse. The resistors are capable of dissipating 100W, so pumping 50A each through them shouldn't be a problem (25W dissipation in each.) They are some eBay specials, though... http://www.ebay.co.uk/itm/400259682230

    I am using 18x IRFZ44VPBF MOSFETs mounted on the large heatsink. I am using these particular ones because they are surplus and going cheap :). On-resistance is not really important, except for maybe short circuit (100A) tests.
     
    Last edited: Dec 11, 2011
  16. MrChips

    Moderator

    Oct 2, 2009
    12,452
    3,371
    I would like to know how do you connect to your dummy load? Isn't the wire or banana jack going to burn up?
     
  17. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    Some very thick wire will be molded into the hooks on the front of the unit, and I'll probably buy some short lengths of AWG 4 for test leads.
     
  18. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    To make the hooks, I bundled together stripped wire and covered it in a liberal amount of solder. Unfortunately, I couldn't quite finish the other one, because I ran out of solder! :rolleyes: Ordered some more...

    Edit: Anyone know how much current solder can handle? These connectors are more solder than copper! (I'm using tin lead 63/37.)
     
    Last edited: Dec 11, 2011
  19. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    4,771
    971
    OMG that is UGLY.. Just go to the darn hardware store and buy some proper ring terminals..
     
  20. tom66

    Thread Starter Senior Member

    May 9, 2009
    2,613
    214
    Heheh, but do you know any ring terminals that can handle 100A? This should be capable of 100A. I'll use them until I find something better.
     
Loading...