APC Back-UPS CS 650 Charging Circuitry Issue

[philipjfry]

Hello, dear magnificent community! 

Having successfully repaired 2 devices that would otherwise have been beyond my abilities in the past, I am once again turning to you brilliant guys and gals for pointers with an issue that I unfortunately can't figure out by myself.

This time, it's about some APC UPSs (UPSies?)  of the venerable Back-UPS CS 650 variety. I know it is an ancient model by now, but it is an amazing workhorse and astoundingly, it is still being sold today. Thanks to the USB connection, it also still does everything that is necessary today. I have a lot of them in the field and there are very few issues with them. The most common and well-known one is that Capacitor 26 (near the 40A fuses) bulges over time, causing the unit to prematurely complain about the battery needing to be replaced. An easy fix that I have applied successfully many times.

There is however another issue that I am seeing on a growing number of units now: The charging circuitry seems to fail. With a freshly charged battery, the unit works fine and does everything it should do (including providing backup power), but even during regular operation, it (completely) drains the battery over the course of about half a day, instead of keeping it topped up. Unlike working units, it also does not turn on if no battery is connected. I suppose this is because the internals would usually be powered through the charging circuitry in this scenario.

Since this model is serving me (or rather, my customers) so well, I know it inside out regarding behavior/quirks/strengths/weaknesses/limits and I would be able to repair not just 1 but probably a whole bunch of units (3 for now, certainly more in the future), it would be really neat if I (or rather, we) could figure out what is behind this failure. I'm hoping it's also something rather trivial such as replacing C26.

I do have basic education (U=R*I) and a bit of experience with repairing electronics (enough to not electrocute myself and to not burn down the house), but this one is once again beyond my diagnostic abilities. What I've done so far (beyond monitoring battery voltage, which behaves the way one would expect if it were never actively charged and used to power the internals) is to physically inspect the components for potential issues (none found) and checking all the fuses (all good). Unfortunately, I was not able to obtain a schematic, which is what a local repair guy insisted on before even looking at the issue (somewhat disappointing, he's not the most motivated person tbh).

So I was hoping that maybe one of you has come across this issue (or a schematic) in the past, or that you may be able to instruct me on how to best diagnose this issue? I should have most if not all the necessary equipment available. The only thing that comes to mind that I don't have is a scope, but I do have access to that too if necessary. I am including 2 pictures for now and I can of course make more if there are any areas of specific interest.

Thanks in advance for any help you can provide! 

[Peabody]

A schematic!
A schematic!
My kingdom for a schematic!

Copyright 1485  Richard III

[philipjfry]

A schematic!
A schematic!
My kingdom for a schematic!

Copyright 1485  Richard III

Yes, it would certainly be advantageous and I've been able to find schematics in the past when I really didn't expect it. Alas, this time it's the other way around. I'm hoping that the charging circuitry for a 12V lead acid battery is simple enough to make do without one?

Two things I forgot to mention:
1. Unfortunately, I only have a working 350VA model at hand right now, which is quite similar but still somewhat different. I'll post pictures of that asap. I should be able to get a hold of an identical unit for comparisons too if that helps.

2. If we manage to revive this unit, I'll make sure to put some more solder on the traces on the bottom of the backside...

[floobydust]

elektrotanya has a few APC schematics. They are a cheap product.
Usually it's the electrolytic capacitors aging because these are on 24/7. Check C38, C26. Look for DC at C26. Fuse F6 too. R131, R131A looks a little dark.

[philipjfry]

Thanks floobydust!

While they are indeed value oriented rather than premium, they are still significantly better than other products that I have seen over the years. Especially some of the China brands. I somewhat like the fact that they represent the minimum viable UPS. I'm guessing they are as simple as they reasonably can be, which is a quality on its own, and they have a comparatively beefy transformer.

I checked out elektrotanya, the closest model I could find a schematic for is the BK650MI, which is relatively close (the code for mine is BK650EI). At first glance, it looks quite a bit different (more premium) on the outside though and also uses the bigger battery (12Ah vs 7Ah), so I'm not sure yet how useful this schematic will be. The part number also diverges quite a bit (640-0214E Rev. 05  dated 1997 vs my 640-0565-Z Rev. 05 dated 2004). I'll check how far the similarities of the charging circuitry appear to be tomorrow, with a pair of fresh eyes.

Today I can report:
C26 (specced 470µF 16V) I already replaced prior to posting. The old one was bulging and down to 300µF, if the capacitance meter can be trusted. The new one charges up to battery voltage if a battery is connected and strictly discharges once no battery is connected. It drains faster if the power button is pressed while no battery is connected. It is completely unaffected by AC power (which it probably shouldn't be, I'm guessing it is downstream of the fault on the AC side).
C38 (specced 22 µF 50V) measures 8.5V when the unit is connected to AC, regardless of it being powered on or off. I desoldered it and it still measures bang on 22µF. It also keeps a charge, I charged it up to 30V and let it sit for a while, just to be sure.
F6 shows continuity. I did previously check F4, F5 (the two 40A automotive style fuses) and F6, as well as the circuit breaker on the back of the unit. F1, F2 and F3 are absent on this revision.
R131, R131A and R120 (all below each other) show exactly the values that are printed on them. The darkening is/was caused by old flux from the factory, which I had to scrape off to get a reading.

[floobydust]

I would check the IC (p/n?) next to C38 likely needs more than 8.5V for it to startup.

[philipjfry]

That's a TI TL3843 with an UVLO of 8.4V. Coincidence? Maybe not! 

I'm also including pictures of the working 350VA model I have at hand. Looks quite different, unfortunately.

[philipjfry]

I just had a minor brainwave and realized that I haven't hit this issue with my InfiRay P2 Pro thermal camera yet. No idea why, I love the thing and am using it regularly, including for diagnosing issues such as this one.

Anyway, as soon as I connect the unit to AC power (without a battery inside), one component (and just this one component) instantly lights up brightly! R56 heats up to over 80°C in less than a minute - that's unlikely to be normal behavior, right?

[philipjfry]

Small Update: I did desolder R56 and it measures 100k\$\Omega\$, as is should, according to its brown, black, yellow, gold color code. So, if the high temperature is irregular, it is a symptom rather than the cause. Since it is built to dissipate heat, I guess it's somewhat normal after all. In circuit, it measures 100k\$\Omega\$ in one direction and 11.5k\$\Omega\$ in the other, which also makes sense I suppose.