Converting HP ProLiant Power Supplies for Amateur Radio Use

Robert Kilian, K6RBK

k6rbk@ostechnologies.net

Inspired after glancing at the DC output ratings on a failed HP ProLiant DL360 G5 power supply unit during a datacenter visit and thinking “wow, this little supply is awfully compact compared to most desktop switchers” and the fact that they’re designed to run at/near 100% output for years on end, I searched the ‘Net for some information on converting ProLiant PSUs under the assumption that surely someone must have already done a conversion.  Sure enough, I stumbled across quite a few threads on various R/C forums and this excellent document from Ohio Packet which is the technical foundation for this document and goes into much more detail on the technical aspects of conversion than I will here.  My focus is the actual modification of HP power supplies and research into the newer higher-output PSUs that I’ve not yet found conversion instructions for (some of the newer ProLiant PSUs can do 12V @ 56A or higher). This is a great way to build a “home-brew“ power suppy for cheap!

HP P/N 194989 Modification Process:

Rather than attempt to source a female connector to mate with the PSUs male connector, I simply soldered some jumpers between the pins that needed to be connected (for complete pinout, see Ohio Packet doc) as shown below:

Figure 1 -- Bottom side of connector

The wide pin on the right (pin 1) is ground (as is the wide pin on the left).  Pins 5 and 9 are jumpered to ground.  Pin 10 (where the jumper on the far left is seen) is connected to the 12V pad on the other side of the connector.  Also connected to the 12V top pad are two 8 gauge radio power leads terminated in ring lugs on the PSU side (only because they were already attached to this particular radio power lead set and provided a clean solder joint) and 30A PowerPoles on the other end.  The small jumper on the right is from pin 10 on the bottom of the connector:

Figure 2 -- Top of connector

The two outer smaller pads are ground; the center pad is 12V.

After getting this far I went ahead and tested the output and noticed the voltage was a little low (around 11.8V) and decided to open up the case to see if I could find the voltage-adjust trim pot mentioned in the Ohio Packet document.  After a quick peek and a little testing, I located it:

Figure 3 -- Voltage adjust trim pot & bypass jumper

The voltage pot is a 2KΩ trim pot shown within the red rectangle (the trimmer on mine snapped off when I removed the “glue” holding it in position).  Before the trimmer fell off, I was able to adjust for a maximum of 12.1V.  In this particular unit’s case, I then jumpered the two pins circled below the pot (thus effectively bypassing the broken pot).

Results:

With these modifications in place, the PSU has no problem supplying enough power to run my FT-857 or FT-840 at 100W in packet mode.  Voltage drop *on this particular unit* is about 0.3V with the radio at full power; still within Yaesu’s spec for the radio (13.8V +/- 15%).  According to the Ohio Packet document, these supplies are adjustable out to 12.8V.  It’s very possible that I have an overly “tired” unit.

Possible improvements:

        - Rather than soldering the 12V power leads directly to the connector, solder screw terminals to to the ground & 12V pads to allow for field-detachable/replaceable leads

        - Connect both ground pads on the top of the connector instead of using only one

        - Add a switch to power the unit on/off without having to disconnect the AC line

HP P/N 411077 Modification Process:

I could not locate any conversion information for this PSU (mine is a failed unit out of DL360 G5; the failure must be somewhere besides the 12V circuit because it will not power up a server but works fine to power a radio with 12V).

Using some clues I found in this thread at rcgroups.com I was able to identify the relevant pins needed to turn this PSU on.

This modification is quite similar to the HP P/N 194989 (and many other PSUs) in that only two pins (on the top of the connector) need to be grounded:

Figure 4 -- Pins 4 & 7 grounded w/ 330Ω resistors

Initially I misidentified the 12V +/- pins based on what my meter indicated; I can only guess with the 12V circuit “off” there must have been a reverse bias on the 12V pins causing the incorrect reading.  Pins 4 & 7 were identified as the pins needed to turn this PSU on thanks to some information on rcgroups.com that mentioned that at least one of the short pins (4 & 5 here) need to be grounded, and that the relevant pins to turn on the PSU will typically show somewhere around 4V when the 12V circuit is not on.  After checking all pins for their voltage values, pins 4 & 7 were the only pins around 4V (3.7 or so on this PSU) so I used aresistors to ground them (just in case I was wrong, the resistors would limit the amount of current going where it shouldn’t be going).

Results:

A quick test of this PSU with no load showed approximately 12.3V; with an FT-857 @ 100W max in packet mode into a dummy load it showed a max voltage drop of 0.5V -- which could be attributed to the DC supply leads being used.  With no fan, the PSU did get warm after around a minute of testing.

Further modifications & testing will need to be performed on this supply to identify the voltage adjustment pots as well as an appropriate cooling method.

Also, because this unit is “broken” (would not power on a server) there may other issues that I have not yet identified, however I suspect the issue is probably related to either non-12V circuit or a fault with the PSU’s ability interact w/ the server’s on-board voltage regulator.

Possible improvements:

- Add cooling

- Identify & document voltage adjustment process (apparently some newer PSUs do not use trim pots for voltage adjustment but rather a varying resistance between two pins)

Robert Kilian, K6RBK

k6rbk@ostechnologies.net

July 9, 2012