Server Redundant Supply Connections

[oliveira_tec]

Hello everyone!

I have a couple of this type of redundant power supply, from a server with 3 of it:

SP-252-TC (PSW-0035-M)
250W
(Ablecom)

They stop working some days ago; I'm trying to fix them... I think is a problem with the cap's that are too close to a resistor that heat's and then they get 'empty'. I've allready change to news one's but my real problem is that I can't get them to real work: I want to make them output the respective output's before installing in the server but I think, like there are too much pins, i'm missing some connections too activate them out of module of the server. Someone knows the pinout of this kind of power supplies? or where I can get some information about the connections?

Thank you very much. (Portugal)

Sorry the 'good' english writing

[370forlife]

Re: Server Redundant Supply Connections

You'll need to find the PS-ON connector on the part that connects to the module, and continuously short it to the large one labeled ground.

Only problem is that it might need a minimum load on it

It dosen't look like it does, usually the label will say what the minimum load need is on each rail.

[hardwareguy]

Re: Server Redundant Supply Connections

One of those capacitors looks like it may have a seal failure on the bottom.

[ratdude747]

Re: Server Redundant Supply Connections

second the one with the seal failure. fix that.

[linuxguru]

Re: Server Redundant Supply Connections

Yup, most probably Teapo SC - it's worth replacing all of them at one shot.

[ratdude747]

Re: Server Redundant Supply Connections

oh, and what leaked out of the bottom appears very burnt. dont be fooled, that is not glue. probably smells great!

[oliveira_tec]

Re: Server Redundant Supply Connections

Hello!

Thanks for your time folks.

I'have allready put two new capacitors picked from a LCD power supply that I dont need it a the time; This caps are not very easy to find here because the small external diameter. The fault of the caps get 'burned' is on the bleeder resistor next to them. My intension is to move it a litle far to a point that the air flow greater.

370forlife, my bad bad luck is that i dont have the module with me so i cant follow the wires to the PS_ON switch
I tried to short all the signal path's to the GND one at the time but I get no positive results. I have in one terminal +5vdc that i think is the +5v_Stand By; I shorted all the signal path's with this output but results are the same: nulls. The supply only get's the Led light on red and the +5v_Stand By; the power outputs still dead.

It seems that this type of supply is more hard to fool, or still there is a major failure.

A big hug to all!

[370forlife]

Re: Server Redundant Supply Connections

Are those genuine rubycon capacitors?

If it won't start still, it might need a minimum load then. That will prove to be very hard to try. You would have to hook up a hard drive or two for a load, and that would be hard with the connector on the power supply.

[ratdude747]

Re: Server Redundant Supply Connections

and the minimum load strikes again...

[archimedesmp]

Re: Server Redundant Supply Connections

I figured out how to turn it on

Let's look at the unit from the back, placed such that the PCB is on the bottom and you can see the PCBs top layer with the markings "+5V", "GND", etc.
I suppose the purpose of these connectors is obvious ;-)

The rightmost connector is the ON switch (corner of the PCB, group of 4), which is sitting at about +5V if the unit is idle. This needs to be pulled low. But unlike with ATX, shorting it to GND is not enough.

Flip the unit and take a look at the PCBs bottom plane.
The connector layout is quite similar, and in fact, the big connectors carry the same voltages (GND, +3.3V, +5V, +12V). There is also a group of 4 (now on the left side) and a group of two (on the right). These are now DIFFERENT than on the other side.
Also, you'll find there is one additional connector in the middle, sitting between the huge GND connector and the +12V connector. That one needs to be connected to GND, too.
The backplane my units came with permanently connects this connector to low, so you should be safe doing the same. There are even holes in the PCB behind the gold tips, which you could repurpose.

There is also a 5V standby (now outermost on the left, part of the group of 4), but the unit runs quite warm just sitting there. So if suppose if you use it in a project you're better off pulling the plug.

My unit erratically was turned off by the controller sitting on the backplane while the ATX connector was shorted out. No idea if that's just because the backplane senses the PSU is idle, or if it was some low-load protection kicking in.

I'm too lazy to draw a pin out diagram, but once it's turned on, you can use your DMM to figure out where to find the non-labled ATX voltages. They should all be there.


Oh, err... and: Sorry for the necro'ing the old thread, but this is one of the the first hits on Google when searching for this unit - and maybe someone might find this helpful. Just let it vanish back into the depth of the forum

[Stefan Payne]

Re: Server Redundant Supply Connections

Aren't there guides for the always on mod on the internet these days?

[PeteS in CA]

Re: Server Redundant Supply Connections

OK, on a don't ask me how I know basis:

* That short pin on the bottom of the PCB is probably the enable pin (shorting it to ground enables the PSU);

* The wide voltage plane on the bottom of the PCB is probably the output return;

* While the date code on the two electrolytic caps are either December, 2004 or 12th week of 2004, I'm pretty sure the coding of the S/N means that it was produced in week 18 of 2005;

* This model has been obsoleted by the server manufacturer.

Speaking generally, short or long pins indicate signals or voltage plane the manufacturer wants to "make" first or last during insertion or "break" first or last during extraction. I remember implementing this sort of thing on a power supply design back in the mid-late 1980s. For example, this PSU was designed so that output return "made" before the enable pin did. Similarly, when extracting, the enable pin would "break" before return did.

BTW, Ablecom is a Asian contract manufacturer that specializes in assembling chassis, with power, backplane, and cooling. This PSU may or may not be Ablecom's design, but it was assembled by Ablecom or a sub-contractor. A lot of US computer companies have a variant of this production process model: have basic chassis and system assembly done by a contract manufacturer and do final integration or configuring done in the US (or EuroLand). Apple and Dell were doing this already in the late 90s.

[eccerr0r]

Re: Server Redundant Supply Connections

This got me looking at one of my server's hot swappable PSU modules, unfortunately it looks a bit different than this. The one I have is a DPS-350MB made by Delta and curious how it was designed. The unit is *so* completely packed with components you can't see light from front to back, I wonder about airflow... probably depending on external airflow or perhaps this unit is very high efficiency?

[PeteS in CA]

Re: Server Redundant Supply Connections

This got me looking at one of my server's hot swappable PSU modules, unfortunately it looks a bit different than this. The one I have is a DPS-350MB made by Delta and curious how it was designed. The unit is *so* completely packed with components you can't see light from front to back, I wonder about airflow... probably depending on external airflow or perhaps this unit is very high efficiency?

You should see the 280mm L x 73mm W x 38mm H 2KW PSU sitting on my desk. It truly is !

[eccerr0r]

Re: Server Redundant Supply Connections

Makes me wonder why my ATX 380W PSU is dumping so much heat right now... plus it has a lot of dead space

Besides that fact, I was wondering if I should get a third one of those server PSUs so I can actually have redundant PSUs. The machine is a 2 of 3 redundancy - need two good units out of three. Alas I don't use that machine...