Corsair CX550 issue with UPS

[whitestar_999]

@gta5 according to the test done by @rijinpk1 his CX450 was able to handle inrush current at almost 100% system load & which is more than my system load(i3 vs pentium dual core,7770 vs 1050Ti) on a UPS that is rated 300W so shouldn't this exclude inrush current theory as well because CX450 & 550 are similar design wise.If his system can work with a 600VA UPS with a CX450 then surely my system should work too with a new 600VA UPS which however as per you is risky because of your inrush current theory.

 

[whitestar_999]

hold up time is measured at 80% load afaik. even if you stress your system components to 100%, you are not going to use the half the power the PSU can supply. 1050 ti doesn't even need a pcie power connector, right?

Yes,1050Ti is without any power connector.It draws power from pci-e slot only.

 

[rijinpk1]

what I think is that since the UPS is a pseudo sine wave one, the output at high load without mains may be too distorted to handle by the active PFC PSU cx550 which is actually expecting more or less a sinusoidal wave. non-APFC PSUs will not have this problem.
A good UPS might solve the issue. This is just my assumption, may be far from truth! Try if you can find a UPS just for testing.

 

[gta5]

@gta5 according to the test done by @rijinpk1 his CX450 was able to handle inrush current at almost 100% system load & which is more than my system load(i3 vs pentium dual core,7770 vs 1050Ti) on a UPS that is rated 300W so shouldn't this exclude inrush current theory as well because CX450 & 550 are similar design wise.If his system can work with a 600VA UPS with a CX450 then surely my system should work too with a new 600VA UPS which however as per you is risky because of your inrush current theory.

There are too many possible scenarios here..

1) what if your current UPS in "reality" has a low threshold for inrush current than the theoretical 300-360 watts ? Maybe even less than rijin's UPS ?

not talking about powering a steady load , but instantaneous surge ..

2) what if batteries are not able to power a higher "instant" power draw but fine for lower power draw like with your fsp as a result your UPS is giving overload beep and dropping load completely ? ( to me this is less likely but since Per sir at JG said this point , so i am just mentioning it )

All we know is your UPS is able to handle your system load and inrush current of your FSP , which is lower than 550..

If that is the case then moving to a new good 600va UPS like apc should solve the problem

But here is why I still think there may still be some risk involved..

Your PSU actually uses a bigger primary cap than 450 , (since your psu has to handle 550 watts while providing similar hold up time of 450 )
the advantage is that for your given 150 watt load you get a longer hold up time than 450...

But increasing the primary cap size increases the inrush current .. If they simultaneously have not increased NTC thermistor(inrush current limiter) then your PSU may end up with higher inrush current than 450.. Like I mentioned previously it mainly depends upon interplay of both these , and some other minor factors and differs from PSU to psu.. While a point can be made that your system load is also lower .. But without actual measurements nobody can tell ..

Rm650x of topgear works fine with his 600va UPS ..

And there is a big point that rijin's PSU is GW , with different design than yours..

And also cx 550 already didn't work with your current 600va .. So... There are reasons that it will work is not 100% guranteed

You should contact him , he is using 1050ti with 600va with 550 ..

Suggestions for UPS

And ask him if it is fine if his UPS is handle under load and his variant

I think there are good chances that a good apc 600va will work fine , but there is a risk involved .. So you better make pre arrangements with the dealer just in case it doesn't work.. And return it.. To be on safe side

Better yet , can't you borrow or "rent " another UPS even if its 600va from somebody for few hours ? See what happens and That will help you in forming decision

This is a very common issue if you search around other forums , and most people solved it by moving to a higher VA ups..

Btw this inrush current is not just a theory , this is what APC have said themselves n their faq I linked before.

Also the point regarding damaging cap .. This is also very less likely given how so many people are using apfc PSU from all brands with 99% ups not being pure sinewave for years without any harm and also even on inferior quality primary capacitors

So your cap degrading "so fast" in just a month , not to mention your primary cap is a high quality Japanese one..possible but highly unlikely to happen this fast

 

[whitestar_999]

@gta5 thanks for the detailed reply.I have posted in that thread for feedback about apc ups with 1050Ti & cx550.The thing is there is no option between APC 600VA & 1100 UPS & I don't want to spend 5.5k on 1100VA UPS.Getting a 800VA UPS from cheaper brands(champion or microtek) is another option but then I think if I need to replace it within 2 years then it isn't worth spending.Getting a cheaper 1000/1100VA ups from cheaper brands also doesn't look much good as difference between their price & 1100VA APC ups is ~1.3k which is not much considering APC much higher quality.

 

[chimera201]

Won't find a better explanation than this:
*pqlit.eaton.com/ll_download_bylitcode.asp?doc_id=11666

Requirement Four—
Transfer to backup power faster than PSU “hold-up” time
Power supply units inside IT equipment have an energy storage device, typically a capacitor, that stores enough energy to keep the device running during very brief power interruptions (milliseconds). This is known as “hold-up” time and depends on the internal capacitance of the power supply and the output power rating. At higher output power, the energy is drawn from internal capacitance faster than at lower output power.
According to IT equipment standards set forth by the SSI Forum, minimum hold-up time at fully rated output power is one cycle. At 50 Hz (most of Europe and Asia), this translates into 20 ms. At 60 Hz (the standard in North America), one cycle would be 16.7 ms. Since most IT equipment is designed for the global market, the minimum hold-up time is 20 ms and may be longer at lighter loads. However, the trend is toward shorter hold-up times. Under pressures to reduce PSU size and cost, manufacturers are designing PSUs with smaller capacitors, which leads to shorter hold-up times. This effect is somewhat offset by the prevalence of redundant power supplies, since each power supply would be loaded to less than 50 percent of its capacity.
A related issue with respect to hold-up time is the peak inrush current (Table 1) required to charge up the capacitor that provides the ride-through capability. When first connected to an AC power source (or when powered up on an already connected source), the equipment temporarily draws a large inrush current that can last for 2–10 ms and be as much as 10–60 times the normal operating current.

Similar to the start-up inrush current, there is also a surge current drawn to recharge the capacitors after short interruptions in power (Figure 8). If the power interruption was less than five ms, surge currents will typically last for half a cycle (10 ms) and will be less than 300 percent of nominal current. For interruptions of 10-15 ms, the surge current could be 700 to more than 1000 percent of nominal current, and can last for 1.0–1.5 cycles (20 ms–30 ms).

Implications for UPS design
UPS performance requirement: The UPS must ensure no interruption in its output that lasts longer than the hold-time of the IT equipment’s PSU. This means that the UPS must have an acceptable transfer time for all transitions between different modes of operation—such as from normal operating mode to battery mode and back again, or between high-efficiency mode and double-conversion mode for new energysaving UPSs.
Note that hold-up time will be different for single- or multi-corded servers, because the more PSUs on the IT equipment, the less power load on each PSU, and the longer the available hold-up time. Looking at it another way, single-corded IT equipment will need a UPS with faster transition times, to prevent unplanned shutdowns and reboots.
Transfer time should actually be much faster than the maximum allowable hold-up time, because the longer the PSU goes without power, the larger the surge current it will draw when it receives power again.
In cases where the PSU is without power for more than 5–10 ms, the inrush current required by the PSU could easily exceed the maximum current output capacity of the UPS inverter, forcing the UPS to shut down to protect its own inverter components.

As a result, Eaton recommends using a UPS with a transfer time of less than 5 ms, if the UPS will be loaded to 70 percent or more of its capacity. Rapid transfer time helps prevent a UPS overload condition that could cause a system failure. To prevent problems with large inrush current during initial start-up or restart, most UPSs will feed the IT equipment directly from the utility source at that time.
Considerations for branch circuit breakers. The AC source supplying the PSU must also be able to support inrush current without tripping any breakers. This factor becomes especially critical if several pieces of IT equipment are served from the same circuit. All of this equipment draws inrush current simultaneously after a power outage. It is important to either size the breakers accordingly (using breaker types that have a trip curve compatible with the expected load), or sequence the start-up to reduce system-level inrush current.

At no point does it suggest to get a higher VA UPS.

So for whitestar load even if you take 8 ms figure as worst case at 100% load

Now this is where you are wrong. You assume the transfer time of the iball UPS as advertised.

 

[gta5]

Now you now see how hold up time is highllllllyyy unlikely to be the cause

Thanks for that link , a much more detailed proof for what I have said

Please read that and my replies again

I have said exactly the same thing they are saying

Here

the inrush current required by the PSU
could easily exceed the maximum current output capacity of the UPS inverter, forcing the UPS to shut
down to protect its own inverter components.

It's after this statement they are saying what you have bolded

As a result, Eaton recommends using a UPS with a transfer time of less than 5 ms, if the UPS will be
loaded to 70 percent or more of its capacity. Rapid transfer time helps prevent a UPS overload condition
that could cause a system failure

They are trying to not overload the UPS of a given size , when it will be running at a higher load of"70% or more ".. [ deliberately fixing 1 part of the equation ] the only way to do this is by shortening transfer time, and thus reducing inrush current as a result of not allowing caps to discharge heavily , which is what I have said previously

Or you can have a long transfer time of ups at the cost of higher inrush current ( means higher VA UPS required to avoid overload )

Now that less than 5ms Transfer time of UPS is highly difficult to get .. Plus they are mainly talking about high loads

Look at apc 600va it has transfer time of 4-8ms typically , that less than 5ms figure will be achieved rarely ,maybe 2-3 times out of 10.. and afaik as i said it depends at what point on sinecurve does the AC power is lost.. Purely luck based afaik

So the only realistic solution is to get a higher VA UPS to sustain higher inrush current with normal higher transfer times of general decent ups

The actual inrush current will vary from PSU to PSU , PC load , how much cap discharged , other things etc and cannot be ascertained without proper testing ..
All we know

1)his other psu is drawing less inrush at full load than what his UPS can handle..

2) While at low load his new psu's inrush current is within limits of his UPS , but not at high load

3) While his invertor has enough headroom to support higher inrush of PSU at full load , that is why it works fine

So this is why I was saying that there are good chances that 600 VA apc can work (""maybe"" much better threshold to power inrush, ""much"" shorter transfer time , thus ""less inrush current""

but still there is risk involved , and may not work , so make pre arrangements to return in case it doesn't work..
Cannot be determined without trying or without having proper measurements


Read the apc faq i linked previously again

Computers containing PFC (Power Factor Corrected) power supplies and their use Back-UPS and Smart-UPS SC with Step Approximated Sine waves.

They say the same thing

Now this is where you are wrong. You assume the transfer time of the iball UPS as advertised.

No , i didnt assume at all

Please read again

That 8ms figure I was taking as "hold up time" of PSU as worst case scenario at full load .. And showing how his hold up time at his pc load will easily exceed 25ms in even worst case scenario.. In reality it will be higher since his load will be slightly less than 150 watts at full load + his PSU has a higher hold up time than that.. Although it can vary a little bit from unit to unit as caps aren't exactly precise to the specs everytime

 

[chimera201]

 

[whitestar_999]

@gta5 I believe this is the statement you are overlooking "Similar to the start-up inrush current, there is also a surge current drawn to recharge the capacitors after short interruptions in power (Figure 8). If the power interruption was less than five ms, surge currents will typically last for half a cycle (10 ms) and will be less than 300 percent of nominal current. For interruptions of 10-15 ms, the surge current could be 700 to more than 1000 percent of nominal current, and can last for 1.0–1.5 cycles (20 ms–30 ms)."
The factor here is the time it takes for UPS to transfer power to psu & not psu holdup time.

Assuming my ups transfer time has degraded to say 10ms(after all 4-8ms is official spec) then UPS now needs to provide much higher inrush current for next 20ms which it is unable to but if it had transferred the power to psu within 8ms then it might have been able to provide lesser inrush current(still more than usual) for next 15ms or so & switchover would have been successful.

@chimera201 from what I posted above isn't it also a conclusion that using a higher VA UPS can compensate for degraded transfer time over the course of usage of UPS.

 

[chimera201]

Assuming my ups transfer time has degraded to say 10ms(after all 4-8ms is official spec) then UPS now needs to provide much higher inrush current for next 20ms which it is unable to but if it had transferred the power to psu within 8ms then it might have been able to provide lesser inrush current(still more than usual) for next 15ms or so & switchover would have been successful.

chimera201 from what I posted above isn't it also a conclusion that using a higher VA UPS can compensate for degraded transfer time over the course of usage of UPS.

I don't think your UPS has degraded. It is working fine with your FSP PSU. But I don't trust the advertised specs by iBall. I believe the transfer time for the iball UPS to be higher than 12ms. No one is going to measure it so they can write whatever they want and they won't get sued in India. The FSP did not have any problems since it had a larger holdup.

Quote from that pdf:
"Transfer time should actually be much faster than the maximum allowable hold-up time, because the
longer the PSU goes without power, the larger the surge current it will draw when it receives power again."

Basically UPS's transfer time < PSU's holdup time (which i have been saying from the start)
In your case UPS's transfer time > PSU's holdup time.

Also no need for 5 ms transfer time. They are just saying that to be extra safe (and to make sales of course). 10ms "maximum" should suffice. Quote from the pdf again for the reason:
"However, the trend is toward shorter hold-up times. Under pressures to reduce PSU size and cost,
manufacturers are designing PSUs with smaller capacitors, which leads to shorter hold-up times."

@rakesh_sharma23 was planning on doing some UPS compatibility testing?

 

[gta5]

Again , ^ Please read the PDF again completely .

Basically UPS's transfer time < PSU's holdup time (which i have been saying from the start)
In your case UPS's transfer time > PSU's holdup time.

Please tell what ""realistically"" do you Assume transfer time of his UPS to be.?.

12ms (50% more than advertised or comparable apc -)

14ms ? ,

16ms ? (100% more or double than what advertsied , apc (that's a lot btw )

18ms ?

20ms?

22ms ?

Even if you assume "that high" ,
that is " still fine" ..
His Hold up time is much more >28-30ms, so highllllyyy unlikely

More than that please show some link.. For that big of an assumption

And more than that high in highly unlikely case it would be a "faulty/defective ups" not hold up time problem because that is more than enough

Quote from the pdf again for the reason:

Let me quote the exact same quote you have quoted completely

Since most IT equipment is designed for the
global market, the minimum hold-up time is 20 ms and may be longer at lighter loads.

However, the trend is toward shorter hold-up times. Under pressures to reduce PSU size and cost,
manufacturers are designing PSUs with smaller capacitors, which leads to shorter hold-up times. This
effect is somewhat offset by the prevalence of redundant power supplies, since each power supply would
be loaded to less than 50 percent of its capacity.

Note that hold-up time will be different for single- or multi-corded servers, because the more PSUs on the
IT equipment, the less power load on each PSU, and the longer the available hold-up time.

Now as mentioned previously longer transfer time of UPS will be responsible for a "high inrush current" , and may overload the UPS but that is not because PSU has low hold up time , that is because of higher inrush current of psu than UPS can handle . With same UPS another PSU with similar holdup time will work if the inrush drawn by it is well within capability of ups

Or

Another UPS with same longer transfer time of previous one but with bigger capacity will work fine with that same PSU because it can support higher inrush current within its limits without overloading

The key factor is inrush here which is dependant on transfer time of UPS , PSU design,how much cap discharged , load and other factors

 

[whitestar_999]

@gta @chimera201 Let's just follow the KISS principle These are my assumed scenarios as per my understanding of source quoted above:
1.system load 200W--CX550--power cut--ups time to transfer 10ms--CX550 caps discharging(estimated time to fully discharge 20ms)--at 10ms ups transfer power--CX550 draws much higher inrush current--UPS overloads

2.system load 200W--CX550--power cut--ups time to transfer 8ms--CX550caps discharging(estimated time to fully discharge 20ms)--at 8ms ups transfer power--CX550 draws higher inrush current(but lesser than in situation 1)--UPS works

Correct the above scenarios as per your understanding.

 

[gta5]

@gta5 I believe this is the statement you are overlooking .

where did you felt that ?

Assuming my ups transfer time has degraded to say 10ms(after all 4-8ms is official spec) then UPS now needs to provide much higher inrush current for next 20ms which it is unable to but if it had transferred the power to psu within 8ms then it might have been able to provide lesser inrush current(still more than usual) for next 15ms or so & switchover would have been successful.

Basically for a given load , the longer the transfer time of UPS , the more the Cap discharges ,
the more the cap discharged , the higher it will draw inrush current to fill it up quickly..

1)You can either get UPS with small capacity with much less transfer time so as to not allow PSU to draw a higher inrush and keep it within limits of that small capacity

Or

2) you can get a higher VA UPS with longer transfer time to support higher inrush current drawn because of longer transfer time and avoid overloading

Without proper measurements all you can do is hit and trial..

But you also have to keep in mind the design of PSU , as inrush will vary from PSU to PSU .. You can see that in that graph..

from what I posted above isn't it also a conclusion that using a higher VA UPS can compensate for degraded transfer time over the course of usage of UPS.

What you have written is right and in accordance with what they have said.. Like I just mentioned in previous paragraph

@gta @chimera201 Let's just follow the KISS principle These are my assumed scenarios as per my understanding of source quoted above:
1.system load 200W--CX550--power cut--ups time to transfer 10ms--CX550 caps discharging(estimated time to fully discharge 20ms)--at 10ms ups transfer power--CX550 draws much higher inrush current--UPS overloads

2.system load 200W--CX550--power cut--ups time to transfer 8ms--CX550caps discharging(estimated time to fully discharge 20ms)--at 8ms ups transfer power--CX550 draws higher inrush current(but lesser than in situation 1)--UPS works

Correct the above scenarios as per your understanding.

Yes right, second will draw less inrush current but question is how do you know that even that less inrush current drawn will be still within limits of smaller VA UPS without proper measurements/numbers and trying first ?
Maybe even the less inrush current drawn can still be higher than what your UPS can handle and if thats the case it will overload again even though inrush current was less than in situation 1

 

[whitestar_999]

@gta5 I thought the source quoted talked only about UPS transfer time irrespective of load on psu so I assumed that if switchover happen within a reasonable time then higher inrush current would still be within UPS limits(aka scenario 2) which you probably didn't consider because of chance factor hence overlooked. We know that CX450 can successfully work with a 300W 6 years old UPS with a new battery & a system load higher than mine so I am assuming a CX550 with an even better 600VA/360W UPS should be fine too.

 

[chimera201]

Do you guys seriously believe that iBall has a better transfer time than APC?

Cause: Transfer time > Holdup time
Effect : Surge

Effect comes after cause and effect won't exist without cause. CX550's holdup cannot be less than 12ms (this is a trusted value) so obviously iBall UPS's transfer time is higher than that. And each ms difference will result in an exponential increase in surge.

Also jonnyguru's post for reference:
jonnyGURU Forums - View Single Post - Facing a strange issue with Corsair CX550

Now even before gta5 quotes this part from that post:

Even a cheap Cyberpower or Tripp-Lite standby UPS has a transfer time of 4ms.

Just know that jonnyguru doesn't live in India. The cheapest stuff in western countries is still better than expensive stuff in India. Different models are sold here and there and he lives in a 60Hz grid. The APC 600VA model itself has 10ms typical(not maximum).

 

[gta5]

Show me 1 post where I have said his ups has better transfer time than apc and we will talk more afterwards , and no I was never going to quote jg because I had already posted even before your post , but you never read it or ignore it..


And read post 2344332 again

Corsair CX550 issue with UPS

Anyways short on time , will reply to your last post later whitestar , all the 3 sources are in front of you, form your own conclusion. The last source is much more technical and detailed .. Read again and compare yourself and see if my all posts are saying the same thing as they are saying or not

 

[whitestar_999]

@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
Also in previous posts it is shown that a 6 years old v-guard 300W/600VA ups(with new batteries) is working fine with a system load more than mine(i3 3rd gen at 100% load & 7770 at 98% load) & I don't think a 6 years old v-guard ups should have a much better transfer time than my iball ups.

 

[rijinpk1]

Also in previous posts it is shown that a 6 years old v-guard 300W/600VA ups(with new batteries) is working fine with a system load more than mine(i3 3rd gen at 100% load & 7770 at 98% load) & I don't think a 6 years old v-guard ups should have a much better transfer time than my iball ups.

a little correction. it is an i7 3rd gen
What my thought as I shared in one of my previous comment is that under higher load with an abrupt interruption, the output signal provided by iball UPS may be too distorted for an APFC PSU to handle even if it lasts for few milli seconds.

 

[rijinpk1]

Also comparing v-guard with iball does not make sense to me. v-guard is one of the finest brand which produces really quality products. trust me, you will find at least one of their products in every single house in Kerala. Their quality is much superior than any other local brands available here.
the quality of internals does matter and I think iball is not able to catch up with an APFC circuitry.

 

[whitestar_999]

Yes I know about V-guard(heard many good things about their voltage stabilizers). I wouldn't have put iball quality at same level as V-guard but only did that because it is 6 years old(aka 6 years old vguard quality=1 year old iball quality) but may be you are right & maybe even 1 year old iball quality is less than 6 years old V-guard quality.