Advice or Schematics for a faulty APC Back-UPS XS 1300LCD?

[skyhawk]

So I picked this thing out of ewaste years ago, and a super-quick check at that time suggested it was working fine.

I finally got around to putting some (brand new and fully charged) batteries in it, and now I discover that it's not charging the batteries. It will run on batteries just fine, but when in online mode the terminal voltage remains about 0.1v below the floating (disconnected) terminal voltage of the battery pack.



There's a YouTube video discussing the same fault (failure to charge) and showing disassembly. I've followed this guide to get the unit apart and checked the same things shown there.

There's nothing obviously blown-out on the board, other than a visibly bulged but not leaking or ruptured 22uF 400v capacitor.  I've checked all the diodes I can find - all have voltage drop in the expected direction, and there's no dead shorts. All the fuses (The little round one and the three automotive-style) are fine. The relays all clicky-click when tapped with 12v or 24v as appropriate.

Someone's written in green sharpie "P" on the black battery connector, which makes me guess maybe someone connected batteries backwards? (More likely probably P for Pooched, though)

I'm not equipped for surface-mount rework, so if it's not a "low-hanging fruit" issue I'm not going to be able to repair it, even if I can detect it.

So I have a few questions:

a) Can anyone share circuit schematics for this board? It's marked 640-1019_REV04
b) What would be the likely damage if someone did connect a 24v battery pack to this thing with reversed polarity?
c) Would the dodgy 400v cap inhibit battery charging?
d) Can anyone suggest any other simple-to-check likely faults?

[Swake]

Where in the circuit is that cap located? Eventually post some clear pics of the board.

The lowest hanging fruit is to change the cap. Bulged means it is dead.

When that is replaced and it still doesn't work, find how to perform a factory reset or something equivalent.

Measure all the diodes and transistors with a DMM in diode mode. The method is slightly different for a  FET, look up how to check a FET on youtube, it's easy ones you know how.
Then check everywhere for shorts in the Ohm position.
After that if still nothing found, switch it on and follow the path of the power till you encounter the zone that causes trouble. Careful, lots of power there that will zap you very hard if you have a moment of inattention.
There wil be some chips that regulate the power somehow. If no one comes up with schematics you might find examples in the datasheet or application notes of those chips.

[skyhawk]

I'm not sure these photos will be any good, but it's about the best overview I can get with my phone.

The bulged cap is right next to the red positive battery cable. I'll order a new cap later today. I've already messed around with the communication protocol, other than resetting the battery installed date, which I think is purely informational as there's no RTC on this thing, there's not much else it can do. It can initiate a calibration pass, but calibration isn't the problem here.

I've had this thing assembled and operating outside of it's case, but other than touching the battery leads to measure terminal voltages, I'm not brave enough to poke around the board itself while it's live.

[Swake]

It would make a lot of sense that the cap is the root cause of the issues. Or at least it is part of it.

If not then:
Would it be that the UPS considers that the batteries are fully charged? Have you tried to put a substantial load on the UPS, at least 300 W to verify it starts charging after a while?

If it cannot measure the voltage of the battery it is likely not trying to charge it. Is that 8pin chip near the cap a UC39432? datasheet here https://www.ti.com/product/UC39432#tech-docs . If it is a UC39432 then please verify pin 6, I believe this pin could be in relation with the voltage of the battery.

Related schematics, maybe there is one that is close enough to your board:
- https://elektrotanya.com/showresult?what=APC+Back-UPS&kategoria=All&kat2=All
- Search for "APC Back-UPS" on this page https://www.eserviceinfo.com/index.php?what=search2

[skyhawk]

Would it be that the UPS considers that the batteries are fully charged?

The front panel quite clearly shows the battery is not fully charged, and it does decrease as time goes on and the battery is more and more drained. The runtime estimate also declines over time.

Is that 8pin chip near the cap a UC39432?

Close, but no cigar? It's a UC3843BN - a PWM controller - https://www.ti.com/lit/ds/symlink/uc3843.pdf

I've gone through those links, and I don't think there's any contenders. Most of the schematics are from the mid-1990s. This board has a lot fewer ICs than most of these.

I probably should mention that I actually own two of these units. One has been in service for I literally don't remember when I acquired it - I've probably had it for over a decade now.

[MLXXXp]

I probably should mention that I actually own two of these units. One has been in service for I literally don't remember when I acquired it

Having an identical working unit can be helpful because you can compare voltages, resistances, waveforms, etc. between them, to possibly determine what's wrong with the failed one.

[skyhawk]

Having an identical working unit can be helpful because you can compare voltages, resistances, waveforms, etc. between them, to possibly determine what's wrong with the failed one.

Yes, but that other unit is currently in-service providing redundant power for my home server. Taking it apart is... not something I'm looking forward to.

I found a schematic that might be at least loosely related to the device in question: It's from around the same time (Schematic is from 2004, board says Copyright 2006), and it's charging circuit includes a large high-voltage capacitor (That is domed out in my unit) across a four-diode bridge rectifier. I've confirmed the same configuration on my board. Interestingly that schematic calls for a 450V cap, whereas mine is 400V. Perhaps because this is a North American model only intended for operation on 120V mains?

[Swake]

Replace the bulged cap. Without a good cap in that place this circuit cannot deliver much power. You could start measuring if the PWM chip does its job and the feedback circuit around C17/Q41 is working, but again, without a good cap it might put you on the wrong path.

While you are at it you also might want to check the other caps. Do you have some means to measure these caps? Capacitance would be an indicator already, and series resistance (ESR) would be very useful in this case.

Funny how they put a 1.6A and 3A fuse in parallel. Whoever did that must have had some trick in his sleeves because that is really not something that can work in practice.

[skyhawk]

Buying an ESR meter has been on my to-do list for about forever.....

Does this look decent and fairly priced?
https://www.digikey.ca/en/products/detail/peak-electronic-design-limited/ESR70/9687220

[Swake]

The Peak Electronics ESR70 is an ok device with a good reputation.

However, my opinion is that it's design is now superseded by recent developments. In the same accuracy league for a fifth of the price and much more capable you'll find RLC meters based on an Atmega microcontroller. Plenty of declinations of this device out there. This is an example: https://www.aliexpress.com/item/1005005237553083.html .Not saying the one of the link is a good choice, verify before bying, it was just a random link. This atmega based device is actually a very clever open source firmware development originally made in Germany if I recall well. They still add features to it. The FNIRSI DSO-TC3 is the 'luxury' edition of this meter, but that is likely going the closed-source road again.

There is one important element that is yet a difference in the advantage of the ESR70: They claim to have a better input protection for when a capacitor was not discharged prior to measuring it. I have no idea whether this is true or not.

My goto portable device is a DER-EE DE-5000. Admitted it is double the price of the ESR70 and maybe not justified in your case, but it really is much much better.

[skyhawk]

That DSO-TC3 seems like a brilliant jack-of-all-trades tool to have in my toolkit.

But the manual suggests it doesn't measure Capacitor ESR, which is the point of this whole exercise?

[Haenk]

That DSO-TC3 seems like a brilliant jack-of-all-trades tool to have in my toolkit.

But the manual suggests it doesn't measure Capacitor ESR, which is the point of this whole exercise?

As a quick test, the C will show you, if it's still in the "good" range. My tester also displays R, but that might differ on the firmware version.
For in depth test, I also have a dedicated tester with pretty high precision, different frequencies and bells and whistles.
Not needed for a bulged cap though. Just replace it, no further investigation or testing needed.

[Swake]

It does, check these screen shots: https://www.eevblog.com/forum/testgear/fnirsi-tc3/msg5014999/#msg5014999
I read somewhere it only measures ESR above 1µF.

As you mention it is a bit of a jack of all trades, good enough in many cases but certainly not perfect. As is the case with any measurement instrument, you have to understand the limits of such a thing and up to what degree you can trust the reported numbers. For ESR measurement for example it will not be very accurate, but you don't care about accuracy, you need to know about the ballpark and eventually if there is a difference between 2 components and that is something the Fnirsi will do very well. Actually this is the same situation with the ESR70. Of course none of these should be compared to a 'real' and much more expensive RLC meter.

[skyhawk]

Replace the bulged cap. Without a good cap in that place this circuit cannot deliver much power. You could start measuring if the PWM chip does its job and the feedback circuit around C17/Q41 is working, but again, without a good cap it might put you on the wrong path.

Digikey's shipment got here a lot faster than I was expecting.

I apologize for doubting you. It looks like that cap is all the unit needed.

I guess I'm biased due to my experience with many PC mainboards with swollen caps that still worked. I don't have the ability to measure ESR (yet), but my old garbage hardware-store DMM measures the new cap correctly, yet cannot produce any capacitance measurement at all on the old one.

[Swake]

Well, that is good news. Happy ending.
And no worries, for the same that board indeed had more issues than only the bulged cap. Good that you checked the diodes and so on.

That computer mainboard is a slightly different situation in that it is unlikely that its on-board power supplies have to provide maximum power from the first moment. Consuming max power is only something that happens with the the biggest possible CPU at 100% CPU load + maxed out RAM and whatever else it supports. Most of the time there is some capacitance left in the bulged caps, not much but might be just enough to start that computer. In your UPS it is a different story, charging batteries it is likely going for max power immediately, without warning and without further monitoring, there is no room for this because it must be a cheap implementation. Whatever capacitance left in that bulged cap is never going to be enough.

[skyhawk]

Fun little update on the failed capacitor:

My DSO-TC3 just arrived. It measures one of the new capacitors perfectly. It thinks the old capacitor is a resistor usually somewhere in the 10Meg-ohm ballpark.

[Swake]

I've got a little box with my trophies...  Lot's of dead RIFA caps in there. Relatively few electrolytic caps. Plenty of chips and transistors with craters or cracks.