My Projects‎ > ‎

Convert HP server power supply for RC use

The following article is nothing more than a project report. I take no responsibility for how others may use this information.
I have been following several threads on RCGroups about converting many types of power supplies for powering RC chargers for a while now, but recently there has been a lot of talk about using server power supplies and so I just had to give that a try.

A big thanks goes out to HP, Dell and IBM!
These companies make servers that serve as the backbone of most companies and even the Internet. Inside each of these servers is a power supply (PS from here on out) and most have redundant, hot-swappable units. To put it lightly, these are well thought out and well made units. This one fact intrigued me but the fact that most were also APFC units really got my attention.

A little about the power supply used here 
I chose to use an HP model DPS-600PB series ESP135 PS that is used in ProLiant DL380 G4 Rack Servers. It is rated at 575W, or 47A at 12V.

#13 points to where a pair of these PSs are installed.

This PS is active power factor corrected (APFC), meaning it will draw about 40% less current than a non-APFC PS. This is a huge plus when running the PS off a generator.

I got my PS off eBay for $18 shipped and then later bought 2 more for $30 in order to take my project to the 24V level, Connect 2 server power supplies in series for 24V.

Modding the power supply to be used outside of a server
My research showed me that I needed to connect 3 pins on the hot-swap connector in order to power on the unit. I used some 22awg solid phone wire to make the connections and once completed, I used a regular computer power cable to power the PS. (see updates)

Next up I had to make the connections to 12V and GND. I used short lengths of Turnigy 12awg noodle wire and soldered them between the spades for each output. There turned out to be a lot of material in the spade connectors, so it was not easy to solder but it was doable.

Once I had those two things done, I was ready to test.

Removing the red handled catch
I didn't like the catch on the side of the unit, so I removed it. I removed the 8 screws from the panel with the labels on it and removed it. Next I took a screwdriver and pried the catch bar up and worked it until it broke off. Once I had it off I placed the panel on the concrete floor and hammered the 2 rivet-like holes flat. After that I reassembled the PS.

Testing the power supply
My first goal for testing this PS was to draw 47A off it and see how it did. I have a Kill-A-Watt, so I connected that to the input and a G.T. Power Watt Meter to the output. This allowed me to monitor the PF, watts and amps on the input and amps, watts and voltage on the output. Below are the results for the test run (see photo of setup).
  • Peak input amps (@ 121VAC): 5.7A
  • Peak wattage: 690W
  • PF at 104% rated output: .98
  • Output wattage: 596W
  • Output amps: 48.3A
  • Voltage @ 596W output: 12.33V
  • PS efficiency at 104% rated output: 86%
How did the PS do? Amazingly well! I found that running both chargers wide open, one charging a 6s pack at 11A and the other charging a 6s pack at 10A (both limited by the 12V source), I pushed the PS to 104% of its rated output, or 596W of its rated 575W output. The PS held 12.33V and did not make a single complaint. No bad smells and no bad sounds, it just sat there and did its job.

These PS units really are amazing. I am very happy with my units and have further plans for them. Stay tuned!

Notes and updates

Nov. 4th, 2011 - Making 24V. Once I had one PS working and tested, it was time to double it up and make 24V. Check out that project here, Connect 2 server power supplies in series for 24V.

Jan. 29th, 2011 - Removing the hot-swap connector. I was reading about someone's build on RGC and saw that the poster had removed the hot-swap board and plug and replaced it with banana plugs. So I set out to do the same. I knew it would not be an easy task but it would be doable. The end result is a much cleaner looking unit.

Apr. 30th, 2013 - Been a long time since I have edited this page and thought it was time to add a few things. When I posted this project little was known about these PSUs. When we started using them we simply tried pins until they worked. Now we know the pin-out and there are many more interesting things you can do with these guys. For starters, we found that shorting pins 6-9-10 together powered them on. Unfortunately pin 9 can be used for adjusting the voltage. Shorting 6-8-10 also turns them on and leaves 9 open for use. Putting a 1K ohm pot between 3-9 allows for output voltage adjustment up to 13.8V. Shorting pins 4-8 will lower the fan speed but the PSU can still raise the speed when needed. 

Removing the pull handle and opening it up.

Removing the catch.

The hot-swap plug. (see updates)

The 3 pins connected together.

The GND and 12V connections.

Chargers drawing 46A or 573W.

The hot-swap connector removed and banana plugs installed.