UPS-protected PC spontaneously restarts

[BillP]

My desktop PC has started to blink. At irregular intervals, up to twice a day now, it shuts down for a fraction of a second and reboots. It's not the failure of house current; the system is on a BE750G UPS and I've tested that the UPS works, and the house lights don't flicker. I can reproduce the problem by running the PowerChute UPS self-test; about a third of the way through, the self-test does something - obviously not intended to produce this result - that causes the shutdown and restart. I'm speculating that a sensor of voltage levels in the PC has gone hypersensitive and shuts the system down as it would for overheating, but I'm not sure it would also reboot. The Windows 7 event log shows nothing except that it was shut down abnormally, no surprise there if the hardware is at fault. Anyone else run into this problem, and how do I fix it? It would be a clue if I knew what steps the self-test goes through that triggers the restart.

[BillP]

1.) Besides a Smart UPS (if the power supply just cannot handle the step approximated sinewave or transfer time), you could try a BR1000G which is a quasi sinewave. We have also suggested this to folks to better deal with Power Factor Corrected power supplies. A lower end Smart UPS with a pure sinewave which while pricier, should also work is something like an SMC1000. I don't think it comes in any smaller VA though, only larger. I guess I suggest this because based on the behavior, the issue could be the UPS transfer time, step approximated sinewave, or both.

2.) I would not suggest this cause you'll have the same issue because of the step approximated sine wave that the BE750G outputs. So when the BE750G goes to battery, the BK650MC would still turn off or go to battery itself.

[BillP]

Thanks!  I'll take a look at the units you suggest in #1, and unplug the BE750G I was charging up.

About the "larger UPS than necessary" issue; given the same operating load on each, does a larger-capacity UPS draw any (or significantly) more power than its smaller cousin?

And how about when there is essentially no load? I find I need to leave the UPS on 24/7, even at night when the PC is off, or else the CMOS goes haywire and the next morning the system tries to boot from a CD drive.

[BillP]

A larger capacity UPS with the same load may give you a few more minutes runtime and maybe draw 1-2W extra for the battery charger. I would say there is no significant difference though, only minor. The BR1000G also offers an LCD display and some automatic voltage regulation functionality.

However, if I use our UPS efficiency calculator, you can view "losses" at 0% load - http://www.apc.com/tool/?tt=5. See comparison below. I'd probably expect 10-15W from the battery charger, assuming you have a power outage and the UPS battery is completely discharged. Otherwise, expect the single digit figures below.

Have you checked out the power settings in your BIOS/CMOS to see if it is a setting causing the weird behavior?

[BillP]

Thanks for following up and sharing your results! I am glad it works well.

[BillP]

I want to update this thread with a new development and ask about a couple of possible solutions.

The story so far, as I remember it up to a couple months ago. I have a Dell desktop XPS435T running off an APC BK650MC since 2009. Worked fine, no problems. About 15 months ago I saw the battery on the UPS was getting old, and I decided to replace the unit with a newer BE750G. Bought it, installed it. Disaster. About every day or so the PC powered off entirely for a fraction of a second and rebooted. I found I could reproduce the problem with the PowerChute(?) software, and that the cause was the BE750G "self-test" in which it transfers power briefly to the battery. Apparently it's not brief enough for the Dell. Long hassle over whether the Dell's power supply lacked the "capacitance" to handle the flicker or if the new UPS was defective.

That was never resolved; I had no way of testing conclusively, at least no way that didn't involve crashing my system over and over. But I had a workaround: I just got a new battery for the old BK650MC and put the new UPS in the basement.

That worked fine through several power outages until about a month ago. Since then, even a brief power loss at the wall socket causes the BK650MC to beep and the Dell to power off and on as though the UPS did not exist. I would like to resolve this, and I have two questions.

1. Since this is an APC forum, let's assume the problem is with the Dell. 🙂  It seems to need genuinely uninterrupted power; is there a UPS that meets this requirement?  (I don't understand all the technology, but I have a device that shows it's possible. My laptop runs off its battery, and the wall current merely keeps the battery charged. I can pull the plug on it with no problem.)

2. Second question: this is a very ugly solution and I'm only looking at it as a temporary fix, but...  Supposing I plugged the BE750G into the wall, and the BK650MC into the BE750G? Then when the new unit does its infamous self-test, the older one should have the capacitance itself to ignore the flicker and keep going, right?  I could just try the experiment except for my reluctance to crash the Dell, so I thought I'd ask here first.

[BillP]

I don't see anything in the BIOS settings that suggest a value to modify. I can't think what sort of option that would be, to turn on such disastrous consequences. Also the PC was happy with the BK650MC until recently and I haven't fiddled with the settings.

My goal is to purchase a UPS unit that even with self-testing will deliver an uninterrupted power stream, so I can put the issue aside and move on to more important matters like what to have for lunch, etc.  The SMC1000 looks like a good candidate, and I added it to my "shopping cart". I know in the past there has been a Trade-UPS option to return an existing unit like the BE750G for some sort of credit, but I don't see that on the order form; can you tell me where to find it?

[BillP]

I know in the BIOS there is something power related that refers to last AC [power] state typically which refers to any power interruption and what the PC will do after it (come back on or stay off). Also, there is wake on LAN which may play a part.

Anyway, you can access Trade UPS from here -> Recycling Options - recycle

[BillP]

Thanks for the Trade UPS citation and your help in general; I expect I will turn in the BE750G for a (an?) SMC1000 and see what happens.

About the BIOS setting, I don't recall that particular option; in any case I think the fundamental problem is not that the PC reboots itself during a cutover to battery (as opposed to staying off), but that it crashes in the first place. At the moment, my guess is there's a finicky power supply in the 4-year-old Dell; but it works fine otherwise and I'm reluctant to replace it on a guess.

[BillP]

I'm happy to report   that the new SMC1000 I just installed can go to battery power again and again without crashing the Dell PC.

I haven't figured out how to configure all the options yet, but I'll open a new topic for that if necessary.

[BillP]

yes, so to me, it seems like the UPS itself is operating correctly and possibly this is not the correct unit for use with your PC. did it ever work with the PC?

[BillP]

can you confirm/deny all of the strange behavior you are seeing is when the UPS is running off of its internal battery?

I think so, though I wish I could be certain. The evidence is:

- PC briefly crashes and reboots - always - some seconds into the PowerChute self-test option

- at UPS startup (when I press the green on/off button) a lightbulb plugged into a battery backup socket displays variable brightness. I just ran a startup test and this happened:
The green on/off LED blinked for about 18 seconds. During the first 9 seconds (roughly) the light was bright; then it dimmed for the next 9 seconds; then the LED went solidly on and the bulb brightened again.
I didn't have a stopwatch the other times I tried this, but I think it was about the same.

[BillP]

the self test is designed to transfer the UPS to battery operation and compare the battery voltage drop while on battery with the attached load to a table it has to see if the voltage goes out of range over time to look for a potentially faulty battery. the the battery voltage drop goes out of tolerance during the self test, then you'd have a bad battery indication.

oh, i forgot to add that the self test itself by default occurs at UPS start up and every 14 days by default. i am not sure if you can change it on this model like some other models.

but i think you are saying you are able to replicate it if you force a self test and it does not necessarily always occur during the scheduled self test.

[BillP]

Thanks! The BE750G UPS is a new unit which supposedly passed all its factory tests, but I notice when I just plug a light into it, the brightness visibly varies at startup. It could be that when the PC was plugged in, it was booting after the startup self-test finished. Before I ship the UPS back as defective, I wonder if there is a way to verify if it's the culprit. Since the PowerChute self-test crashes the PC, I can't use the results from that.

[BillP]

this UPS outputs a step approximated sine wave. when its "online" it is just passing through utility power.

can you confirm/deny all of the strange behavior you are seeing is when the UPS is running off of its internal battery?

[BillP]

i think what you are seeing is just typical behavior with a step approximation to a sinewave.

there is some info on that here: http://forums.isxusergroups.com/thread.jspa;jsessionid=1ED58EF7009B43461A1C6C862AE5C2E2.node0?messag...

this is also why you see the light flickering when the UPS is outputting a step approximated sinewave.

[BillP]

That might explain the light bulb's varying brightness. But it still leaves me looking for an explanation of why the PowerChute self-test crashes the PC.
Basically the UPS seems to be causing the exact symptom it is supposed to protect against.
So either it is faulty, or the PC is faulty in being hypersensitive to an "acceptable" variation so far as I can tell.
I can't think of any other explanations, and I need a way to isolate the problem to one or the other unit. Hmmm...

[BillP]

here is what i'd recommend. it could be a few different things causing issues for the PC - perhaps the transfer time when the UPS goes from online to on battery with a combination of it hating the step approximated sinewave.

i'd suggest doing a pull plug test. if you can afford to do it with the computer, i would have it running on the UPS then remove the plug from the wall on the UPS and watch the UPS transfer to battery and see if you see the same behavior. I'd also then repeat this test with something else, like the light bulb or a clock radio to see if you see any issues with that. this would be a little different than a self test and i'd be curious of the results since it is demonstrating a power failure. we could also see if it is indeed only occurring when the self test happens.

[BillP]

i'd suggest doing a pull plug test

I did that earlier with the light bulb; I should have mentioned. I'm not comfortable doing it with the PC attached; the last time the self-test crashed it, that scrambled my RAID array and it took 5 hours to rebuild. I just did the pull-plug test again with the light bulb, and got the same results as before; the bulb dims when on battery, and brightens when back on AC.

[BillP]

did it ever work with the PC?

No; but there is a bit more to the story. The UPS is a new one, a BE750G. The older BK650MC I had before worked fine through outages and the like, but a couple weeks ago it failed. Because of its age, I figured the battery was gone (which still is likely), but instead of replacing the battery I thought I'd upgrade. The new "upgrade" is the one that causes the blinks. I've ordered a new battery for the old unit. But I also noticed an issue with the new one; you know how the buyer has to connect up one battery wire on delivery? When I did that I had a @#$%^& time with it, and on closer inspection I see I didn't push the connector all the way on. I've since done that, and now I need to set a time when I can risk a test of it. BTW, thanks again for your help with this!

[BillP]

did it ever work with the PC?

No; but there is a bit more to the story. The UPS is a new one, a BE750G. The older BK650MC I had before worked fine through outages and the like, but a couple weeks ago it failed. Because of its age, I figured the battery was gone (which still is likely), but instead of replacing the battery I thought I'd upgrade. The new "upgrade" is the one that causes the blinks. I've ordered a new battery for the old unit. But I also noticed an issue with the new one; you know how the buyer has to connect up one battery wire on delivery? When I did that I had a @#$%^& time with it, and on closer inspection I see I didn't push the connector all the way on. I've since done that, and now I need to set a time when I can risk a test of it. BTW, thanks again for your help with this!

[BillP]

no problem. and thanks for the background. BK650MC is a completely different platform and may be related to the root cause of the issue - not sure.

if the lamp or light stayed on during previous steps then i don't know if the battery connector is the issue. worth a check though of course. and yes, i know those battery connectors are difficult to connect with lack of space oh and the length of the cable 😞

[BillP]

did it ever work with the PC?

No; but there is a bit more to the story. The UPS is a new one, a BE750G. The older BK650MC I had before worked fine through outages and the like, but a couple weeks ago it failed. Because of its age, I figured the battery was gone (which still is likely), but instead of replacing the battery I thought I'd upgrade. The new "upgrade" is the one that causes the blinks. I've ordered a new battery for the old unit. But I also noticed an issue with the new one; you know how the buyer has to connect up one battery wire on delivery? When I did that I had a @#$%^& time with it, and on closer inspection I see I didn't push the connector all the way on. I've since done that, and now I need to set a time when I can risk a test of it. BTW, thanks again for your help with this!

[BillP]

did it ever work with the PC?

No; but there is a bit more to the story. The UPS is a new one, a BE750G. The older BK650MC I had before worked fine through outages and the like, but a couple weeks ago it failed. Because of its age, I figured the battery was gone (which still is likely), but instead of replacing the battery I thought I'd upgrade. The new "upgrade" is the one that causes the blinks. I've ordered a new battery for the old unit. But I also noticed an issue with the new one; you know how the buyer has to connect up one battery wire on delivery? When I did that I had a @#$%^& time with it, and on closer inspection I see I didn't push the connector all the way on. I've since done that, and now I need to set a time when I can risk a test of it. BTW, thanks again for your help with this!

[BillP]

Would there be any reason to think that another type of UPS like the BR700G would have a shorter cutover interval when switching to battery power? (I still need to test the present unit now that I reseated the cable, but in case that doesn't work I'm looking at alternatives.)

[BillP]

i was asking around. do you know what the sensitivity is set to on the UPS if you check the powerchute software? make sure it is set to high. if its on medium or low, that may increase the transfer time to battery by a few milliseconds. i think that is the root cause here and it is worth a try.

but, all Back UPS have a similar transfer time so i don't know if another Back UPS would be the best depending on what we find with the sensitivity. another thing thing that was suggested to me was the age of the computer you have - maybe the capacitor that would normally hide this type of issue in the computer power supply was drying up so then it cant hold the computer up during the 4-6ms transfer time. i know your computer is running Windows 7 so that did not sound likely to me really..

if we found that this is a higher class power supply or something that just requires a better transfer time, our next suggestion would be a Smart UPS but those are a bit more $$.

[BillP]

The sensitivity is the default level; I haven't altered it.
But after some more experimentation, I have to conclude that this BE750G unit is just defective. 😞
I reseated the wire to the battery and verified that it was fully into the connector; then two days ago I reinstalled the unit to my PC. It worked well for two days, then a couple hours ago it dropped all power to the PC briefly. I brought the system back up and ran a self-test. About a third of the way through as measured by the progress bar, the green LED on the on/off button began to flash and the UPS dropped power for a couple seconds as near as I could tell, way more than milliseconds. So my best guess is that the UPS takes that amount of time to switch over. I didn't try unplugging the UPS; basically the unit is causing the very problem I bought it to avoid, and I don't want to turn into a test lab. Every time this occurs it damages my RAID array and I have to rebuild it, a 5-hour job.
I purchased a replacement battery for the BK650MC, and I'm using it now. I don't need two UPS devices - especially one that dies periodically - so I suppose I will try to return it for a refund. Before I ship it back, I'd be happy to perform any tests that don't involve crashing my PC.

[BillP]

im sorry to hear that this has been a problem and i completely understand. my thought is that this issue is specific to the PC because basically what you've said so far is the lamp works when the UPS switches over and it seems to be an issue with the transfer time/and or step approximated sine wave output.

i can't blame you for going back to the older UPS since it works. beyond that, i suggest returning it because i just don't think there is anyway to get it to work with your PC power supply unfortunately. we have other models that may work better when you are ready to use something else but im glad you have the BK650MC for now.

[BillP]

I'm doubting that the PC is the weak link here, though I haven't found a way to isolate the fault one way or the other. This morning the BK650MC gave a loud beep and - nothing else happened; the PC stayed on through the brief outage. (Sigh of relief.)

Since an incandescent lamp is probably a poor test, I used a night light with (I think) an LED, and it perceptibly blinked off and back on when I unplugged the BE750G.

So all I can tell for sure is that the new UPS and the 3 year old Dell XPS desktop don't work in harmony. My guess is a manufacturing defect in the BE750G; which has features I like, but it has to, you know, work. If there are other models I should consider, I'd be happy to take a look for the future; but for the moment the new battery in the old UPS seems to be doing the job. I assume I contact tech support for a return authorization for the BE750G? Thanks again for all your help.

[BillP]

Oops; duplicate message. I got repeated Server Error notices when I tried to post this earlier.

Message was edited by: ronks

[BillP]

ill see if I can delete those extra messages.

i don't think its the fault of either the UPS or the PC - I think it is a compatibility issue based on the information provided.

did you purchase the UPS via APC directly? what i mean is tech support is only going to be able to do RMAs - meaning if they find there is a fault with the unit, then they will provide you with a new unit and have you ship the old one back. sales will only take back the unit if you purchased it from us directly and it is within 30 days. even with that, i think there is a restocking fee associated with it but i haven't done it in years so i am not 100% sure on that specific procedure in order for you to get $$ back.

[BillP]

did you purchase the UPS via APC directly?

Yes, when the battery failed on the older unit I decided to upgrade to a newer model from APC; then when it caused such problems, I bought a battery for the older one, also from APC. Which is what I'm running on now.

Not that it matters much, but I'm unclear on how a compatibility issue could be involved; I would have thought there was at least a de facto standard for power continuity, and that if BE750G's were incompatible with Dell XPS desktop units it would be widely known by now. I'm no expert in this, though.

[BillP]

i also don't support these units all day along but we have surely seen issues with particular power supplies.

here is just one example:

http://en.community.dell.com/support-forums/desktop/f/3514/p/19262542/19443657.aspx

[BillP]

Hmm, we're trying to eliminate possible causes and this Optiplex post introduces a new one; it's like swimming upstream. But there does seem to be a finite set of suspects, and maybe a way to cross some off the list. I'm seeing these candidates:

1. This BE750G unit is not switching over to the battery quickly enough (I think you said that should be 4-6 milliseconds).
2. All BE750G units have the problem; that seems unlikely.
3. The power supply in this 34-month-old PC lacks the ability (the capacitance) to bridge a 4-6 msec gap. I've run bootable hardware diagnostics on the system and it passed, but I'm not sure if this specific feature is tested.
4. This BE750G unit, or all such units, don't put out a proper sine wave when on battery power (and the BK650MC does, because it never caused this sort of outage).

You might be able to eliminate #4; do the two UPS models produce different types of output when running on battery?

And how long does the self-test, when it switches to battery power, stay there? The PC goes dark and reboots within a space of 2 seconds at most, I'd say; if the UPS remains on battery during the reboot, even for a few seconds, then I can't see how it's a square-wave problem.

As to #1 and 2, I have the QA test results for this unit but I don't see a specific test for speed of transfer to battery. The descriptions are cryptic and they all say PASSED; none of them says "4 msec" or anything like that. Do you know if that's tested?

[BillP]

hi again,

correction: this particular Back UPS is uses a "standby" topology versus some other Back UPS products that use "line interactive" which is closer to 4-6ms. for this particular UPS (BE750G) I would say it could be as good as 8-12ms. i double checked with our Back UPS product expert to verify the transfer time point.

i also believe that the power supply in your desktop is likely to be power factor corrected which may not work well with this UPS either (see FA158939 @ http://www.apc.com/site/support/index.cfm/faq/ for more info on that)


BE650MC and BE750G are completely different. BE750G is a high frequency design. BK650MC also did not have a self test function and would only go to battery during an outage. i am not sure how many outages you have but perhaps the issue did not show as much due to the lack of an automatic self test. i also don't know if you specifically have tested this recently (simulating a power failure on the BK650MC)

ultimately, if you are going to get another UPS in the future after the BK650MC, i would definitely research Dell XPS power supply compatibility to see which UPS models other users are using since my gut feeling is that there are 1-2 issues here we've discussed in particular with the BE750G that are specific to your model of power supply and boil down to compatibility due to the design and type of the UPS and the specs and requirements for the power supply.

[BillP]

Thanks! I am asking over in a Dell forum about power requirements for the XPS system (capacitance, waveform). The sheet I have rates the power supply at 475 watts; no info on power factors. When I used Powerchute on the BE750G, it showed a draw of about 133 watts in normal operation.

To bring some closure to this, what I think I will do is keep the BK650MC as a reserve and replace the BE750G with a BR700G or BR1000G. I see the BR's are line-interactive rather than standby, which may resolve the problem. Every unit seems to be "stepped approximation to a sine wave", but so is the BK650MC which works. And after Windows 8 comes out and stabilizes, I may replace the Dell which has problems of its own, being on its fourth motherboard. And if that doesn't work, I will probably move to another state and change my name... 🙂

[voidstar_apc]

The Dell XPS 9000 in particular has been notorious for not working with budget UPSes.

You may want to consider a new power supply (like [this guy over here|http://forums.isxusergroups.com/thread.jspa?messageID=16801䆡]) rather than a new UPS because PSUs are usually cheaper and you'll be able to use whatever UPS you want with it.

[BillP]

LOL ok. well please let us know if we can help in any other way.

[Anonymous user]

I have an old Dell XPS 600 with 650 watt power supply that works fine with APC BE 725. But this machine dates to before the 2007 changes that make these reasonably priced APC units tend to not work with Dell XPS machines.

I just purchased a Dell XPS 8500 special edition from Dell Small and Medium Business sales person. She insisted the very unit you have will work with XPS 8500. It has a 460 watt power supply and I have read that sometimes these work and sometimes they don't. I purchased one from Dell but I have doubts as to whether it will work. I don't want a Smart UPS because of price and they take up way too much room that I don't have.

[BillP · Accepted solution]

Just a thought, but I think APC's UPS Pro series is less expensive than their Smart UPS series; though more $$$ than the base UPS products. If the gap in switching between wall current and battery is a concern, the Pro series looks better per the specs. My older BK650MC has worked well through a couple of brief outages and the shiny new BE750G is in the basement; if it doesn't play nice with the next PC I get, I'll probably move up to a BR700G or the like.