Does an oversized PSU work somewhat as a UPS for very brief power interruptions?

[SlowSpyder]

Pretty good storm outside, power is flickering. My microwave and stove have reset their clocks twice, my Logitech Z5500's have turned off from the power flickering, but my desktop has stayed powered on so far. I have a 1300 watt PSU, my machine idles around 110 watts with the current set up. If I had a smaller PSU, do you think with the brief power interruptions my system would have turned off? Just a random question thought during a storm. 🙂

 

[Mr Evil]

The ability for a PSU to maintain its output when the mains is interrupted is called "hold-up time". The ATX spec says that this should be at least 17ms. I would expect larger PSUs to have longer hold-up times simply because they have larger capacitors in them, which store more energy.

 

[SlowSpyder]

Kinda what I was thinking, the large capacitors may give a brief amount of output that can outlast a small power interruption.

 

[Red Squirrel]

Sorta, but I would not count on it, we're talking miliseconds here. Heck, I've seen situations where a UPS relay does not trip fast enough and PC still ends up rebooting. I had it happen myself recently with one of my servers. Was doing power maintenance and I threw the main breaker, and I did not throw it fast enough and it arced a bit causing a brownout throughout the house. This killed the power to the server despite the UPS picking up the load. Basically it was enough to make it reboot. Still kinda stumped about that one though as it was only that one server that was affected. I've seen laserjet printers do the same too, the bring the voltage down low enough that PC fails, but not low enough for UPS to trip online.

 

[Puffnstuff]

You should also consider that a power supply is subject to ripples in the supply. This is why I run an apc with avr (automatic voltage regulation) which not only supplies battery power when the mains drop out, sag or spike but also cleans the power it sends to the pc. I keep my laser printer off the battery side of my apc since I'm only running one but when I was using two it had its own unit.

 

[BonzaiDuck]

Sorta, but I would not count on it, we're talking miliseconds here. Heck, I've seen situations where a UPS relay does not trip fast enough and PC still ends up rebooting. I had it happen myself recently with one of my servers. Was doing power maintenance and I threw the main breaker, and I did not throw it fast enough and it arced a bit causing a brownout throughout the house. This killed the power to the server despite the UPS picking up the load. Basically it was enough to make it reboot. Still kinda stumped about that one though as it was only that one server that was affected. I've seen laserjet printers do the same too, the bring the voltage down low enough that PC fails, but not low enough for UPS to trip online.

Wouldn't that derive from the voltage range settings in the UPS software as well as the aging of batteries?

Excluding anomalies with my KVM switch which could cause me to mis-diagnose a problem, I'd always noticed that batteries needing replacement were evidenced during an outage when the PSU switched, and the system would either freeze or partially freeze -- meaning that the mouse was frozen.

It sort of throws me that the OP doesn't mention using a UPS. I consider those as insurance "investments" in my equipment. They're extra trouble, batteries need replacement from time to time, and the initial outlay could be more than $100. But I can see what can happen to folks who use a surge-protected power-strip for a desktop computer.

On the other hand, the OP's observation and the response comment seems to be an argument in favor for the purchase of PSUs that far exceed workstation requirements. I've got a 650W UPS (Seasonic) that is more than adequate for the system with 2x GTX-970 SLI (overclocked) and the CPU overclocked. I've got about 100W to spare at the highest load against the spec, and some of these PSUs are capable of sustained loads 100W higher than their rated spec.

 

[JEDIYoda]

No!!

 

[Red Squirrel]

Wouldn't that derive from the voltage range settings in the UPS software as well as the aging of batteries?

Excluding anomalies with my KVM switch which could cause me to mis-diagnose a problem, I'd always noticed that batteries needing replacement were evidenced during an outage when the PSU switched, and the system would either freeze or partially freeze -- meaning that the mouse was frozen.

It sort of throws me that the OP doesn't mention using a UPS. I consider those as insurance "investments" in my equipment. They're extra trouble, batteries need replacement from time to time, and the initial outlay could be more than $100. But I can see what can happen to folks who use a surge-protected power-strip for a desktop computer.

On the other hand, the OP's observation and the response comment seems to be an argument in favor for the purchase of PSUs that far exceed workstation requirements. I've got a 650W UPS (Seasonic) that is more than adequate for the system with 2x GTX-970 SLI (overclocked) and the CPU overclocked. I've got about 100W to spare at the highest load against the spec, and some of these PSUs are capable of sustained loads 100W higher than their rated spec.

Aging batteries would not do that, at very worse, it would simply not work at all, but in my case I still had about an hour of run time and probably more left. (4 hours usually). I think it has to do with bad luck, and an unclean voltage cut. The UPS has a certain threshold and the voltage may dip slightly above that, but still not enough for the PC to handle it, and it might already be on the lower end of the sine wave at that point so you end up with missing an entire cycle and UPS only acts when it sees that a new cycle is not starting. By then it's too late. At least that's my theory. I could be wrong. Dual conversion fixes that, but that's $$$. Nice thing with dual conversion though is you can just keep adding rectifiers, inverters and batteries and grow with time.

That said, my parent's UPS probably IS due for new batteries, they have not been changed since I lived there and I've been away for like 7 years. For them it's mostly just to get them through blips and such though and it normally works fine for that. But when they go to print that's when it fails. (printer is NOT plugged into it, just in same circuit)

 

[2is]

If we are talking about a millisecond or two maybe, but for any practical/real world purposes, absolutely not. I think it'll be a very rare event where you experience a power outage that was too long for a proper sized PSU but short enough for an over sized one.

 

[Torn Mind]

Any proper shutdown will last for more than 10 seconds. An oversized PSU will not run for more than 1 second, and even that is optimistic.

 

[Rifter]

we are talking milliseconds here people, so NO.

 

[Torn Mind]

Given that active PFC has migrated to nearly every quality PSU (the garbage from Logisys and Diablotek, et al are not quality), these PSUs can run even with 100 or even 90v AC. So, they can "survive" some brief brownouts. But I wouldn't take chances like that. The brownout could last too long and poof, there goes your work.

The common stove has no fancy conversions of AC at the point the super thick copper screws into the unit. Lose power and poof, everything turns off. For microwaves, it is similar. There is a big start capacitor but little else in terms caps or even needing to stay on for 17ms.

 

[NeoPTLD]

Sorta, but I would not count on it, we're talking miliseconds here. Heck, I've seen situations where a UPS relay does not trip fast enough and PC still ends up rebooting.

There are several large capacity electrolytic capacitors in a PSU and they tend to gradually dry out. PSU capacitors could be dying. Most plug-in UPS are not truly uninterruptible. There's a transfer switch that switches between utility and inverter power. It takes some milliseconds for inverter to come up to voltage and the switch to transfer. Battery in UPS could be dying too and sagging excessively to be able to support the load. You might recall that lights dim more and starter turn more sluggishly as the battery degrade. In my experience, UPS battery becomes unusable after 2-3 yrs.

Given that active PFC has migrated to nearly every quality PSU (the garbage from Logisys and Diablotek, et al are not quality), these PSUs can run even with 100 or even 90v AC. So, they can "survive" some brief brownouts.

Most PSUs are designed for one SKU for the world. They have to support down to 90v unless they want to exclude Japan where the outlets are 100v.

 

[Red Squirrel]

There are several large capacity electrolytic capacitors in a PSU and they tend to gradually dry out. PSU capacitors could be dying. Most plug-in UPS are not truly uninterruptible. There's a transfer switch that switches between utility and inverter power. It takes some milliseconds for inverter to come up to voltage and the switch to transfer. Battery in UPS could be dying too and sagging excessively to be able to support the load. You might recall that lights dim more and starter turn more sluggishly as the battery degrade. In my experience, UPS battery becomes unusable after 2-3 yrs.

Yeah I suppose caps that are degrading could definitely be an issue though I have seen it on relatively new PSUs. Recently I had it happen with a Supermicro server that is decently new and being a server, has a high quality PSU. Most UPSes switch in <16ms, but that makes it possible to miss an entire cycle. Does not help that most UPSes arn't pure sine wave, nor dual conversion. (always running on inverter)

Here's what it looks like when a system switches to UPS power:



It can very, this is another:



It would have been interesting to see the wave form when I did experience a failure. Come to think of it, I need to check my parents' house, they have the problem all the time with their laser printer kicking on, it browns out the circuit, but the UPS does not quite trip fast enough then the PC locks up. (printer is obviously not plugged in UPS, but in the same circuit as it). I probably should check the batteries in theirs...

As for mine, it has 4 deep cycle marine batteries, so no problems there. They are crappy tire batteries though. 😛

 

[Torn Mind]

Most PSUs are designed for one SKU for the world. They have to support down to 90v unless they want to exclude Japan where the outlets are 100v.

Prior to Active PFC, PSU's had a little switch for 120 or 240 volts to control a voltage doubler circuit.

They always made units that is a compatible around the world. But prior to Active PFC being put into PSUs, the input voltage was manually set and not autoranging.

 

[NeoPTLD]

Is this helpful at all for anyone?
http://www.apcmedia.com/salestools/RMUZ-7DTKRC/RMUZ-7DTKRC_R1_EN.pdf

I used to have my desktop PC on UPS, but I've been without it for a few years as power outages are quite rare and the replacement battery that only lasts about 2 1/2 years are quite expensive.

Prior to Active PFC, PSU's had a little switch for 120 or 240 volts to control a voltage doubler circuit.

They always made units that is a compatible around the world. But prior to Active PFC being put into PSUs, the input voltage was manually set and not autoranging.

I was not talking about auto ranging between low and high ranges. The low range would still have to handle 90 to 140v so it can operate with a 10% margin from line voltage. The lower range being 10% below 100v in Japan and upper range 10% above some places that run 125v.

10% below official voltage is there to account for voltage drop in wiring under load and other appliances in the house starting up. Power supplies are supposed to be able to provide the hold up time under the toughest conditions. Heavily loaded at 90v 50Hz input.

 

[Rubycon]

Absolutely not.
Transfer time should never approach hold time or your reliability is compromised.
Ideally, there should be zero transfer time and the UPS should always provide nearly distortion free sine wave at its output terminals. 😉

 

[dave_the_nerd]

Those kind of power spikes'll fry your PSU if you're not plugged into a decent surge protector or power filter. At which point you've spent enough that you should probably have a UPS anyway.

Or you can save money up front and risk it. Most people do. But it's a risk.