2

I have a LITEON Server PSU which I would like to use for an electronics project. For the PSU to 'start up' I assume I need to connect a certain combination of the small pins (that can be seen on photo 2) together (similar to how one 'jump starts' an ATX PSU (example: https://forum.overclock3d.net/showthread.php?t=394).

The power supply specs: Brand: LITEON Model: Cisco Catalyst 3K-X 1100W AC Power Supply C3KX-PWR-1100WAC V02 enter image description here enter image description here

I have tried finding what combination of pins I need to connect together to 'jump start' this PSU but have had no success in getting it to work thus far.

I have measured the voltages over the small pins relative to the two 'power' prongs/pins (the big contacts on the left and right side of the 16 small pins), the two 'power' pins have a voltage of 0V across them (since I have not yet managed to turn on the PSU). Some of the small pins have a voltage relative to the 'power prongs', I have measured all small pin's voltages relative to the left side 'power' pin(negative multimeter wire on left side power pin). The voltages are as follows: enter image description here

I noticed 2 of the 16 power pins are shorter (these pins' location is marked in the image below): enter image description here

Perhaps it is not a case of simply connecting two of the pins (or connecting 1 of the 16 pins with GND) but instead it might need a signal originating from the server chassis in order to turn on.. In that case I might be out of luck since I don't have the chassis, only the PSU unfortunately. (it was suggested in the following link that a digital signal may be needed: https://www.techpowerup.com/forums/threads/how-to-turn-on-power-supply.147023/ (check post #16))

It was also suggested on the same page that there might be a way to open up and find a way to turn it on that way: "Personally I'd open it up and have a look at it. Chances are you can manually trip a relay or jumper a couple solder joints and get it to turn on" (post #7).

It was also suggested that if it were a matter of connecting 1 pin to ground to get it to start this might be a good method of doing it: "Personally i would hook one end of a ~1k resistor to GND. Probe each wire with the other end. when u find PS-on it will turn on. " (post #16). Is this a good way to do it, could this potentially damage the PSU?

I have added some photos of the internals but if more detailed photos are needed please let me know. If any of my explanation is unclear/incomplete please let me know.

Any help will be greatly appreciated, thank you very much! :)

enter image description here enter image description here enter image description here

Maarten -Monica for president
  • 330
  • 1
  • 4
  • 21
  • @TonyStewartSunnyskyguyEE75 Thank you for your comment. Could you please elaborate on what you mean by 'bootloader ?' ? (I am quite new to this). By looking at the document I assume your suggestion was (if indeed this PSU needs a digital signal to be sent to (some of) its pin in order to turn on, to generate the required signal using the info found in the document? And to somehow send that data to the appropriate pins via my pc? My assumption could be totally wrong, please do correct me if so. Thank you once again! Note: I unfortunely only have the PSU not the server chassis. – Maarten -Monica for president Jan 15 '20 at 19:20
  • @TonyStewartSunnyskyguyEE75 ok no problem. Could you please explain what you mean with "Bootloader ?" though? That would be fantastic. THank you – Maarten -Monica for president Jan 15 '20 at 19:23
  • It may need two pins grounded to turn on. Do _not_ take it apart. With it switched off, measure resistance between each pin and negative power terminal. Thus you can filter out ground (0 ohms) and ground sense (tens ohms). There is likely a group of two pins with equal positive voltage and resistances - i2c bus. The remaining pins with positive voltages are likely candidates for grounding (through a resistor!). – A.K. Jan 16 '20 at 02:16
  • This could be useful: https://www.rcgroups.com/forums/showpost.php?p=40032474&postcount=4091 – A.K. Jan 16 '20 at 02:21
  • 2
    Also, measure resistance between positive pins with the same voltage to reduce the number of signals (zero ohms). I would expect there is a standby output (highest voltage - 3.3V?) and the remaining ones are pulled up input/output signals. The one with 0.133V is likely a sense input, but not sure which one. – A.K. Jan 16 '20 at 02:43

2 Answers2

2

For the PSU to 'start up' I assume I need to connect a certain combination of the small pins (that can be seen on photo 2) together (similar to how one 'jump starts' an ATX PSU

I don't think that assumption applies to cisco-specific server PSUs like yours. It's not unlikely that some of these pins supply a small supply voltage, large enough to enable a mainboard controller (BSC), which talks some proprietary digital protocol through some of these pins with the power supply. That expectation comes from the fact that some datacenter-grade hardware has access to sensors inside the power supply (to predict failure, for example), so there's some controller that does some data exchange with the mainboard anyway. Whether that's the case for your supply – hard to tell. But since server PSUs never needed to be turned on with a simple switch on the front side of the server case (many don't even have such a button!), it's likely your "short method" wasn't implemented.

Randomly shorting out pins sounds like a viable way of tripping fuses on a power supply, or even damaging it or yourself – that beast can supply 1100 W, and that will definitely make a small wire very, very, very hot very quickly.

By the way, if you've never been in a server room – these things can (and this one will) be very loud. Really, not something I'd try to use for anything that's powering anything in a closed room, or not very loud, anyway (e.g. a CNC drill).

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
  • Thank you for your suggestions! Regarding 'randomly shorting out pins' ... I should explain more clearly: In the post I was referring to in my original question I read that I probably need to connect one or multiple of the small pins to Ground in order to turn on the power supply (such as is the case for this (other) PSU: https://static.rcgroups.net/forums/attachments/2/9/5/9/9/a3955180-237-6800%20PSU%20WIRING.jpg). – Maarten -Monica for president Jan 15 '20 at 19:42
  • If I were to use a very thin wire (with a resistor in it) in order to make the a fore mentioned connection this wire should burn through before the PSU's fuse(s) would blow or other damage could occur I think? Would you still recommend against this? – Maarten -Monica for president Jan 15 '20 at 19:42
  • Is there any way of finding out if my PSU needs a digital signal or simply to connect a few pins to GND in order to turn it on? I should perhaps try asking the manufacturer. – Maarten -Monica for president Jan 15 '20 at 19:42
  • @Maarten-Monicaforpresident no. seriously, the main point of my answer is: connecting pins will not start this PSU. Don't try it. Full stop. – Marcus Müller Jan 15 '20 at 19:43
  • Good suggestion regarding the sound. I might need to add some bigger fans running at lower RPM to get it to run less loud (making sure the cooling is sufficient for the PSU at full load) – Maarten -Monica for president Jan 15 '20 at 19:43
  • ... which makes this more costly / effort than just buying a PSU that you can use right away. It will be hard to find fans that are much quieter than the ones used here: these aren't your average CPU fan. This is *forced* air with a bit of pressure. – Marcus Müller Jan 15 '20 at 19:44
  • My plan for the PSU is to use it to charge a Li-ion battery pack (72V nominal, 20S, 10 parallel). I have checked prices for suitable PSU's/chargers for this purpose and it would cost me approx 100-120euro (new). I already have the server PSU and would need to add a 20 Boost converter to get correct voltage. So I thought it might be worth it to try and get it started... – Maarten -Monica for president Jan 15 '20 at 19:49
  • good point regarding fan pressure, good to know! The noise will not be a huge issue I think, I can live with it. – Maarten -Monica for president Jan 15 '20 at 19:49
  • um, assuming you want to actually get something close to the 1.1 kW that this supply can source into your batteries: you're aware that this is going to be quite a lot of costly copper and silicon in that boost converter? Doesn't sound like a good plan – rather work from a voltage that's too high downwards. That's what your Lithium charge controller does, anyways. For your procjet, 120 € in power supply does sound pretty cheap, so um, go with that. – Marcus Müller Jan 15 '20 at 19:51
  • I was planning on the Boost converter linked below (costs 19euro approx) Assuming the listed specs are accurate it should be able to handle 1500W https://nl.aliexpress.com/item/33022430770.html?spm=a2g0o.productlist.0.0.767d2822FqtuUR&algo_pvid=63294fd0-00c1-4b55-b059-e35c5da18b76&algo_expid=63294fd0-00c1-4b55-b059-e35c5da18b76-7&btsid=2af813bc-cd0a-4945-b111-72f851c9319b&ws_ab_test=searchweb0_0,searchweb201602_3,searchweb201603_53 – Maarten -Monica for president Jan 15 '20 at 19:55
  • 1
    I'm sorry, I most definitely don't trust aliexpress project descriptions, especially when it comes to power supplies. – Marcus Müller Jan 15 '20 at 19:56
  • 1
    Where is the datasheet that any reasonable supplier of a boost converter would offer, which shows how much current you can draw at what temperature from what output voltage given what input voltage. Missing such a datasheet, none of the claims made on that product page mean anything. – Marcus Müller Jan 15 '20 at 19:58
  • Also, does that toroid even **look** like you can push 1500W through that core? – Marcus Müller Jan 15 '20 at 19:58
  • tbh I would not know but judging by your last comment I am assuming the answer is NO. Thank you for the insight. – Maarten -Monica for president Jan 15 '20 at 20:01
  • well, compare the size of that inductor with that in your server supply. The aliexpress device claims to push 1.5× the power through a core that's very much smaller. – Marcus Müller Jan 15 '20 at 20:04
  • @Maarten-Monicaforpresident I would like to assume it’s based off a standard ATX or server one and see if pulling down one of the pins to ground would start it. Do you measure any voltage on any pin? – winny Jan 15 '20 at 21:11
  • @winny I will check if there is voltage on any of the pins relative to GND – Maarten -Monica for president Jan 15 '20 at 21:13
  • @winny I have measured the voltages of the pins and added the results to my original post (3rd and 4th image). Please take a look. Thank you! – Maarten -Monica for president Jan 15 '20 at 22:37
  • @Maarten-Monicaforpresident Good. I would take a 1 kohm resistor and try pulling that 0.133 V to ground. If that does not work, try pulling each 0 V to 3.3V using the same resistor. Chance of damaging it should be very small but not zero in case Marcus is correct and it’s more complicated than your standard power supply. – winny Jan 16 '20 at 06:51
  • @MarcusMüller once again thank you very much for your answer, tips and insights/technical details. I have accepted answer since it described the pinout/startup procedure that worked for my psu but your answer is also greatly appreciated! – Maarten -Monica for president Jan 17 '20 at 21:25
2

I have seen this pinout on other Cisco power supplies - no guarantee that it would agree with this model though so YMMV. That said the values you observed agree with my past experience, so you may be in luck. YMMV, IANAL, you break it you bought it, etc. -

Looking "into" the connector from the back (same as your pictures)

enter image description here

With no AC connected, check in resistance mode between:

  • C4 (output high side remote sense) and P1
  • B4 (output low side remote sense) and P2

You likely will see some measurable resistance between the pins. Most PSU manufacturers put a resistor between sense and output inside the PSU in case the external remote sense lines are left floating.

If you're brave, enable could be B3 (one of the short pins you noticed) - pull this pin to B2 and the output should come alive. If remote sense is needed, you will likely see the output shoot high then go to zero (OVP latch-off) meaning you will need to connect C4 to P1 and B4 to P2.

(FWIW, I2C clock might be B1 and I2C data might be A2)

Adam Lawrence
  • 32,921
  • 3
  • 58
  • 110