Jimmy wrote:
The Powerwalker sure has lots of specs, but it's pretty difficult for me to judge whether it's better or worse than others, because the others don't give that level of spec.
Did you already read the cyberpower cp900epfclcd datasheet here?
https://www.cyberpower.com/eu/en/Fil...DS-18080023-01

By the way, I amazon-ed a little bit to search an USA cp900epfclcd's "brother" and found the cpXXXpfclcd "family". As you can see the "e" is missed and I guess it stays for european (?). If yes, in the sense if there're no other important differences, consider this
https://www.amazon.com/CyberPower-CP...dp/B00429N18S/
You can find infos he
https://www.cyberpowersystems.com/pr...e/cp850pfclcd/
and its datasheet here
https://dl4jz3rbrsfum.cloudfront.net...P850PFCLCD.pdf
The first difference I read is about the on-battery wave output: just "sine wave" and not "pure sine wave" as the EU family (cpXXXEpfclcd) BUT if we look at amazon buyers' comments, we can get these graphs:
-CP1500PFCLCD
https://www.amazon.com/gp/customer-r...SIN=B00429N19W
https://www.amazon.com/gp/customer-r...SIN=B00429N19W
https://www.amazon.com/gp/customer-r...SIN=B00429N19W
https://www.amazon.com/gp/customer-r...SIN=B00429N19W
-CP1350PFCLCD
https://www.amazon.com/gp/customer-r...SIN=B00429N19M
We'd trust as usual. Unfortunately there's nothing similar for the 850VA model within comments. My only concern is I guess there could be a certain possibility cyberpower built the "upper class" models (eg. 1500VA) with a better electronics than the "lower class" ones as the 850VA. I say that because AFAIK certain brands do this job: different models quality within the same models "family" (eg. a laptop series); what do you think about this cyberpower case?

I share your same doubts about the powerwalkers because for now it's hard to find many feedbacks to "study". Just this 800VA model
https://www.amazon.it/BlueWalker-Pow.../dp/B00K791C1U
has 101 ratings and they're "quite good". My "favorite" comment is
https://www.amazon.it/gp/customer-re...SIN=B00K791C1U
and tells the main problem is the transformer is not cool down enough by the heatsinks so the UPS won't last so much.

By the way, there's also the PSW powerwalker series
https://powerwalker.com/?page=select&cat=VI_PSW&lang=en
and it's more exspensive than the SW series I wrote above. Honestly I don't understand the differences, can you help me? Let's consider the 750VA model
https://powerwalker.com/?page=produc...120081&lang=en
with its datasheet
https://powerwalker.com/datasheets/P...SW%20Promo.pdf
I don't think the difference SW / PSW is just "sine wave" / "pure sine wave" because the SW generates pure sine waves too, so what's different to the SW series brothers?

The one that claims "PWM" and "98% efficiency" would
be a class D digital one which "draws the waveform with narrow
square waves", then follows that with a low pass filter to
only pass the fundamental. (At least, that's the one principle
I can imagine doing this.) The low pass filter is in effect
the regeneration filter. Because it's PWM, the transistors
in the output stage are either open (OFF), or saturated ON, which
is where the "digital" terminology comes from. Saturated transistors
dissipate little heat, and then it's the regeneration filter
that's the only questionable part of the design (how is it
done, does it generate heat). The 98% efficiency, that tells
you how many watts the inside of the box dissipates. Not much heat.

They also make stereo equipment that way. Class D with filter.

As for the one that has the overly hot transformer in it, it
could be a ferroresonant design. The heat could be a sign of
eddy currents and harmonic content being driven into it. I don't
know the operating principle well enough for ferroresonant,
to say more about it. Generally, if a design is not efficient
and is kicking off heat like that, that's the kind of
design that needs "constant fan" for as long as the heat
is present. If the users find it hot with the cover off (where
at least it gets convection cooling), it's going to be much much
worse with the cover on.

You can see in the oscilloscope waveform with the glitches in it,
that's some sort of crossover distortion. But it might not be
exactly the same as you'd see in stereo equipment. That glitch
would represent a bit of harmonic content, which generally would
not be too bad for an ATX active PFC load, but I suppose would
make an inductive load absorb some of that. You might just
barely be able to hear that in an absolutely quiet room,
with say a turntable motor as the load. The UPS itself
might make more noise than the turntable motor though.
Strictly speaking, that waveform is not "Pure Sine", which
implies certain operating principles, it's just a non-sine
source that is a better approximation than Stepped Sine.
That's why you'd call the output "Sine Wave" output, because
it's not quite as good as Pure Sine. It's a good enough
sine wave, a cost conscious compromise. With Pure Sine, the
method used (whatever it happens to be) is inherently glitch free.

Pure sine, used to start at $1000 or so. Any price movement
we see there today, is the evolution of the concept. The
method used in the old pure sine, might not be the same as
the ones we see today. That's because Class D only became
a fetish in more recent times. There are some 500W computer
speakers that are Class D, but two stage Class D, with one
stage being a kind of programmable power supply, the second
stage drawing the final waveform, with little thermal overhead.
The drop across the final transistors is only a few volts at most.

Paul