I was reading this article

CyberPower LX1500GU-FC UPS Tear-Down
https://www.tomshardware.com/picture...tear-down.html

I have bought a CyberPower LX1500GU3 recently.

Of course, when the power goes out, a UPS uses the battery as a source of power.
It looks like some of these UPSs output an AC where it doesn’t look like a sine wave at all.
It looks more like a square wave except that every other square is inverted.

Is this bad?
I tried it on my desktop, 2 laptops and they are able to run on the UPS while wall power is cut.

How does this effect the DC 3.3 V, 5 V, 12 V that the power supply of a desktop delivers?
Do they remain absolutely flat?

From the article you linked...

That sounds awful to me! And I have a CyberPower 1500VA UPS under my desk but have never looked at the waveform until now... oh my...

All UPSs will output a "modified sine wave", usually constructed from a stepped square wave, where the number of steps and their placement is designed to effectively knock out the worst of the harmonic distortion, at the expense of higher component count. Higher power systems will typically have a 12 step waveform notched at the peak and a THD of a few percent, smaller units are usually a bit worse. but 46.8 percent - that literally means 46.8 percent of the output power is in the harmonics. I am confused by the peak number being lower than the total and wonder how they have actually spec'd that or if it is a typo. I cannot find THD in the specs for mine.

The "squareness" of corners on the waveform is due to the higher harmonics and can be very hard on some power supplies, especially switching power supplies - which is just about every thing when it comes to computers.

All of that said, I cannot conclude this unit is actually a potential problem, but it does cause me some concern at the moment. I won't sleep well tonight, thank you very much!

If it were an online UPS where the load runs on the inverter all the time, I would not plug anything I have into it. But with standby operation where you are only on the inverter when the power goes out, and only for as long as it takes to shut down... a better case, but still.

Square (or trapezoid or any other non-sine) wave is not good for inductive loads i.e. transformers where higher harmonics will dissipate causing excessive heating and buzzing, but transformers nowadays are rare. Your typical computer PSU doesn't use them and won't be affected.

Info below I got from Mr. Bard (Google AI)

If your computer power supply is labeled as green i.e. has some sort of active Power factor correction circuit then you probably want a true sine wave UPS. As posted a standard switched mode PSU should tolerate a modified sine wave without problems and I have used both types with computer equipment for years. Sensitive electronics always gets plugged into a true sine wave UPS.

Most electronic devices prefer pure sine. The USP's don't make square waves (many do but sold as pure sine,) they make a modified sine wave that many systems are able to filter.

I thought when they say modified sine wave, that it was actually a bunch of steps to approximate the hill (going up) and ditto for the hill going down.
But this one is pretty brutal. It is a single big step.

Square wave:
So, there is a huge induction that happens at the rising step
then it is flat (no induction happens)
then there is a sudden drop in voltage so there is a huge induction again.
then it is flat (no induction happens)
There is a sudden drop in voltage into the negative rage. So there is a huge induction again.

Transformers are not the only inductive components and there are other effects.

Any device which must filter its power source to some degree, which is every electronic device we own, will be affected in one way or another. Absence of high inductance passes the energy in the harmonics along to other components, and ultimately to the load.

Switch mode power supplies, which is pretty much every computer PSU these days, can become much more active in the presence of higher harmonics as they attempt to satisfy their load, and limits, from their source. This can manifest in the form of excess heat, capacitor degradation and also in the form of higher noise levels on the output voltage which can then translate into data errors.

The effects increase in proportion to the harmoinc distortion, so the rule is:
And the question is not whether a given device is affected, but by how much and in what way it is affected.

THD approaching 50% strikes me as far from where you really want to be.