Hitachi power supplies

[redtails]

A couple of months back, I bought a bunch of second hand power supplies off of eBay. They were Hitachi server power supplies, labelled HS1182 (2 of those) and PPD5002 (1 of those). In total it costed me like €30 for three of them including shipping, so as for monetary loss on my behalf, it wasn't so bad. In fact I believe I got an incredible deal, seeing what an equivalent power supply would cost. Now I know what people think about using computer power supplies for lab purposes, though my expectations for these things were a bit higher than your average Chinese power supply designs. The scarce documentation claims it's been designed by Hitachi's, though the casing itself claims "Shindengen Electric MFG corporation LTD, Thailand". Your guess is as good as mine who actually designed and produced these things. It is very well possible that Hitachi bought the retail rights from Shindengen Electric co. If you have more info on these things, don't hesitate to reply. As with lots of server hardware, there is no public documentation available.


I won't advertise the eBayer who sold me this (he has tons more of these second-hand industrial surplus things, it's crazy what he has for sale). All I'd like to say is that the box was nice and as for the second-hand nature I must say that it's not really that beat up. I've heard true horror stories of people buying surplus stuff online, though my experiences with surplus electronics have thus far been good.


The size of these puppies (HS1182) is, a bit strange. You'd expect them to be 19 inch due to the wide-spread usage of 19-inch racks. But the width is 26cm, or 10 inch. The front panel has two extruding metal parts and has a  total width of 28.5cm. I'm not sure what's going on with these dimensions. It's about 4cm thick and 25cm long. The dimensions of the PPD5002 is even weirder (just 20cm wide, 4.5cm thick and 23cm long).


The sole reason I got these surplus devices was for their rated maximum values. Yep, it really says 12v 74a and 5v 58.8A. This is one gigantic 1kW SMPS. They're pretty specific about the input voltage and frequency (200-240v ac at 47 to 63 hz). I guess it'll work in Europe, Australia and most parts of the world that use the 220-240v system. I found it a bit weird, because it implies they have a separate model for 100-120v counties such as the USA and Japan. What's further interesting is the datecode. It says 0305 which is presumably february 2003. There's a little green led, a soft powerswitch and some pin headers on there, some with jumpers. I have no idea what the pin headers are for. If anyone knows, don't hesitate to reply.


When I opened it up, I was pleasantly surprised that there are two calibration pots for the 12v and 5v rail. According to my trusty old fluke 73, the 12v pot allows you to adjust the voltage from 11.5 to 13.7v. The 5v rail can be adjusted from 3.6v all the way up to 6v. The thing just falls short of the 13.8v automotive industry magic voltage, though I can only imagine these things to be wonderful for high-powered li-ion emulation. Of course you can do the 5v necessary for many TTL devices. I'm a bit disappointed it doesn't go down to 3v3, but alas.


This would be the big money-shot of this thread (see bottom for full-size). Notice the absolutely massive stabilizing capacitors and all the separation transformers. What I find strangest are the blue wires that are wound around those absolutely massive chokes. They're connected to points only labelled "S" and "F". I'm not sure what they do or what they're for. If anyone knows, please tell me, I'm very interested in it! Most of the chips cannot be identified because they are strongly pushed against various heatsinks. The ICs I can identify are a whole lot of KCH30a10 diodes (these are 30a diodes for rectifying the hf "AC" from the switching transformers). There's a whole load of D5L60 diodes on them (made by Shindengen). The 12v rail has 2 diodes labelled S60SC4M. The datasheet says they're Schottky-type rectifiers again made by Shindengen. there's some SIP-style devices on vertical riser boards, but they're very hard to read. One says "5332M", which I assume is some power-factor correction device.  I've not yet found out how this SMPS precisely works. I can only assume that the grey square devices that are heavily held in place on the heatsinks play a major role. The grey components are about the only components producing heat. If I listen very closely to it, I hear some switching noise coming from it, though it's barely audible. On the front there's a little green led and a soft power switch, implying there is some control circuitry that controls the on/off state.



The last thing I'd like to note is that, as with many power supplies, it's not always "on". There's a soft power switch on the front of this device, but it doesn't switch the power supply on unless there's two pins connected on the back side. I guess this prevents the PSU from wasting electricity when the servers aren't switched on, or something like that. I have some experienced with these Hitachi PSUs; switching them on usually meant connecting one of the pins to the 12v anode (12v and 5v rail do NOT share anodes).  It usually only takes a couple of minutes before you hear a loud click from a relay. The very obvious down-side of this thing are the strange AMP/molex connectors. I don't have any connectors that even remotely resemble these things, nor do I have any information on where to get them. If anyone knows more on these exotic molex connectors, please tell me! I usually just stick breadboard cables in them, they hold quite firm. There's no way I could use the full 58/74A through breadboard cables, though. I was seriously considering desoldering the AMP connectors and soldering on something else with a decent standard connector, not like these things have much value anymore.

Although I'm not going to use these things to power up any servers or harddrives, I'm overall very pleased with the solid-steel construction and engineering quality that has gone into this thing. It's too bad I have to bodge the thing with a little red wire to boot it up, but apart from that there wasn't anything bad inside that I could see! Anyway, this is all what I wanted to write. If you want more pictures or just discuss this: a reply is just a click away!

[T4P]

The "absolutely" massive caps are mains filtering caps not stabilizing caps 

Unless you're talking about the numerous parallel caps that are so that they can make the PSU very short in height but also increase the ripple current capability significantly in parallel
But from what i see there aren't any fans right? seems VERY efficient to have only so few heatsinks for a 1kW PSU

Were they forced induction cooled when they were in the server rack or what?

[GeoffS]

Any chance of you resizing those photographs so they can be seen?

[BThunderW]

Correct me if I'm wrong, but if these are switching power supplies, don't they always have to have a load while on? Otherwise they'll blow. When I converted some computer power supplies to benchtop supplies, I always made sure to put a high wattage power resistor across each rail.

I've seen an unloaded PS blow up before.

[redtails]

Any chance of you resizing those photographs so they can be seen?

I'm uncertain what you mean. Most images are 800×534, and the largest one needs to be big to give a nice overview. Or did you mean the images are too small?

The "absolutely" massive caps are mains filtering caps not stabilizing caps 

Unless you're talking about the numerous parallel caps that are so that they can make the PSU very short in height but also increase the ripple current capability significantly in parallel
But from what i see there aren't any fans right? seems VERY efficient to have only so few heatsinks for a 1kW PSU

Were they forced induction cooled when they were in the server rack or what?

Oh right I used the wrong term! The three 450v 390uF capacitors would be mains filtering caps, you're correct. The 12v and 5v rail both have their collection of parallel filtering caps. Yes, in a typical server setting, these things would be situated above or below a storage server (I think they're called "HDS", or Hitachi Data Systems). Each rack would have its own dedicated cooling; a combination of forced-air and heat dissipation by fluid flow. The small heatsinks are tightly screwed on top of the top cover, further increasing the heat dissipation to the outside world. I currently don't have the equipment to rev it up to 70 amps to see how it handles at full load, nor do I actively cool it. It barely gets warm to the touch when I keep the load below 10a. It will probably require active cooling when forced higher. Thanks for your comment on usage of the right terms!

Correct me if I'm wrong, but if these are switching power supplies, don't they always have to have a load while on? Otherwise they'll blow. When I converted some computer power supplies to benchtop supplies, I always made sure to put a high wattage power resistor across each rail.

You're not wrong about this, you're very correct about this, in fact! Cheap computer ATX power supplies don't have separate rails. Without load on 5v, the 12v will hover around 9-10v in most designs. Same applies to the 3v3 rail. They are dependent on one another to function correctly. It turns out that in this server PSU, the rails are entire separated and do not influence each other. If I set the 5v to 5.35v, it will maintain that value even if I short-circuit the 12v (very bad idea but alas). I can't speak for its inherent stability, as I don't own an oscilloscope, though it's still functioning after all this time, so it can't be that bad I hope. There are a bunch of cement 10W resistors on the board (they're hidden under the flat sheet-metal heatsinks. My assumption is that these provide the minimum load to assure good functioning

[GeoffS]

Any chance of you resizing those photographs so they can be seen?

I'm uncertain what you mean. Most images are 800×534, and the largest one needs to be big to give a nice overview. Or did you mean the images are too small?

The largest image is nearly 3000x3000 pixels which is difficult to see on an average screen.
One suggestion is to include images as attachments rather than inline that way the fourm software will provide suitable thumbnail images that can be expanded when viewed. The other option is to provide a link that can be followed to the site where the image is stored. This gives the user the choice of viewing or not.

[krish2487]

Notice the absolutely massive stabilizing capacitors and all the separation transformers. What I find strangest are the blue wires that are wound around those absolutely massive chokes. They're connected to points only labelled "S" and "F". I'm not sure what they do or what they're for. If anyone knows, please tell me, I'm very interested in it!

Try "Start" and "Finish"
 

Most of the high frequency ferrites, especially multiple winding transformers are directional/polar in nature.

The "S" and "F" are used to denote the start and end of the winding, though it is usually mentioned in the construction/mechanical details of the transformers, it is not uncommon to see it labelled on the board too!!!

Hope that was of some help.

[SeanB]

I would guess they are a current sense coil, probably used by the active PFC circuit to measure the load on the supply. Otherwise they could be used to do load sensing for report to the server as to power supply loading.

[prodiv]

Hello all!

Does anybody know how to turn on the HS1182 power supply? I have one and I'll appreciate any help you provide.

Thanks in advance!

[prodiv]

Well, 20 minutes with tester and I found the answer. So want to share...