Re: RE: UPS lifespans

On 5/29/2024 9:49 AM, Carl wrote:

An n=1 anecdote :-). I bought a 550 VA APC UPS in 2015, in 2022 it started
beeping that the battery was low so I replaced it but the unit still beeped.
Wouldn't charge the new battery, tried various things including charging the
battery separately and reinstalling it but no joy. The manual said there was
no reset function, it should just come back to life when the new battery was
plugged in The old battery still tested ok, down some capacity but not dead,
so I gave up and bought a 600 VA APC and it's still going after 2 years.

There is a "reset" process for APCs that doesn't see much attention.
And, I am not sure what it ACTUALLY accomplishes.  Likely just ensures
the MCU forgets EVERYTHING (whereas their firmware implementation likely
doesn't know how to accomplish a "total purge" of state.
[I think they just use dinky little 8051's or equivalent.  And, you know the
firmware isn't all that complicated.  But, the idea of wasting time reverse
engineering it is not particularly exciting]
Google, "APC brain dead" (no idea who decided to call it "brain deading"
vs. "reset"...)
Their battery management systems suck -- they often overcharge or, in your
case, fail to see the battery ("Connect Battery" prompt on units with a
display)
I've got one, now, that claims 999 (!) minutes of available up-time if
power were to fail with the current load -- wanna bet it thinks the
current load (this machine, modem, router, etc.) is "0" and "999"
is its notion of "3 digit display infinity"?
[Even if that were the case, it should be abundantly clear to the designers
that the UPS won't support ITSELF -- in the absence of power -- for that long.
So, why would it ever display something like that?  Speaks volumes about the
quality of their engineering!]

Room lights flicker maybe 2-3 times a year and go completely off long enough
to annoy the microwave clock every couple of years so pretty clean power
here. The battery spends it's entire life just floating fully charged. Don't
know how long it would last if it was being cycled frequently.

Most of their models have a "self-test" feature.  The obvious implementation
is to deliberately isolate from the ACmains and force itself to resort to
battery power.
Of course, if the battery is incapable of supporting the load (or the UPS
itself) this will create a REAL power failure, crashing your loads regardless
of what you may have been doing when IT decided to run the test.
A smarter approach (for a "battery check"), of course, would be to provide a
small load to the battery -- even while it is being charged -- to infer the
SoC of the battery without forcing the user to be suddenly at its mercy.
This test can be disabled or its frequency altered.
It would also be nice if the battery management treated each of the batteries
as an FRU (yeah, I'm sure they want to treat them as a PACK as they are in the
battery selling business and just sell UPSs to create artificial demand!  :>)
I am amused, though, that I've not (yet) encountered any failures in the power
electronics.  OTOH, I deploy the units with plenty of margin (it would be interesting to plug a circular saw into one to see how well it protects
itself!)