Why the overvoltage when run on a rectangular wave UPS

[Deleted member 191766]

This guy has found excessive voltages on the main capacitor of the APFC part of a power supply when run on a rectangular wave UPS

https://digilander.libero.it/hibone/UPS/PSU_and_UPS_Compatibility_Test_Report.pdf

but why is this happening?

 

[OneMoar]

because its rectangular wave you sort of answered your own question ........

 

[dont whant to set it"']

Aren't them all rectangular wave? Yes them are.

 

[R-T-B]

Aren't them all rectangular wave? Yes them are.

There are sine wave and modified sine wave (stair step) so no.

True square wave is pretty low tier now.

 

[Deleted member 191766]

because its rectangular wave you sort of answered your own question ........

But WHY is this happening

 

[R-T-B]

But WHY is this happening

Because squarewave gets/stays ultra high at the wave peak? That's about all there is to it.

 

[Bill_Bright]

Aren't them all rectangular wave? Yes them are.

No, not at all. As 10 seconds with Bing Google would show.

BTW, I am not following that digilander link. My security apps don't like it.

And FTR, they are not rectangular. Nor are they square. As R-T-B notes, they are either pure sine wave or stepped, typically called simulated, modified or stepped approximation wave forms.

But WHY is this happening

Did you try Bing Google? It really is amazing what folks can learn and how quickly (much faster than waiting for a forum reply) they could learn for themselves if they just took a few seconds to type their questions in their favorite search engines. Others include DuckDuckGo, dogpile, Yahoo and others.

These search engines are great to verify what we think we know before posting inaccurate information too. I highly recommend them.

***

Batteries output DC. AC power supplies don't run on DC. So the voltages need to be inverted and turned into AC. That is not easy (or cheap) because a simple inverter does not provide a clean enough wave form so it has to be simulated or stepped and that takes more processes.

Think of a digital image. Curves are not really curved. They are jagged little steps. As the steps get smaller, the more the curve looks like a real curve, but they are still stepped.

See https://www.kstar.com/indexproblem/...n't really a sine wave, modified or otherwise.

Fortunately, power supplies are easily capable of dealing with stepped approximation wave forms - and have for decades. Don't let anyone try to convince you that you must go with a "pure sinewave" output UPS. There is a lot of marketing "hype" out there claiming pure sinewave UPS are better and a lot of hype claiming active PFC power supplies need pure sinewave power. That is pure marketing hogwash! Don't fall for it! That's all it is, marketing hype by those companies who make pure sinewave UPS trying to take the market share away from APC, who sells mostly "stepped approximation" output UPS.

Any 1/2 way decent AC/DC power supply can handle the "stepped sine wave" or "stepped approximation" waveform just fine. They have for the last 25+ years with no problems so there is no reason to believe the much more reliable and robust power supplies of today can't either. They are much more capable at handling power line anomalies than PSUs of yesteryear.

And that's not just computer PSUs, but the power supplies for our modems, routers, switches, monitors and more. The only time there really is a need for a "pure" sine wave output UPS is when used to support highly critical and sensitive health monitoring and life support equipment as found in hospital intensive care units.

 

[R-T-B]

Heck, I ran AC fan motors in the past on my tripplites stepped sine wave. Other than a small RPM reduction, even they had no issue, and motors are technically the place you should see it most.

I have seen true square wave (which can be problematic) in really cheapo UPSes and battery inverters, but that is the least of my list of concerns about using them. If you buy a decent UPS, this’ll be a nonissue.

 

[Bill_Bright]

Heck, I ran AC fan motors in the past on my tripplites stepped sine wave.

Yeah, I have my 3/4 HP garage door opener on my 25 year old APC 900VA UPS. Works fine. But motors are not high speed digital devices either.

 

[newtekie1]

But WHY is this happening

First, it isn't necessarily overvoltage unless the cap isn't designed to handle that voltage. Not knowing the PSU they used personally, and not knowing the circuit, I don't know if that cap is actually being overvolted. I will say, most consumer PSUs with APFC tend to use 400v caps for the bulk cap, so this would be driving those caps a little hard.

As to why this is happening, it is likely due to the voltage spikes at the beginning of the square wave forms. These spikes go beyond the normal output voltage, and that is what is causing the cap to be overcharged. You can see these spikes in the oscilloscope shots.

I don't think it would be much of an issue for the small amount of time the PSU is likely going to be running off the UPS, and the small time of the spikes.

Heck, I ran AC fan motors in the past on my tripplites stepped sine wave. Other than a small RPM reduction, even they had no issue, and motors are technically the place you should see it most.

I've done the same. Running motors off modified sine/square wave tends to make the motors hum pretty loudly, at least to me. But, hey, when the power is out I'll live with the loud hum if it means I have a fan pointed at me. I'll also say the fan motor did get a little warmer than usual when running off a square wave, but it wasn't dangerous.

 

[Deleted member 191766]

Not to be a pain... but one of the voltages was 497V (Model: 1000VA) and no beginning spike showed on the oscilloscope trace.

I have a theory, but the circuit diagram is at home, and I'm not sure my reasoning is sound; so I wanted to know what the folks here thought.

 

[newtekie1]

Not to be a pain... but one of the voltages was 497V (Model: 1000VA) and no beginning spike showed on the oscilloscope trace.

I have a theory, but the circuit diagram is at home, and I'm not sure my reasoning is sound; so I wanted to know what the folks here thought.

But the oscilloscope they are using is also a piece of junk, so it's easy that the spike just isn't showing up. Note on the oscilloscope shots that are missing the spike, the rise and fall lines are either missing or half missing too.

 

[R-T-B]

Yeah, I have my 3/4 HP garage door opener on my 25 year old APC 900VA UPS. Works fine. But motors are not high speed digital devices either.

Oh no, not at all. But you'd think logically speaking AC motors would be the first to exhibit issues due to how they work. But I get your point, a digital computer has so many little things that it has room for failure in it's own right.

Point is, being of even halfway decent build quality, none should have an issue.

 

[Deleted member 191766]

... and not knowing the circuit ...

I finally got to the circuit diagram

 

[Bill_Bright]

Oh no, not at all. But you'd think logically speaking AC motors would be the first to exhibit issues due to how they work.

IMO, the biggest issue with AC motors is the HUGE current demand they place on the circuit when first spinning up until they get to their operational speed. That is, it takes a lot more energy to get a motor (same with a gas engine) up to full speed than it does to maintain that speed.

It is a similar problem with laser printers. They should not go on a UPS because laser printers require a lot of energy to fuse the toner, but even more so to heat up the fuser during the initial power on process.

But I get your point, a digital computer has so many little things that it has room for failure in it's own right.

Not only that, but a spinning motor has momentum and will keep spinning for a short time if there is a dip or short duration sag in power, and consequently recover quickly after the anomaly passes. High speed digital electronics are likely to come to total stop (halt) state and will wait in that state until signaled to start again.

 

[newtekie1]

I finally got to the circuit diagram

That isn't what I meant. Yes, I know what a APFC circuit design looks like. When I said I didn't know the circuit, I meant I don't know what component values they are using in their design. If they are using a 500v capacitor then it's not being overvolted in the link from the original post.

 

[Deleted member 191766]

Good point; wish the author of the article

https://digilander.libero.it/hibone/UPS/PSU_and_UPS_Compatibility_Test_Report.pdf

had reported the values.

But what I am after is understanding WHY the voltage gets higher, and the answer is not 'because its rectangular wave '.

Because squarewave gets/stays ultra high at the wave peak? That's about all there is to it.

As I understand it, it is a rectangular wave whose height matches that of the sine wave and whose width is such that the resistive heating is the same as the sine wave it replaces; i.e. it is built to be the same height.

Did you try Bing Google? It really is amazing what folks can learn and how quickly (much faster than waiting for a forum reply) they could learn for themselves if they just took a few seconds to type their questions in their favorite search engines. Others include DuckDuckGo, dogpile, Yahoo and others.

These search engines are great to verify what we think we know before posting inaccurate information too. I highly recommend them.

Of course; that is why I am here to inquire of the collective brain.

I actually have a theory based on the circuit diagram I attached, but I am not sure that theory is sound.

Apologies that it took me so long to get back with a reply.

 

[newtekie1]

But what I am after is understanding WHY the voltage gets higher, and the answer is not 'because its rectangular wave '.

And I already gave you that answer.

 

[silkstone]

I thought that square wave input was much better than sinusoidal voltage for devices which use a switched mode power supply?
In theory they should.

 

[Deleted member 191766]

And I already gave you that answer.

Not all waveforms had spikes.

 

[Bill_Bright]

I thought that square wave input was much better than sinusoidal voltage for devices which use a switched mode power supply?
In theory they should.

There is some truth to this - though in this case "better" should probably be defined as "easier" rather than implying sinewaves are not as good or even bad.

The reason why is because in this scenario, DC can be thought of as basically a square wave that has been rectified - that is, one polarity (the negative side, for example) has been flipped so now you have 2 positives for a constant positive with a "flat top" on the waveform. That is a lot easier to achieve than a sinewave which is really constantly varying voltage. Also, when producing DC voltages, you want to remove (suppress) all AC components (such as ripple) riding the DC output voltage. This is a lot "easier" to obtain when you don't have to employ as much filtering circuitry when you already start out with a "flat top" waveform.

While this article was written way back in 1985/1989, it is still applicable today:
UPS - UNINTERRUPTIBLE POWER SYSTEM WAVEFORMS
IS A SINE WAVE NECESSARY?

This is why the push for pure sinewave UPS is nothing more than pure marketing bullfeathers. The only real reason pure sinewave UPS have become more popular in recent years is their prices have [finally] come down so they are competitive with stepped approximation or simulated sinewave (the traditional type) UPS.

 

[newtekie1]

Not all waveforms had spikes.

And apparently, by your logic, not all the wave forms had rising edges either. We know they do, but if that junk scope doesn't show it, it must not exist, right?

 

[Deleted member 191766]

By the same logic; if the scope doesn't show it, it is still there?

My guess is that these are unloaded outputs.

I thought that square wave input was much better than sinusoidal voltage for devices which use a switched mode power supply?
In theory they should.

Well, the instantaneous currents can be higher, and so the heating losses more.

 

[newtekie1]

By the same logic; if the scope doesn't show it, it is still there?

My guess is that these are unloaded outputs.

Yes, because you can see artifacts of them on the scope. All square wave AC outputs have a spike. The design and components necessary to suppress the spike are so expensive, that implementing them is just as expensive as just using a pure sine wave inverter.

The square wave waveform always has a rising edge that overshoots the voltage(spike), then a small falling edge that dips below the desired voltage, then a small rising edge again that gets pretty close to the desired voltage. The reality is that the voltage actually continued to oscillate the entire time until it drops back to 0, but the oscillation is so small after the initial spike it doesn't matter. I have never seen a square wave UPS that didn't have these spikes.

On the pictures that are missing the spike(and part or all of the rising edge) still have the secondary smaller rising edge that happens after the spike. So we can use that to assume that the spikes are still there. Again, yes, this is only an assumption because of the garbage scope, but it is the best assumption based on the facts we know. /thread

 

[Deleted member 191766]

Fair enough

Here is my conjecture: when the voltage and current match in phase the PFC control has nothing to do and one gets a certain voltage on the capacitor; but

• If the PFC circuit is active (trying to phase match) it will be pulling current off (see the attached circuit diagram)
• Now, since the circuit is designed not to waste energy the output DC I*V must be the same... so if one is reducing the current out, the voltage out must increase.

By trying to understand the cause I was wondering if there was a way to avoid the problem; in this case phase matching to a rectangular wave is probably not worth the trouble.