PC shuts down during UPS mains->battery transfer

[nap]

Hello!

I have a Dell Poweredge T20 attached to a new CyberPower 900PFCLCD true-sine UPS that (only sometimes) unexpectedly resets when I unplug the UPS to simulate a blackout under heavy PC load. Having a PC reset from a power outage while plugged into a UPS sort of defeats the purpose.

The PSU is a Liteon L290EM / (Huntkey HU290EM-01) PSU.
The computer at its maximum operating load consumes 250W at the wall which should be around 208W DC or ~75% of nominal PSU load and 40% of nominal UPS load.

Reading some reviews on this topic I suspect that the power supplies hold-up time at load is shorter than the transfer time of the UPS. I suppose I could replace the power supply but it has fairly high efficiency at low loads (77,5%@29W DC which is exactly what the Dell uses in 24/7 idle) which is a quality I haven't found in many other units that would fit in its place (ATX w/ max length 140mm).

The PSU has a 170μF/400V primary capacitor, is possible to tell whether this is sufficient for the 8-10ms transfer time that my UPS takes according to reviews? If this is well dimensioned, perhaps the UPS is faulty?

[stj]

Re: PC shuts down during UPS mains->battery transfer

170uf sounds like a joke,
even old 300w stuff from the 32bit days used 330-470uf.

you could also check / uprate the caps on the output-side.

[nap]

Re: PC shuts down during UPS mains->battery transfer

Ok thanks, I'll try to take pictures of the whole PCB over the weekend.
Do you think replacing the primary cap with something like this:

http://de.rs-online.com/web/p/products/8442871/

should solve the problem? I'd have it replaced by my local electronics store because I'm worried about soldering the primary side. Maybe I should just buy a new PSU...

[stj]

Re: PC shuts down during UPS mains->battery transfer

post the details of the 170uf cap.
make, series, height, diameter.

[momaka]

Re: PC shuts down during UPS mains->battery transfer

170uf sounds like a joke,
even old 300w stuff from the 32bit days used 330-470uf.

Well, you are comparing apples to oranges. PSUs back in the "32-bit" days often did not have Active PFC. Thus, they often had two 200V capacitors on the primary side connected in series (due to voltage doubler circuit needed for 120V countries and/or half-bridge topology designs). In a series connection, the total capacitance of two capacitors becomes half. So two 200 V, 330 uF caps in series would be equivalent to a single 400 V, 165 uF cap.

With that said, two 330 uF caps in series (or one 165 uF, 400V cap) is about right as the *minimum* input capacity for a 250 Watt power supply.

Thus, it DOES make sense to upgrade the primary cap of your PSU to something bigger. 220 uF should yield some improvement, but 270 uF should be much better. And if you go with a 330 uF cap (that is, one rated for 400V still), the hold-up time will be excellent.

HOWEVER, there is one caveat: if the PSU has an AC detection circuit, then it will shut down as soon as AC power is removed/disconnected from the PSU. In such case, the size of the input capacitor will not matter at all. I found out about this little thing while working on a ThermalTake Smart M Series M850W PSU build by CWT. This same PSU/platform is also used for the Corsair CX750M, so it does the same exact thing. Thus, it is possible that many other CWT PSUs have this circuit, and in fact may not be unique to CWT.

So in short, if your PSU has that circuit, then the input cap size will not matter. But I suppose it is still worth trying if you can get a primary cap for a low enough price and either do it yourself or find someone to do it fairly cheap.