WARNING, DANGER OF DEATH
Note - if you are going to play around with mains powered switched mode power supplies, please only do so if you know what you are doing. By their nature, they rectify and store mains level voltages internally, so there is a strong chance there is >300v DC sat on some capacitors, even after you have powered them off and unplugged them. This can also be present on some of the naked internal metalwork, such as the heatsinks. This can kill you. Take note, and do this at your own risk.
All information here is for information only. You have been warned!
I have a pair of Delta DPS-650EB server switch mode power supplies. These came as a pair in a Research Machines 'RM Dualserv', which is so old, and tbh, large and loud, it is no longer any use to anybody. However, each of those PSUs can kick out a healthy 54A of 12v each, and that might be of use to me.
I now know how to power these up! I didn't discover the details - I found somebody on eevblog who appeared to have powered one up, and they kindly shared the details. You can find that thread here.
Bottom line, connect pins T21 (PSKILL?) and T20 (#PSON) to GND. I used 1kohm resistors to do this, but in the original they were just direct shorts. If you are going to use one of these signals as an on/off switch then use T20/#PSON.
Note, after a few seconds the fan spins up fairly fast, and does not decrease, even with no load attached. These are probalby not the quietest PSUs out there.
Originally I did some digging and tried to see if I could get the units to power up outside of the server. Bottom line answer was no. During my 'probing around' (where I had really passed the point of caring) I think I damaged one of my units, as now even with the above two pins connected it clicks and goes into an 'orange LED' error mode.
I'll document what I know here just in case it saves somebody else some time.
The RM Dualserv comes with dual redundant PSUs. They are moderately compact, heavy, Delta switch mode PSUs, each capable of delivering 12v at 54A as their main output. They are rated for 655W or thereabouts.
At one end the unit has the 3pin IEC input connector and the 3 status LEDs. At the other is the fan and the output edge connector.
The edge connector has the fairly normal double sided setup, with two large strips for the GND and +12V (presumably), and 15 single 'pins' (7 on the bottom and 8 on the top, unusually).
Reading around (and if you are messing with this stuff, then you need to read the rcgroups posts, there are normally a couple of pins you need to ground to get similar units to boot. They will be PSON (to turn the unit on), and PSKILL (which indicates if the unit is fitted to a server rack or not for instance).
Searching around the nearest 'equivalent' I could find was a reference to a 750XB unit, that has the same number of pins and a similar 'look'. That unit appears to only have a PSON. Some details can be found in the relevant Intel manual and also in this useful post.
However, testing around the pins with a 1kohm resistor to ground, and then a 330ohm resistor, and then just a plain short produced no results, apart from I think I may have found the +5vsb line, as touching that to ground with a short caused the PSU to go into a reset clicking cycle until I removed the link.
Ultimately, searching around I found that the RM Dualserv was fitted with a Delta RPS-650 'cage', that looks to transform the 650EB pinouts into a more standard 'ATX' style power supply:
That cage contains some reasonably complex extra 'gubbins', mostly I suspect as it needs to take the 12v and convert it into +3v3 and +5v at 24A, so has a few extra reasonably chunky voltage regulators in it. It also must contain the power on circuitry.
I put one of the 650EBs into the frame, and wired the green 'power on' line on the main ATX connector to ground, and voila, out came 12v.
The 'pins' compose of edge connector pads on both sides. They are labelled partly on the PCB as 'T' and 'B' numbers - let's take those to be Top and Bottom. The labelling starts at T1-T25 and then continues back on the other side as B26-B50.
The below table tries to represent what I've found so far, which is not much:
Pin | Function |
---|---|
T1 | +12v |
T2 | +12v |
T3 | +12v |
T4 | +12v |
T5 | +12v |
T6 | +12v |
T7 | +12v |
T8 | +12v |
T9 | GND |
T11 | GND |
T12 | GND |
T13 | GND |
T14 | GND |
T15 | GND |
T16 | GND |
T17 | GND |
T18 | +5sb ??? (shorting to GND resets PSU) |
T19 | |
T20 [short pin] | #PSON ? |
T21 | PSKILL ?. Possibly fan control. 330ohm to T20 slows fan down, even when 12V not powered up |
T22 | |
T23 | |
T24 | |
T25 | |
B26 | |
B27 | |
B28 | |
B29 | |
B31 | |
B32 | |
B33 | GND |
B34 | GND |
B35 | GND |
B36 | GND |
B37 | GND |
B38 | GND |
B39 | GND |
B40 | GND |
B41 | GND |
B42 | +12v |
B43 | +12v |
B44 | +12v |
B45 | +12v |
B46 | +12v |
B47 | +12v |
B48 | +12v |
B49 | +12v |
B50 | +12v |
Note, as far as I can see there are no 'No Connection' (NC) pins on the header - here are some closeup pictures:
These are loud. Even with no load, when powered up in the RPS-650, a single unit ran its fan up to full speed, or so it seemed. Too loud for me to have living in my den/office. Maybe if I find a need for a big PSU far far away (in the garage for instance), it might be workable.
If you do manage to get one of these powered up, inside you will find a couple of small potentiometers that will probably allow you to alter the output voltage, if you need more than 12v. One of the pots (I will bet the on on the RHS) will do that. The other alters the over current trip level I suspect.