High Current Bench Power supply

       I recently found myself needing a high current bench supply for a few reasons: Bench testing high power car amps, running power soak tests on repaired amps or even running my car stereo in the garage when I'm tuning or listening without the need to have the engine idling all the time to name a few. High current lab supplies >10A are very expensive so I found another option. RC guys have been using these surplus Dell server power supplies to run their high power hobby chargers so I got one and tinkered with it.

       This project is basically just converting a surplus server power supply into a useful bench power supply for testing amps. This one is for the Dell 2850 server and is rated at 12V 57.3A output and a pinout is shown below.

       When I got it, the internal pot (VR601) is cranked all the way counterclockwise. This gave an output voltage of 12.8V unloaded. To trick the PSU to current share by connecting the 3.3V to the current share pin using a 10K resistor, I was able to raise the output voltage to about 13.17V unloaded. But this is only good for up to about 10-15A where it then drops back to the 12.8V set voltage as the load current increases which would be fine for testing amps and simulating voltage droop of a stock electrical system.

       But I didn't want to stop there. Being a bench PS, I will need displays for voltage and current. There are a bunch of digital panel meters online that can fit any budget, style and accuracy. Instead of ordering and waiting for another one, I went a different route. I had a couple of those cheap Hobbyking P0 Wattmeters and since they can do >100A, I thought it would be a nice prebuilt solution.

       I took one and tore it apart. The "shunt" was a 6inch lenght of #12 wire which measured to be about 1mOhm but since copper has a rather high tempco, I replaced it with a strip of 1mm stainless steel plate. Stainless steel has a much lower tempco (not perfect but pretty low compared to copper) and a higher resistance so I can get away with a shorter lenght.

       I also tore apart a broken cheapie MP3 player and dissected the LCD display to get to the LED backlight and assorted light filter films with it. Cut them to size and hot glued it behind the original wattmeter LCD and there you have it. a white backlit display. Check out the pictures of how I put mine together.

Dell PSU nameplate Model NPS-700AB.

Here is the dissected wattmeter and broken MP3 player.

An isolated DC-DC converter powers both the wattmeter and the added backlight LED.

DIY connector using perf board and two-row female pin header. The other black box is the DC-DC converter.

Plugged into the powersupply connector. This allows access to the powersupply control pins.
12ga silicone insulated wires for the high current output is soldered directly to the pins.

Test setup.

Nice white backlit display.

Attached everything to a piece of acrylic to test the PSU performance.

Side shot of the test setup.

       If you noticed, there are two switches. One switch acts as the main on/off switch by connecting PS_KILL to ground to turn the PSU on. The other switch is used to turn the main 12V output on/off by connecting -PS_ON to ground to turn the main high current output on.
       In theory this was nice, but in reality, I found that that turning everything off (PS_KILL → open) is not good since the PSU still consumes a bit of power but the fan isn't running so it got very hot doing nothing so I recommend keeping PS_KILL shorted to ground and unplugging the PSU when not in use.

I decided to remove the other switch so here's the clean up process.
Replaced the original fan with a thermally controlled fan and turned it around.
Blue thing sticking out the fan motor is a thermistor. This one has built in thermal speed control.

Back view. Here, you can see the DIY current shunt and wattmeter PCB.

Plugged in the auxiliary connections for control and meter PSU.
Meter PSU is derived from the standby 3.3V supply then used the DC-DC converter to step it up.

Output turned on. I raised the output voltage by adding a 10K resistor between the 3.3V and current share pins since there are no internal pots that adjust the output voltage.

Tested it by hooking it up to the car battery with the stereo and air compressor running with the engine off.
I was using crappy wires with alligator clips that's why the reading seems a little low.

Power supply front. The switch turns the output on/off.
The binding posts are very old bakelite ones which have all solid brass hardware.

Side view of the cleaned up Bench PSU.

Front.


       4 May 2014:

       Moar Powah!!! I did some more tests hooking it up to my car battery using real 8ga wire with real copper in it, running all the lights, stereo and air compressor on, I was able to load it up to the max. With the pot (VR601) cranked all the way counter clockwise and a 10K resistor from 3.3VSTBY to current share pin, this is what I was able to get.

       Without having to remove the board from the case, I probed for a point where I could artifically raise the ouptut voltage. By adding a 27K resistor between the left contact of VR601 to ground (see pic) I was able to raise the voltage up some more. I can take it up to 13.7V but then the PSU shuts down due to the OVP kicking in. Keeping it at 13.5V lets it operate reliably without the protection tripping. VR602 sets the threshold where the voltage folds back to the set output voltage (when the PSU is not forced to current share).

And once that was done, this is what I was able to get. It does get pretty warm with that much loading so I don't recommend running it for a long time at that current.

       For even more power, I tried removing the resistor I added to force it to current share. Turn VR602 fully clockwise and set the output voltage to 13.55V and loaded it up.

       It worked once and I was able to take the pic above but later on, the protection trips and the fault LED kept lighting up. This PSU has a very extensive protection circuitry that it is easily tripped so I returned back to the previous setup with the forced current share and voltage set to 13.55V unloaded. It is still plenty for testing amps or when I'm doing something in the car that needed a continuous supply without having the engine on all the time so it's not too bad.


       5 Aug 2014:

       I read comments on the wattmeter that I used and a lot mentioned that it suddenly stops working for no reason.
I thought I might be lucky since mine was working fine until one day, the display went blank.
So I took it apart to replace it with a dedicated panel meter which displays both voltage and current.

       Here is the new panel meter that I will use. It uses an external shunt but the supplier forgot to include one in my order.
I will contact the supplier about it but I don't plan to use the stock shunt here since it is too big for this application.

       Used the mill to cut a new large rectangular hole for the DPM.
I also replaced the switch with a smaller toggle switch since there is not enough room left for the old one.

       Using a bench supply with CC mode, ammeter and millivoltmeter, I measured the old shunt to be 1.31mOhm.
I will have to change it since the meter is designed to use a 100A/75mV shunt which is 0.75mOhm.

       Drilled new holes and cut the DIY stainless steel plate shunt.

       Unfortunately, I miscalculated and it was lower than I was hoping for at 0.66mOhm.

       Loaded the supply and calibrated it with my ammeter. Fortunately, the adjustment pot has enough range that I was able to make it spot on.

       I also decided to do the temp controlled fan mod on this PS as mentioned in a PDF I found in the internet.

       The LM34 temp sensor is wedged between the fins of the main PS switching devices.
(I later added some thermal grease to it for better thermal contact.)

       I also put back the original fan since the replacement was noisier vibration wise.

       Here it is in use charging a large battery bank.

       I do miss the watts display on the old readout but I'm hoping this one will last longer than the old one.


       06 Jan 2015:

After getting cut up messing with this stiff welded stainless screen, I managed to make some protection screening

It did make the PS look nicer rather than bare looking.


Page created and copyright R.Quan © 26 Apr 2014.