DSP-600pb Characterization

I got myself 4 old server rack power supplies and I’m working on a project to utilize them for a high voltage source. These pieces of hardware are amazingly cheap off ebay which makes sense given the number of servers around the world.

Each DSP-600pb is rated at ~12V and ~47A meaning a total of 575W of power. The goal is to utilize 4 of these within an adjustable configuration to provide power for any project that I can come up with. Hopefully 2.3kW is enough power!

An interesting feature of these power supplies is that the output voltage can be regulated. This post https://endless-sphere.com/forums/viewtopic.php?t=47415 shares an amazing project put together for high current charging of lipo batteries.

I undertook my own experiments to determine when the power supply becomes unstable and looses regulation or triggers the protection circuit. I utilized an electronic load and a 100k pot that I slowly adjusted.

Characterization Setup. Electronic load connected to server power supplies.
Connections on DSP600-PB connector. On/Off mini switch and 100k Potentiometer.

The results of this test can be seen in the charts below. I was in a time crunch so I tested only 2 out of the 4 supplies which should give me enough data to get a general sense of the output adjust range. Overall these power supplies have a rather limited adjustment range.

The one thing I did notice that although these supplies can be stable over 30 seconds, they tend to droop over time. And when they hit the threshold voltage they will automatically shutoff. From these tests its evident that the useable adjustment range is only around 10.5 – 13.5V. Note that I didn’t try going above 12V, but from my research raising the voltage seems to be easier.

v2.0 EFNMR Power Supply Test

Finally received my new boards from oshpark and started putting together the stage 1 power supply.

Stage 1 is based around the LT3472 and converts the 5V raspberry pi rail into 2 +/- 8 V rail. Noticed an issue when I put it together:

  1. The negative rail was around -6.2V rather than -8V. Looking closer at the datasheet it looks like I misread the equation. Should have set the negative feedback pin to 320k instead of 270k.
  2. Fortunately the Stage 2 drop out voltage of the LT3093 is only around 190mV so there is plenty of headroom between the -6.2V and the final -5V rail.

Did a simple resistive load testing to validate things were still stable up to the expected current draw. The total power draw was a bit higher than expected but wasn’t too crazy to cause too much concern.