Corsair CX550 issue with UPS

[chimera201]

@chimera201 from whatever I read I understood your transfer time theory but I also have doubts regarding your 12ms holdup figure of CX550 as I understand holdup time is directly proportional to system load which in my case is never even reaching 250W so shouldn't it be ~20ms in my case.From the source you quoted my understanding is that assuming even a 20ms holdup time of CX550 & 15ms transfer time of my iball ups it is still an issue because of much higher surge current needed for switchover.Of course with APC ups with 8ms transfer time I am assuming the surge current requirement will be within the limits of UPS.Also the typical transfer time for CX series is given as 4-8ms.
*www.apc.com/salestools/VGUA-9WHCL9/VGUA-9WHCL9_R0_EN.pdf

Check this specs out (also check the fine print):
*www.apc.com/salestools/JGNY-9TW79F/JGNY-9TW79F_R3_EN.pdf

BX600C-IN : 10 ms 'typical'

Also check this:
*www.apc.com/salestools/JGNY-96UETV/JGNY-96UETV_R0_EN.pdf

BE700Y-IND : 5 ms typical, 10 ms maximum
^ This costs twice.

iBall UPS : ? maximum

I'll say this again, no one is going to measure it, they can write whatever they want.

And yeah VGuard may not be better than APC but is certainly better than iBall in quality.

 

[gta5]

1)

Why take 250 w as system load ?

Intel Pentium G620 Sandy Bridge 2.6GHz CPU Review - Page 15 of 17 - Legit Reviews

60-65 watts for whole system with g620 + 75 watts for gpu + 40 watts for hdd ..

That's not more than 175 watts .. In reality it will likely be lower .. Since you said your gpu gets loaded 65-70% iirc , and i am not sure if madvr taxes cpu to 90-100% during video playback

550/175 * 9.9ms = 31ms hold up time in theory.. Take a few ms off for practicality

So hold up time is at least 25ms , in reality likely more because of low load . You can check the example 250 power supply hold up time graph I linked earlier

So hold up time is not the problem that is clear , And UPS having more than 22-23 ms transfer time is possible but less likely if its not faulty or defective . your actual hold up time is likely to be in the 28-30ms if your actual load is less than calculated max load..

What is your inverter, I don't think it is apc right ? And it works at full load, talking about transfer time here

 

[gta5]

2)

Now check this thread

Weird Power Issue [PSU/UPS]

Slight typo in 2nd case , he likely meant 3 minute backup with CM

Same issue ,

In my short on time search, I couldn't find hold up time figures of those PSU

But Seasonic in general has very good hold up times as they use big caps..
S12ii 520 also uses good sized caps

So unless there is a review showing this , I doubt that brand new seasonic there has a significantly lower hold up time than his much old CM.... both the actual hold up time would likely be atleast 20ms for his load o , so transfer time of "both" his original UPS having more than 20ms , you judge the likelihood yourself

Because the 2 UPS, both year old works good with same load with old cooler master


3)

Then look at the last poster with gtx 960 , in the end he bought an intex 1000va and works fine .. Previous one only worked at low loads or with 7770 gpu


4)

s12ii 620 working good with iball 1000va , with a 290watt system load with overclocked Sandy bridge

Looking for a Pc from 60-70k

Form your own conclusion

 

[whitestar_999]

@gta5 That was a good thread but still it does not conclusively prove anything.Please bear with me & see the below points:

1.In that thread one user was successfully using S12II 520W with a 650VA APC ups with higher system load than mine.

2.In that thread GTX960 user might have issues with UPS relay as the person who replied to him said that he had the same issue & his ups used to work with psu sometimes but not at other times.

3.rijinpk CX450 works with an even lower 300W 6 years old v-guard UPS(with new batteries) which is the strongest argument against inrush current at switchover because his system load(100% 3rd gen i7+99% 7770) will definitely create a much higher inrush current demand at switchover than what his 300W ups can handle as per this theory(suggest a theory that can explain this without making extreme assumptions,I am assuming CX450 to be similar as CX550 even with GW/CWT differences)

4.May be 1000VA ups of intex/iball etc are better quality wise(being costlier after all) & have much less transfer time compared to their cheaper 600VA models.

4.Let's assume for a moment that your theory is correct then it would mean that many people using widely recommended psu here & elsewhere should be facing issue but that's not the case(aka 1050Ti/similar is more common than 1kva ups).

 

[gta5]

You cannot conclusively prove anything without proper measurements and some testing , all you can do is ascertain the most likely reason

1) every PSU will have a different inrush current curve this can also be seen in that graph in pdf , plus every same VA ups may not have same tolerance and can vary..

2) that other person "sometimes" working and sometimes not , means his inrush or transfer time is likley very close to the edge.. Maybe the new ups one just had marginal better transfer time or tolerance to cover "everytime"

Relay problem is less likely for 960 user because 7770 gpu was working fine, and his PSU was seasonic 520 so I seriously doubt hold up time was a problem

3) first his PSU is a different design from diff oem , with diff efficiency even and likely have diff inrush curve

second his UPS may have different tolerance for higher inrush than what your ups Can handle

Third It may have much less transfer time than yours , just an example maybe 11-12ms for him and 15-16ms for you so even though his load may be higher , assuming similar PSU design and UPS tolerance just for assumption , because his UPS switched earlier and yours late so the extra load he had may have been mostly offset by your ups switching late and that inrush is well within his UPS handling range ,


It is likely a combination of factors above

4) Or the fact that these are 1000 va UPS , "for sure" capable of handling much higher power draw than their lower counterparts and thus eliminating the issue for most cases...

4) this issue is much more common with 500-600 watt PSU I have read for those facing , generally not with 450 watt psu which most people normally havr, then most people buying 1050ti like cards have APC , and generally those buying higher gpu or 500- 600 watt psus go with 1000va UPS search around you will find some more

This is why i am saying you have good chances that an apc 600va will work mostly i am assuming because of "significantly " higher tolerance and better transfer time , but no "100%"guarantee Because of inrush , as mentioned by both apc and in that PDF ..

5) cant you arrange another apc ups or atleast any other ups to try it out ?

 

[gta5]

try these 2 tests with Cx 550...

""Fully charge ups batteries first"" very imp

1) you said it works fine at low load ,

you have to be as fast as possible in this test

1)run your PC at idle , with your video player , settings , file loaded etc ready ,

then cut the AC power to UPS manually , as soon as it switches to battery power , ""instantly"" play the video file , put similar load and see what happens

and
this is separate test

2) remove those 4 hdds , use only 1 OS HDD and see if that is able to take the load while playing video

 

[whitestar_999]

try these 2 tests with Cx 550...

""Fully charge ups batteries first"" very imp

1) you said it works fine at low load ,

you have to be as fast as possible in this test

1)run your PC at idle , with your video player , settings , file loaded etc ready ,

then cut the AC power to UPS manually , as soon as it switches to battery power , ""instantly"" play the video file , put similar load and see what happens

and
this is separate test

2) remove those 4 hdds , use only 1 OS HDD and see if that is able to take the load while playing video

Somehow forgot to did 1st test but did 2nd test & it worked.With only 1 hdd connected I started playing a video file using ~88% of my 1050Ti with cpu usage ~77% & then manually tuned off the power to UPS & then turned it back on again after few seconds,no shutdown/overload.Does using additional 3 hdd(two of which are 5900rpm) really capable of producing enough load to be the cause of this issue?

P.S. UPS is always connected to mains/inverter as PC is almost 24*7 on so I am guessing batteries are always full(except on certain days & for few hours when there is 4-5 min of power cut because of some house work resulting in removing of mains/inverter power to ups).

 

[gta5]

Do also the first test ," with all 4 hdds connected"

Also can you do 2nd test again with 1 hdd? This time do it 10 times ( won't take much time ) ( simply on and off AC power manually 10 times while your PC is playing video file just give a gap of 30 seconds between each set of switch off and on..

So play video , after removing 3 hdd ,
Then switch AC off , 2 sec gap then AC on

Give a gap of 20-30 seconds

Then again AC off and turn it on again after 2 seconds..

Wait 20-30 seconds

Again switch off AC , 2 sec gap and then on again.

See how many times out of 10 does it work..

Yeah if its always connected with very less power cuts , then batteries would be fully charged

 

[whitestar_999]

Well I did turn off & on AC power to ups 4-5 times with a gap of 10-50 seconds between each attempt(that lasted 2-3 seconds). I will be able to do the 1st test tomorrow only as system is now running with fsp.

P.S. in 1st test how fast is "instantly" because no matter what I don't think I will be able to play the file before 2 seconds after hitting play button in player.

 

[gta5]

Yeah as fast as you can ... 2-3 sec is fine

For the next separate test , you can try adding 1 hdd at a time and test with that 4-5 times to find the max load/limit of hdds that will be supported by your UPS

 

[whitestar_999]

^^Will do but what how much load 3 hdd can offer anyway & what exactly will be proved by this experiment if it turns out to be correct(aka 1 hdd no issue,4 hdd issue).

 

[whitestar_999]

Yeah as fast as you can ... 2-3 sec is fine

For the next separate test , you can try adding 1 hdd at a time and test with that 4-5 times to find the max load/limit of hdds that will be supported by your UPS

I did the 1st test with CX550 & strangely everything is working fine now!I connected all HDDs then turned off power to UPS & immediately start playing video file with no issue(tried this 4-5 times). Then I tried my initial test by switching off power to UPS while playing video file & this time everything worked as it should.I will update on this after 1-2 days of usage but as of now it looks like everything should be fine.

The only thing I can notice about current situation is that mains AC supply is steady with no voltage fluctuations(aka no voltage regulation by inverter).

 

[gta5]

It appears that the inrush drawn is very close to edge of your UPS capacity.. So sometimes it is working and sometimes not ( varying transfer time thus varying inrush in accordance with load , depending upon at which point on sinecurve you are loosing AC- luck based ) and your PSU is fine and hasn't degraded..


Correct me if I am wrong , but your initial posts gave the impression that you were assuming your fsp to be non-apfc.. It is actually an APFC PSU..

 

[whitestar_999]

So assuming this to be the case is it fine to get a 600VA APC ups in future as its transfer time will definitely be much better than my current ups.

You are right,I just never thought that a 2011 psu would have apfc!

 

[sinhead]

It appears that the inrush drawn is very close to edge of your UPS capacity

Bro. I read the APC link you posted on page 1. So as per the formula on the link - " PFC power supply’s rated output power and multiplying it by 1.25" -

Seasonic s2 520 should draw 480*1.25 = 600W. But if the PC "Load" wattage is 275 watt only, then will the PSU still exceed its rated max power because of inrush current?

please read about my config and power calculation on my thread - Config for 3d Modelling + Rendering.

I would like to save money by buying a 650va ups if that is sufficient.