NMC Lithium Coin Cell Replacement?

[BillP]

I can find no recommended schedule for replacement of the on-board battery on the NMC1 and NMC2 units that I am own.

I remember hearing that the battery is only used to keep the clock ticking in the event that the UPS is not supplying power to the NMC.  I am reconditioning one of my UPSs with an NMC2 that seems to "misplace" the correct time when the UPS is off and the battery pack is disconnected from the unit.  The date and time doesn't reset to some initial state but, rather, jumps a day or two into the past.  I'll try replacing the coin cell and see if that corrects the loss of time.

What are the accepted correct indicators, if any, that this battery needs to be serviced?

Thanks much!

[BillP]

Hi Joe,

This coin cell battery is only used to keep the time when power is removed from the NMC. It should last for several years, if not past the design life of the NMC itself. 

The battery is not serviceable but the user and you'll find it is not physically removable. On the older AP9617/18/19 NMC's you could actually remove it but because of safety agency changes in recent years (UL, etc), the coin cell battery is not removable as then it'd be considered a different class of product. I am not sure how they keep it in if is solder or something else but you won't be able to get it out to change. On any generation of the NMCs though, this was never something the user should've had to worry about.

For issues with time, there is another component on the NMC board that may cause this problem and that is also not serviceable. I refer to it as the time crystal and if it is lose or damaged, you can see issues with it always setting the same (wrong) time.

[BillP]

Hi Angela and thanks for the information.  Yes, on closer inspection, the tabs on the CR2032 battery is not a "socket" but, rather, welded tabs making any attempt to desolder and remove not a good idea especially if it is a multilayer board.

Design life is an interesting specification to be sure.  As you are probably aware, some of us tend to try and keep things going well past their design life as in the case of the 18 y.o. SU2200 that I try and keep out of the UPS graveyard.  I'm safe with this one, however, in that its AP9617's battery can be replaced.

Take care!

[BillP]

Hi Joe,

You're welcome. And by design life, I had at least 10 years in mind knowing that some others would push it even further 

[BillP]

Angela, I have another additional bit of information that I'd like to run by you.  I've been working with Sreenath Jana, APC Case #25481930, regarding the NMC2 loosing the correct time.

What I found was that, if the card was pulled out of the UPS for a length of time, the RTC on the card would not update but would, rather, suspend itself.  I noticed this in the event log date and time entries when I turned off the UPS, in this case an SMX3000, at which time the event logger noted the date and time.  I pulled the card out of the UPS and a full four hours later I reinstalled the card and powered the UPS back up.  Rather then the event logging time keeping mechanism incrementing the full 4 hours, it only advanced maybe 21 minutes which was entirely unexpected if the battery backed up RTC was actually keeping time.

NTP resetting the NMC restored proper time and it continues to keep good time but only if installed in the UPS and powered up.

Do you think this is normal operation?  Somehow I don't think so.

[Anonymous user]

On 2/8/2016 3:38 PM, Joe said:

Rather then the event logging time keeping mechanism incrementing the full 4 hours, it only advanced maybe 21 minutes which was entirely unexpected if the battery backed up RTC was actually keeping time.

This definitely isn't normal operation. It sounds to me like the clock oscillator is running severely off frequency, which could certainly happen if the onboard button cell battery is nearly exhausted. You could do a multimeter check on the battery. The clock circuit draws so little current that it ought to run pretty much until the battery is really gone...probably down to under half a volt.

I too was very surprised to find the AP963x series NMC2 didn't have a replaceable clock battery as the earlier AP961x/AP9606 cards did. Unfortunately, I see a lot of these cards being killed by leaking batteries later in their lives. Depleted coin cell batteries have a bad tendency to leak and any such leakage would probably destroy the flash ROM underneath the battery. I've got an idea about replacing depleted batteries on these cards, though it would certainly void any warranty the card would have and must be done entirely at your own risk. If you're interested, let me know and I'll post my idea.

Here are the results of a test on a loose AP9631 that I've got kicking around. Its battery looks a little soft, though certainly still quite capable of keeping good time. (I do have a much better meter, that one was just handy.)

And here's a better look at the negative battery contact (closest to the camera, positive runs across the top of the battery and faces away from the camera).

To the lower right of the battery (rectangular black object with four leads) is the 32.768 kHz oscillator used as a timing reference for the RTC. You can also make the test from the circuit board's underside, where the battery leads go through the board.

[BillP]

Hi William,

The battery on my NMC2 reads +3.10 VDC which is not bad at all.  Unless the energy from the battery is not making it to the clock oscillator circuitry for some reason, there's something else going on with my NMC2.

As the unit is obviously well past the two-year manufacturers warranty and if APC is unable to assist me, I may purchase a battery mount as used on some computer mother boards for the bios backup battery and "kluge" a fix of my own.  I've seen cold solder joints on lots of equipment in the past so maybe a simple "reflow" of the solder joints may be in order as well.

Thanks again for the response!

(P.S.  In case someone is interested, the info on my NMC2 AP9630 is s/n ZA1036017092, H/W Rev 05, Mfr. date 9/1/2010 and is used in an APC SMX3000RMLV2UNC, s/n IS1135000543, Mfr. date 8/25/2011)

[BillP]

Hi Joe,

I am not sure what would cause this. The only components on board I know deal with time are the coin cell battery and the little time crystal thing I mentioned. It is one of those rectangular shaped components on the PCB that could be loose or mis-aligned. I don't have an NMC with me today to double check what the component # is but you can double check those rectangular shaped components. Other than that, I am not sure what is causing it.

[BillP]

Thanks Angela.  William had a pretty good picture of the clock oscillator circuit (black, four legged and rectangular).  I have to think that, if the oscillator works when powered by the UPS, it should work when powered by the battery.  I'll check closely the battery mounting and see if there is a "blocking diode" that disconnects the battery when there's UPS power available.  Something my have opened up for whatever reason.

Joe

[Anonymous user]

I can't say for certain as there is no public documentation, but I'm thinking that the AP963x SoC has both a hardware and software RTC. A transistor, possibly built into the SoC die itself, would likely switch on the battery assuming that it is ever switched off. The current draw is so low that it wouldn't have to be.

Since your battery seems to be more than strong enough (my AP9631 is younger than yours, having been assembled in the 30th week of 2011, S/N 5A1130T01248, and its battery is definitely not that strong) something may be wrong with the oscillator crystal.

[BillP · Accepted solution]

Hi William,

Maybe the fact that the battery in my unit is so high is indicative of it not ever being used or otherwise disconnected from the active circuitry.  Yes, without documentation in the public domain much of it is guesswork.  I tried the simple things of re-flowing the solder connections on the battery and to the crystal oscillator package but no joy.  I attempted to check the actual voltage on the crystal oscillator package from both sides of the battery to all four leads of the crystal unit and was able to measure only about 0.5VDC to two of the leads (I believe the ones closest to the battery).  My common sense says that somewhere near three volts should have made to to at least one of the legs of the crystal and I will compare readings to a known good NMC2.

The good thing is that it works fine when the UPS is connected to the line or the UPS battery is connected.  In retrospect, I can't imagine a time where I should really care about what the NMC2 is doing outside of the UPS when there is no power being applied.  OCD probably!

Thanks for the thoughts!