[Question] UPS compatibiliy with Active PFC PSU

[Nuus]

I was just about to buy an APC Back-UPS Pro BR1000G until I read all of the reviews for it. While it is a good UPS it is not compatible with new PSU's that use Active PFC due to it producing a simulated sine wave output. That being said I am looking for help with finding a good UPS that producing a true sine wave output.

 

[j_r0dd]

Look at Cyber Power. I'm using the CP1000PFCLCD and couldn't be happier.

 

[Black Ninja]

I was just about to buy an APC Back-UPS Pro BR1000G until I read all of the reviews for it. While it is a good UPS it is not compatible with new PSU's that use Active PFC due to it producing a simulated sine wave output. That being said I am looking for help with finding a good UPS that producing a true sine wave output.

Get APC with pure sine, smart ups. The best for the money is SMC models , after that SMT , and SURTA. First one(SMC1000) you can get around $240. I wouldn't trust other brands , but this is just me.

 

[Ericloewe]

My Back-UPS pro works just fine with my Seasonic G-550.

 

[Tywin]

First, to answer your question: we use APC Smart-UPS of various sizes throughout our lab. They have pure sine output and we've never seen a failure (yet, anyway). Network management cards can be expensive if you need one of those.

Taking a step back though, ask yourself if you actually need pure sine output. Pure sine wave output is better yes, but it does have drawbacks, namely power efficiency and initial cost. The big problem with step-sine output and APFC power supplies is that you end up hammering the input caps. On the plus side for North American users, most (all?) PSUs are designed to work on both 120V and 240V (i.e. 110V and 220V) mains, so the input caps have a lot more margin when running off a 120V step sine. Most modern APFC PSUs should work with step-sine UPSs, but some won't. It may even come down to specific combinations of UPS and PSU. The only way to know for sure is to test your particular combination under a number of different loading scenarios (recommend you do this without your pool drives). Anecdotally, my Back-UPS Pro works just fine with my Antec CP-850 with APFC.

You'll probably also want to read the fine print; if the UPS vendor says Active PFC is not supported, any device liability insurance it may come with could be voided.

 

[Black Ninja]

My Back-UPS pro works just fine with my Seasonic G-550.

I used to use like yours in the past , and switch to Smart APC for 2 reasons: Pure Sin wave , and better hold time. If you don't match your psu with those 2 factors your load can be dropped and the reason of having UPS will be non-existent. I am familiar with this particular model PSU. I guest it comes down to how important the reliability is for you and how much you are willing to spend for it.

Best way is to get APC that don't "switch" when power goes out , and then any PSU will be free from "dropping the load effect".

 

[Ericloewe]

Hold up time of the PSU really isn't the problem on anything worth buying. Even the crummy CX430 had no trouble with that.

 

[Bidule0hm]

Yeah, for example the CS 650 is apparently well under the ATX spec... (11 ms instead of 16). There is some posts on this subject here ;)

 

[Black Ninja]

Yeah, for example the CS 650 is apparently well under the ATX spec... (11 ms instead of 16). There is some posts on this subject here ;)

I know that. That's why I got rid of that power supply. You are absolutely correct it's 11ms , can't even meet ATX Spec of 16 ms.

 

[Black Ninja]

Hold up time of the PSU really isn't the problem on anything worth buying. Even the crummy CX430 had no trouble with that.

It depend what you mean it's not a problem ?:

1.You know what is the hold up time of CX430 and you thinks it's in ATX specifications
OR
2. You don't know , but it's now problem because it won't spoil your lunch ?

 

[Ericloewe]

It depend what you mean it's not a problem ?:

1.You know what is the hold up time of CX430 and you thinks it's in ATX specifications
OR
2. You don't know , but it's now problem because it won't spoil your lunch ?

It means it has worked in practice.

Of course, the real problem is the PSU, not the UPS. If I come across a PSU that won't hold up on a decent UPS (APC Back-UPS Pro or higher/equivalents), I'll get rid of it.

 

[Black Ninja]

It means it has worked in practice.

Of course, the real problem is the PSU, not the UPS. If I come across a PSU that won't hold up on a decent UPS (APC Back-UPS Pro or higher/equivalents), I'll get rid of it.

When you said:" It worked in practice" , some people said "My FreeNAS works with with ZFS and 2GB ram without problems.." . I guess it comes down to what you are trying to accomplish: Build a system against all odds and hope for the best ,or make proper system that has to work because it's important to get it right without taking chances.

For hold up time it's not a just a PSU problem , because if you buy a UPS that doesn't switch , any PSU will work , and hold up time will not be concern anymore , when your UPS it's not doing any switching not risk of load dropping.

 

[Bidule0hm]

For hold up time it's not a just a PSU problem , because if you buy a UPS that doesn't switch , any PSU will work , and hold up time will not be concern anymore , when your UPS it's not doing any switching not risk of load dropping.

You're right, but you shouldn't need to correct a PSU problem by changing another element (it's like if a PSU can't meet the power it's intented for and you reduce the load to make the PSU happy, it shouldn't be that way). Personally I'll never trust a manufacturer who isn't even capable to make PSUs that meet the ATX specs. In this case it's to save a few cents by using smaller caps, it's not that they don't know what they are doing. You can take this PSU, stick in it good brand bigger caps (for the 2 main caps) in place of the crappy Capxon (or whatever other chinese brand they find in the back of the alley this particular day) and suddenly the PSU will meet the hold time spec (plus the caps won't fail in 2 or 3 years) :)

 

[Black Ninja]

You're right, but you shouldn't need to correct a PSU problem by changing another element (it's like if a PSU can't meet the power it's intented for and you reduce the load to make the PSU happy, it shouldn't be that way). Personally I'll never trust a manufacturer who isn't even capable to make PSUs that meet the ATX specs. In this case it's to save a few cents by using smaller caps, it's not that they don't know what they are doing. You can take this PSU, stick in it good brand bigger caps (for the 2 main caps) in place of the crappy Capxon (or whatever other chinese brand they find in the back of the alley this particular day) and suddenly the PSU will meet the hold time spec (plus the caps won't fail in 2 or 3 years) :)

I couldn't agree with you more.:) With ATX power supplies you can find info on hold up time , voltage regulating , etc and you can get good power supply that you know it will meet or exceed ATX spec. Example my favorite Corsair AX760 is great power supply and hold up time is 19ms, but my routers , switches, phone routers, etc devices are also connected to the ups. And these devices have either external small power supplies or build in internal for whom it's almost impossible to find info on hold up times. We do want our switches, routers, phone systems, to be running from UPS too right ? After all the are the backbone of the whole network. So for that case ( even if if don't like the idea much) the only solution is online APC, it will never switch so hold up times on any device will be irrelevant.

If you have any other solution you can suggest it will be appreciated. I am not crazy about getting online ups for 2 reasons: very expensive and efficiency is bad, but I can't think of any other way to make sure switched, routers , etc will hold their load on switching ?!

 

[Nuus]

So this has been very informative to read about and I appreciate all the feedback. The problem is I'm only keeping up with maybe half of what is being said.

My problem is I don't know if my PSU-UPS are compatible. My build I'm working on buying is slotted to have the SeaSonic G 550W Gold PSU and I originally wanted to get the APC Back-Up Pro BR1000G. Upon reading of the possible sine wave problem I switched to the CyberPower CP1000PFCLCD. So now I'm stuck here out of my depth still trying to figure out which UPS would be best with my system and if I need to change my PSU.

 

[Bidule0hm]

In fact I think the routers, switch, etc. PSUs are designed to be put on UPS. And a simple test is to unplug your UPS a few times, if everything is happy each time it's ok ;)

@Nuus From what I know there is no problem with non-sine wave UPSs et active PFC PSUs but it's not confirmed, I haven't read enough documentation to be sure. There some here if you want to read about it ;) I also have a non-sine wave UPS (and it's not even a stepped sine, it's roughly a square wave) with an active PFC PSU and I never had a problem ;)

 

[Ericloewe]

Us 230V people are the only ones likely to be bit in the ass by non-sinewave PSUs.

The basic problem is that stepped approximations need higher voltages than that of a sinewave to get the same RMS value. These higher voltages may be nasty for the main filter capacitors on PSUs, depending on how good the approximation is and on how nasty the capacitors are.

 

[Bidule0hm]

The main caps are generally rated at 200 V each (and they are in series) so 400 V total (I also seen (very rarely) the 2 caps in parallel and rated at 400 V). Unless the UPS produces peaks that are more than 400 V (or the PSU is utterly crappy) it's ok. My very crappy UPS use a pseudo square wave that is 370 V max so unless better UPSs use even crappier waveforms than that, it's ok :)

The current however might be a problem, with the UPS I have peaks at 3.5-3.6 A and only 0.7-0.75 A on mains. In my case I'm not worried because the PSU is far oversized for the load but just because I like when things are the most perfect they can, I thought about adding some low-pass filtering to smooth a bit the waveform and reduce the current peak, but I haven't done that yet.

However this is a bit off-topic and has nothing to do with the PFC :)

 

[Black Ninja]

So this has been very informative to read about and I appreciate all the feedback. The problem is I'm only keeping up with maybe half of what is being said.

My problem is I don't know if my PSU-UPS are compatible. My build I'm working on buying is slotted to have the SeaSonic G 550W Gold PSU and I originally wanted to get the APC Back-Up Pro BR1000G. Upon reading of the possible sine wave problem I switched to the CyberPower CP1000PFCLCD. So now I'm stuck here out of my depth still trying to figure out which UPS would be best with my system and if I need to change my PSU.

Your power supply is very fine. I have one like that, and hold time is 14ms , just 2 ms under ATX spec , but I think is still excellent power supply , it rated very high. Look this: http://www.techpowerup.com/reviews/Seasonic/G550/

As for the power supply , me personally wouldn't by anything but APC UPS. For less money SMC1000 is around $240 for pure sin wave, if it's too much I rather keep BR1000G with stepped wave, I've had one before I switched to pure sin.

 

[Black Ninja]

In fact I think the routers, switch, etc. PSUs are designed to be put on UPS. And a simple test is to unplug your UPS a few times, if everything is happy each time it's ok ;)

@Nuus From what I know there is no problem with non-sine wave UPSs et active PFC PSUs but it's not confirmed, I haven't read enough documentation to be sure. There some here if you want to read about it ;) I also have a non-sine wave UPS (and it's not even a stepped sine, it's roughly a square wave) with an active PFC PSU and I never had a problem ;)

I don't like to test it that way. You don't test poisoned drink by taking few sips and if you don't die , then just bottom up.:) I think it's lame, no wonder why we don't reliability these days, perhaps that's how they testing these days.

Also switching time of UPS varies base on voltage variations , it's not always constant. Could be 4ms next time could be 8ms.