Problem with server UPS system...

[Madscientist267]

The batteries I use for the system are SLIGC115, 4 in series for 24V...

(wow just realized I posted an erroneous reply on another thread, but anyway)...

Every time I put them to a test, I get the following:

Within 10 minutes, the UPS units (APC SmartUPS 1400) are complaining vocally that the batteries are dead... still pumping power almost a half hour after alarm, but why the alarm?

I'm using 2ga wire as interlinks, with a 100A master link between the two 12V 'halves', and 50A glass cart fuses with 10ga wire going up to the units (there are two).

The batteries are rated at 120 minutes @ 75A, which I'm pretty certain I'm not hitting, probably closer to 20A or so total, and they are bitching up a storm. They're golf cart batteries for christ sake.  

More details available upon request, but rest assured, the wiring is not getting hot, and as far as I can tell, there might just be some parasitic loss in the 50A fuse holders? Dunno.

Next step is to wire up a meter across the main output terminals of the batteries and see what is happening there. I have a feeling since these are brand new (about 4 months old?) that the batteries themselves are not the problem, and it's something in the distribution wiring.

Think some uber caps bypassed with several progressively lower value caps, right down to bypass (say 4700, 3300, 2200, 1500, 1000, 470, and finally a few 0.1uf ceramics) might help with this, right at the boards in the UPS units?

I'm at a loss here. I think I'm going to run them this time until the system drops offline to see just how long they take to go truly dead, but I'm convinced (without more measurement) that the wiring is the problem.

Ideas?

Steve

[Madscientist267]

It's now at the hour mark (past where the UPS units started warning that the batteries are dead)...

Does anyone know if these things are 'profiling', thinking that the OEM 18AH batteries are still in the circuit?

Baffling...

Steve

[bob g]

i have 4 of the 1400xl rack mounted ups units, all powered from a set of gc2 agm batteries,
the wire is 4 gage up to the oem input wires which i brought out the front.

i can power all 4 ups units off those batteries without any issues, i suspect you have a voltage drop somewhere
that is showing up as a problem battery as seen by the ups system logic.

you do know you can silence that dreadful beeping?  just wait till it starts to beep and push momentarily the start button
that will shut the beeper off.

btw, i can startup an 8 inch bench grinder (3/4 hp induction cap start motor) off of one ups, although the surge is hard
and the warning beeper goes nuts , it will start it up and then the beeper goes back to the normal periodic beep, that is unless
i have disabled it.

i can't imagine that you should be having any problems unless you have a low voltage problem because of a bad connection, too light of a cable (10 gauge is too light in my opinion for more than a couple feet) or fuse holder connection problem?

bob g

[Madscientist267]

Now closely approaching the 2 hour mark (since alarm activation, still powering 4 computers.)

With the APC units, you can only hit the 'shut the hell up' button when there is enough reserve calculated by the unit to keep it quiet.

Once it goes into panic mode, it wont shut up unless you clip the peizo beeper (which I've been known to do on the 12V, ~650VA versions)

The wire (before the split for the fuses is 2x 10ga, but only for like a foot, then ~2ft at 1x 10ga, through the fuse, a 30A auto-reset breaker, and up into the units.

Steve

[Madscientist267]

Here are the pics of the battery bank just before install... The grey connector at the end hooks into one of identical properties, with about a 9 inch run to the board inside the UPS. There was no modification to the 'internal connection', just re-routed out the back for easy external bank connection.


The Bank's connection details. 50A fuses followed by 30A auto-reset breakers.




The complete bank (with 100A master link visible) sitting next to the 8D (which used to run the 12V version, but has been repurposed as my 12V nominal master RE storage battery).



Seems simple enough, yes? Both UPS units attribute them as 'dead' at about the same time. This may mean nothing since after the fork, they are identical.

Steve

[dnix71]

The alarm on APC UPS units is normal. They weren't made to run like that. Clip the speaker if it bothers you to much. I don't know about the limits of the UPS itself.

It may overheat if run too long. It isn't an Outback inverter made to run 24/7/365.

[Madscientist267]

Yes, but unlike it's smaller cousins, they have a fan on sizeable heatsinks, keeping things cool. They have no problem running for these lengths of time, just the damn alarm going off because they think the bank is 'dead'. It's obviously not... Now working on 2.5 hours... Still holding everything as if there is no actual problem.

I'm convinced it's a voltage drop outside the bank, but until I can pull the servers out, and set up a meter to monitor them, I won't be thoroughly convinced that it's not the batteries either.

I'm going to let them run until the web servers drop, and mark that time, then pull the servers out so I can stick the meter on them and then let them charge all the way back up and run it again to make sure its not a battery problem.


See why it's a pain to add the meter at this point? The batteries are back behind all of this... 


Steve

[Madscientist267]

3 Hours... Running...

Shrug...

Steve

[Madscientist267]

Now 4 hours... seriously doubting battery problems at this point...

Steve

[Madscientist267]

Very closely approaching 5 hours here (within the next 5 minutes), and of course still running (dunno if I mentioned this, but one of the machines the UPS lair is running is the box I'm posting these results on).

Interesting thing has begun happening in the last hour; One of the UPS units has been borderlining out of 'panic' mode... Slipping into '1 LED left' mode instead of '1 flashing LED left' mode.

I'm loosely attributing this to battery heating (they are packed close together enough, basically touching each other, just as in the pics above).

Curious to see if this persists... and will it shut both UPS units up by getting both of them at least out of panic mode.

Steve

[rossw]

Very closely approaching 5 hours here (within the next 5 minutes), and of course still running (dunno if I mentioned this, but one of the machines the UPS lair is running is the box I'm posting these results on).

Interesting thing has begun happening in the last hour; One of the UPS units has been borderlining out of 'panic' mode... Slipping into '1 LED left' mode instead of '1 flashing LED left' mode.

I'm loosely attributing this to battery heating (they are packed close together enough, basically touching each other, just as in the pics above).

Curious to see if this persists... and will it shut both UPS units up by getting both of them at least out of panic mode.

Steve

I don't know if it's universal across the range, but the mid-range APC rack-mount 3KVA UPSs I have all do/did that. When you replace the batteries, you need to tell them about it and make them do a recalibration of the battery state.

I've long since decommissioned my server room that had a bunch of these UPSs, and I really don't miss the dramas of changing batteries every couple of years, I'm sure there is info on the net about it.

You had to connect to the SmartUPS via RS232 (standard models) or via telnet (network enabled models) and issue commands from the interface there.

[Madscientist267]

At 6 hours now, I'm conviced firmly that there is nothing at ALL wrong with the batteries.

I had wondered based on previous experience with these UPS units that they might be profiling the battery discharge curves for the higher capacity batteries,,, I'll find out more when I have the terminal voltage directly monitored via a DMM, and see what happens during discharge.

I'm going to cut it now and let them charge back up 'naturally', then disconnect the servers so I can top off the water, equalize, rest, then bring it all back up. I figure 24H after initial disconnect, everything should normalize, an get it all back to 'normal', ready to run the test all over again. Hopefully after this, the internal gauges should read closer to actual SOC rather than beeping for 5 hours plus.

The beeping on the one UPS is almost gone at this point, with probably 5% or so 'panic' instead of constant like earlier.

If this doesn't work, I will try talking directly to the UPS units and letting them know that there's a much higher cap battery bank installed on them.

Steve

[DamonHD]

What's your live TV channel?  CNN Battery Headline News?  More interesting than the usual 24-hour news I imagine!  B^>

Rgds

Damon

[bob g]

whatever  you do , "don't" cut the leads to the beeper, it will result in a unit that will go into fault mode and not startup!

if you have to do anything, just fill the beepers hole with rtv silicone, while it won't silence the sound it will dramatically reduce the decibels to a more agreeable level.

it also occurs to me that maybe this whole thing will resolve if you allow the internal charger to recharge the battery bank, it will take forever, perhaps a couple days off the main/utility power but i suspect that is when the profile info is calculated for the bank.

the xl series is built to accept add on battery packs, up to about 10 x the original capacity iirc, and i don't think you have to tell the unit
via the rs232 port anything,,, i know i haven't had to but i also recharged the batteries using the internal charger the first time and to maintain the battery bank periodically. otherwise i recharge via the trigen's alternator.

fwiw

bob g

[Madscientist267]

They actually always have... I've never used an external charger on them.

That was part of my thinking however in draining them so far. I have let them go about as long as an hour in the past to see if they were really going to fail, but never have. Obviously after 6 hours, there's plenty of juice. I plugged 'em back in at the 6 hour mark and they once again show full, so I'll do a short test in a few days to see if the gauges drop any slower. They made it all of 10-15 minutes last night before going 'panic' on me.

Now, if the 1400's are similar in behavior to the 3KW units, how would you go about getting it to recognize extra capacity? Or did it just do it on it's own?

There's no 'expansion' port of any kind on these (other than that card module that I've never seen installed in one), I just disconnected them from the board IIRC, took out the little doghouse piece of plastic for the expansion port, and zip tied the cables up to the back of the unit and plugged em in to the external banks.

I used to run 2 x 12V lawn/garden batteries (U1-3) and they didn't do too bad as I recall. They were cheap to replace and had decent run time. As I recall, when the UPS was complaining, it wasn't very long before it actually shut down. But my cable runs were a bit shorter than they are now... ? Like 6" instead of 18 or so, same size wire.

Steve

[Bruce S]

Mad;
 Sorry for not seeing this thread, but I have a pretty good idea of why you're not seeing the readings correctly.
APC units have software profiles. With those 18Ahr batteries the 1400 is seeing it and going off it's profile. The 3000XLS series has the same problem.

IF you have a serial cable that possibly came with these, you can do a couple things that is pretty cool.
These would have also come with a believe it or not floppy disk 1.4M version. and a CD
On the disk there are a couple DOS based programs on the CD there is even more cool stuff.

You'll need to run the program to "tell" your system there is more batteries connected.
Better yet, you could install the management program from the CD and do a load test, this will update the system to match your battery depth.

IF you do not have the disk or CD you can grab the serial number off the APC unit and go to their website www.apcc.com and download the management software for free.

With that software you can even control the alarms and shutoff levels.

OR PM me with a serial number and I'll sort through their website and return the PM with a link.

Sorry for the belated reply 

Cheers
Bruce S