Server PSU Hack - EMC2 API4SG10 12VDC 300W From a Dell PowerEdge M1000E

[Mads Barnkob]

A guide on how to make the, EMC2 API4SG10 (Dell# 071-000-482) , server power supply run outside of its server enclosure. A simple bridge between control pins next to the 230VAC GND pins is required.

/>
Schematic for the modification

http://kaizerpowerelectronics.dk/electronics/server-psu-hack-emc2-api4sg10-12vdc-300w-from-a-dell-poweredge-m1000e/

Additional information on where it came from is available in the server teardown video, follow the link below
*Dell PowerEdge M1000E Blade Server Teardown*
I came by a discarded full size Dell PowerEdge M1000E server rack. Full SAN, 2 large blade servers, 4 or 6 backbone switches, all power supplies and fans. This is a system that is about 8-10 years old and had a new price of around 1000000 DKR (166666 $US).
It was all unloaded in a big mess into a container and I only had time/space to take a blade server, all power supplies and fans with me. Hard disks all go to destruction and backbone switches and the M1000E cabinet itself is too big and clumsy to have any real value in the work shop.

Discussion of the teardown: https://highvoltageforum.net/index.php?topic=1151.0

[petespaco]

Hello Mads.
I am just curious---
  Is this power supply's output completely isolated from the mains?

Have you done any work to make these server power supplies voltage adjustable over a wide range?
Examples:
12 volt supply:
3 to 12 volts, at full rated current

48 volts supply:
12 to 48 volts, at full rated current

Pete Stanaitis
---------------

[Zipdox]

Hello Mads.
I am just curious---
  Is this power supply's output completely isolated from the mains?

Have you done any work to make these server power supplies voltage adjustable over a wide range?
Examples:
12 volt supply:
3 to 12 volts, at full rated current

48 volts supply:
12 to 48 volts, at full rated current

Pete Stanaitis
---------------

Computer power supplies are switch mode power supplies so they're galvanically isolated from the mains. But the ground of the output is almost always connected to mains earth.

Adjusting the output is possible and you can achieve a couple volts of adjustment but these supplies are so complicated that it's not worth your time.

[rikkitikkitavi]

Hello Mads.
I am just curious---
  Is this power supply's output completely isolated from the mains?

Have you done any work to make these server power supplies voltage adjustable over a wide range?
Examples:
12 volt supply:
3 to 12 volts, at full rated current

48 volts supply:
12 to 48 volts, at full rated current

Pete Stanaitis
---------------

Computer power supplies are switch mode power supplies so they're galvanically isolated from the mains. But the ground of the output is almost always connected to mains earth.

Adjusting the output is possible and you can achieve a couple volts of adjustment but these supplies are so complicated that it's not worth your time.

At 12V supplies GND often is tied to protective earth. It is many times possible to break this connection for a floating supply.

At 48V server supplies they are normally floating from protective earth.

The smaller power supplies can possibly be adjusted  of they are not of the resonant type, but a PWM type. The larger units, some 500+ W are almost always resonant topology and hence very difficult to adjust over a wider range. It would be far easier to build a step down converter with PWM topology.

[Mads Barnkob]

Have you done any work to make these server power supplies voltage adjustable over a wide range?
Examples:
12 volt supply:
3 to 12 volts, at full rated current

48 volts supply:
12 to 48 volts, at full rated current

Usually these 12V power supplies can only be adjusted within about 11-13.2V and with some hackery, some of them can get to 13.8V, but then the fun also stops.

Maybe you make a huge lossy capacitive divider on a 48V power supply, to have taps for 12,24,36 and 48V

[petespaco]

Maybe you make a huge lossy capacitive divider on a 48V power supply, to have taps for 12,24,36 and 48V

Yes, that's what I have been thinking about doing.
I have been saving up MOT's with that idea in mind.  I usually remove the secondaries before I store them.
Also bought a couple of car battery chargers with a similar thought in mind.  They often have center tapped secondaries, so by adding  full wave bridge, I can get 24 to about 30 volts from them, plus the variability I get from the tapped primary.

Pete Stanaitis
---------------

[petespaco]

More on cheap adjustable "high power" supplies:
I just found this video where the guy uses a triac (dimmer circuit)  input to a transformer.  I don't know if that could be scaled up or not.

/>
Pete Stanaitis
---------------

[Mads Barnkob]

More on cheap adjustable "high power" supplies:
I just found this video where the guy uses a triac (dimmer circuit)  input to a transformer.  I don't know if that could be scaled up or not.

Pete Stanaitis
---------------

It can.

10 years back or so, we made a variable 10kVA 3 phased transformer from a isolation transformer and a 20 kW frequency inverter. It worked, but what a noise! We actually feared destroying it from vibrations before overload

[petespaco]

Sorry!
I forgot to add the link to the youtube video for the Triac-driven adjustable power supply:

/>
Pete Stanaitis
---------------

[rikkitikkitavi]

More on cheap adjustable "high power" supplies:
I just found this video where the guy uses a triac (dimmer circuit)  input to a transformer.  I don't know if that could be scaled up or not.

/>
Pete Stanaitis
---------------

https://www.st.com/resource/en/application_note/cd00003856-triac-analog-control-circuits-for-inductive-loads-stmicroelectronics.pdf

Then one must consider that phase control on a 50Hz transformer might pose some heavy burden on the feeding outlet, fusing, triacs and transformer itself. The larger the more stress.

Elektor had a 200X2W power supply project in late 80ies with a variant of the triac control above to phase control two transformers to reduce power loss. A friend did build it and o boy, did the transformers hum.. It must be very dependant on teh transformers build quality, leakage inductance , primary resistance, supply impedance etc etc.

@Mads, are you beeing ironic about a capacitive divider on a recitified stabilized 48VDC or am I just beeing unusally slow and miss something out?

@petespaco, have you considered a variac+stepdown transformer+recitifier+lots of capacitance?
I have a brute force  one, I threw it together when I needed to test a few DC motors from a mobile wheelchair. With a little inductance to it makes pretty good DC actually.

[petespaco]

@petespaco, have you considered a variac+stepdown transformer+recitifier+lots of capacitance?
I have a brute force  one, I threw it together when I needed to test a few DC motors from a mobile wheelchair. With a little inductance to it makes pretty good DC actually.

Yes.  That's where I started.  But my goal was/is to do it "on the cheap", and variacs are usually pretty expensive, especially in large sizes.  I have some around here in the 3 amp to 10 amp range (120 volt single phase). but none that could handle 2 to 4 KW.

[rikkitikkitavi]

O well, I think that phase controlling a transformer in teh 2-4 kVA range would entail some quite high currents. might be difficult. I dont know of any practical solution to this.

But woudl be very interested if such existed.

[petespaco]

But would be very interested if such existed.

Now you have me thinking.  I have seen quite a number of used forklift battery chargers at our local online auction house.
They often sell for $30 to $100 dollars.
DC Output voltages vary from 12 to 48.
Many are single phase, often 220 volts input.
Some are 208/230/460.
Some are three phase, 230/460.
Hmmm----
If they have transformer primary  taps, maybe one could play around with the connection scheme to give them at least some variabiity???
And, for the 3 phase models, maybe even more opportunities for variability???

I'd consider rewinding the primary, adding taps, but most of the ones I seen pictures of have (I'm pretty sure) the secondary wrapped around the primary, so you'd have to tear the whole thing apart to do that.

Maybe 3 or 4 12 volt car batteries and 4 separate battery chargers is the way to go.
Then you'd have 12, 24, 36 and 48 available like I did early on with my induction heater experiments.

Pete Stanaitis
---------------

[rikkitikkitavi]

Yes, that is for sure a method when you can accept discrete steps. Tap selection, primary or secondary can med made with either relays or SCRs.
3ph xformes can probably very versatile.

I have a couple of laboratory powersupplies, 30V 10Amp that select secondary winding depending on set voltage to reduce power loss.

I for myself is trying to get my head around this beauty of aboutish 1,8-2kVA.
(primaries, no not election, to the left of my thumb, secondaries opposite, I estimate 50% of these have usable voltages but unfortunately the insulation is resin impregnated)

BTW, thanks for your work on induction heaters. Much useful.