Announcement

**willawake** · 04-03-2008, 09:00 AM

Re: square output UPS vs. bulk capacitor voltage

2 x 200v in series = 400v
i didnt understand why you have 157v mains.
your ups sounds ok
silverstone's was interesting

**i4004** · 04-03-2008, 10:43 AM

Re: square output UPS vs. bulk capacitor voltage

2 x 200v in series = 400v

and 2 x 213 = 426V
<wink>

mains 157V?
"first off let's see voltage on bulk capacitors when we're powering the psu off the mains"
bold added now(only thing i was measuring was voltage on caps)

it's apc ups, yeah, but i didn't know it was producing such a simple waveform, i thought that at least it got few more 'stairs' between 0 and the peak.... but it's the same all across the ups market...'true sine' models aside...

btw. had a power failure at the time your fortron failed?
<smile>
on silverstone's pdf 1000VA ups produces almost 100V overshoot...woohoo....

**willawake** · 04-03-2008, 11:39 AM

Re: square output UPS vs. bulk capacitor voltage

ok i get it now. i read everything again carefully

**Per Hansson** · 04-03-2008, 11:55 AM

Re: square output UPS vs. bulk capacitor voltage

Read the thread at Johnny's

I made quite some effort to explain it all there, why it happens and so on...

**i4004** · 04-03-2008, 12:46 PM

Re: square output UPS vs. bulk capacitor voltage

i like it simple

1-"we can also see square has larger amplitude(square needs to have larger amplitude to produce same power as sine)."

2-"only the (square) voltage peaks are charging the caps, so it's natural to have increased voltage here"

**zandrax** · 04-03-2008, 04:08 PM

Re: square output UPS vs. bulk capacitor voltage

I still don't understand what's the cause: the high transient in voltage steps (the peak) or the stepped technology?
Does make sensa adding a voltage limiting device between the ups and the psu ( I imagine the equivalent of a thermistor applied to voltage instead of current)?
However, I agree with i4004 about the deceiving "Stepped sine wave" name: it's a lie because it's a stepped version of the square wave, there is nothing related to sinusoidal wave at all.

Zandrax

**kikkoman** · 04-03-2008, 04:32 PM

Re: square output UPS vs. bulk capacitor voltage

the *stepped* square wave doesn't make sense to me in the first place. we're powering a SMPS here (which is the purpose of every UPS).
correct me if i'm wrong, but wouldn't a real square wave make more sense? it gets rectified anyway.
at least the duty cycle could be increased (thus lowering the amplitude).

and these voltage spikes are HUGE. they sure don't help the issue... it can't be impossible to reduce them a bit.

**davmax** · 04-04-2008, 12:09 AM

Re: square output UPS vs. bulk capacitor voltage

I could not resist this one. The whole idea of having a "modified square wave" sometimes called "modified sine wave" is simplicity and low switching loss. The modified square wave principle is supposed to set the peak of the square wave to equal the peak of an equivalent sinewave whilst the duration of the square wave is set so that the RMS power is equal to the equivalent sine. If these are set correctly there should be no problem with either capacitor max voltage or the correct power delivery.

The square wave in the pic shows an overshoot which can cause a lightly loaded or noload cap to charge too high.

If cap voltage is too high under load then the dimensions of the square wave are incorrect.

**i4004** · 04-04-2008, 01:44 AM

Re: square output UPS vs. bulk capacitor voltage

zandrax, what will you get if you run such square thru the rectifier?
and what value will that dc have if we've said that these pulses have higher apmlitude than sine?
got it?

davmax i would say here it's more important what happens without load(better to say with very light load of +5vstb powering the pc..ie under 5w or so).
what happens is that psu is plugged to ups and this situation continues for prolonged periods of time(if the power outage continues). and that is not good.
i mean it's not instant disaster or anything, but it's not good for those caps.

**davmax** · 04-04-2008, 02:24 AM

Re: square output UPS vs. bulk capacitor voltage

1. The PC input caps can only charge up to the peak value of the waveform no matter whether square, triangle or sine.

2. The UPS square wave peak voltage should not be more than the normal sine, if it is there is a problem with voltage control in the UPS. The pulse width does not vary, it is fixed at a well establish peak to RMS ratio. The peak voltage is set by the voltage of the DC supply for the output switching transistors, this voltage should be controlled by voltage regulation so if it is not right it is either not adjusted correctly or there is a fault. Where the sample of output voltage is taken for control will depend on the design.

When you consider the principle it is quite simple.

**i4004** · 04-04-2008, 05:13 AM

Re: square output UPS vs. bulk capacitor voltage

this voltage should be controlled by voltage regulation so if it is not right it is either not adjusted correctly or there is a fault.

no.
my device is working properly ie just as it was designed to, but the problem is, it's wrong design.

this situation repeats with ALL ups' (measure voltage on yours) so it's not a malfunction of any sort....

they are simulating rms of sine by increasing peak of this square rubbish.

true, solution should be to have same peak value as sine, and to compensate(to have same rms value) width of the pulse should be extended, so it looks more like this

and less like my first pic in this thread

and apc sine output ups' seem to have ridiculous prices....

**davmax** · 04-04-2008, 05:31 AM

Re: square output UPS vs. bulk capacitor voltage

So it appears to be bad design ie wrong peak to RMS ratio. Very strange that they should produce a product this way. It is very simple and costs no more to get it right.

You certainly have raised the propsect of hidden danger is UPS units.

**starfury1** · 04-04-2008, 02:22 PM

Re: square output UPS vs. bulk capacitor voltage

Yet to fully read links and the above but on the Liberty PSU

As I have said, I wonder what the story is on how many of theses go boom on an UPS (and yes it is stated SINEWAVE inverter)
and how many go boom on mains only

As I have said in other posts,
these use A-PFC which I suppose is built around the fact its feed with a sinewave not a square wave (or something closer to one)
but I don't know if this is the real reason they go "boom", just suspected it

Also this raises an interesting point in that if you go and buy what you "think" is a sinewave inverter because of the use of the terminology.

anyway on "misleading" found this PDF here

So humm yeah to me thats a very broad use of the term "sinewave" to me anyway

this,
(if that is whats produced as output)
look closer to a sinewave

Stepped sinewave Quasi sinewave

http://www.jet-electro.com/Html/faq-inverter.htm

interesting thread I4004 thanks for raising this

Yeah it looks like ringing in that square wave, correct me if I am wrong here but isn't that got to because of some reactive element
(component) inductance,capacitance?
(guessing further you are going to get some depending on situation and load, regardless)

Yep dont know much on UPS units either,
except only from a form concept and what they are suppose to do

Cheers

**zandrax** · 04-04-2008, 04:25 PM

Re: square output UPS vs. bulk capacitor voltage

Originally posted by i4004

true, solution should be to have same peak value as sine, and to compensate(to have same rms value) width of the pulse should be extended, so it looks more like this

and less like my first pic in this thread

Now I understand, i4004: the dangerous part is the big offset between sine and stepped wave on step extremes.

Zandrax

**davmax** · 04-05-2008, 07:34 AM

Re: square output UPS vs. bulk capacitor voltage

I4004 has raised what seems to be a very important issue regarding modified square wave UPS units that have been tested on PC power supplies.

Summarising

His testing and that of PC power supply manufacturers reveals that the modified square waveform in these inverters has an excessive peak voltage resulting in voltage that will cause the input filter capacitors of a PC power supply to be put at risk with excess voltage.
- The normal peak voltage of a 115VAC sine is 163 volts however the APC UPS units tested delivered 197 -199 volts to the 200V rated capacitors in a non PFC Enermax, a 22% over voltage.
- The capacitor voltage in a Fortron power supply with passive PFC reached 213 volts, a 30.7% overvoltage or 7% higher than the non PFC power supply.
The evidence is strong that the UPS peak voltage is too high and a risk to computer power supplies. The modified square wave method implemented with the correct peak voltage of 162 volts and the appropriated pulse width for the correct RMS voltage will not present any problem to a computer power supply even with the +7% caused by passive PFC (174 volts on capacitors). Here is a link that graphically displays the correct sine to modified square wave relationship, where the sine peak is not exceeded. http://www.pcguide.com/ref/power/ext/ups/funcOutput-c.html
It is incredible that there is a possibility the APC and others are delivering UPS product that would create this overvoltage risk. It seems only right to contact APC to receive their explanation and any remedy.

This information and the link to this Badcaps thread has been sent to APC for them to respond.

**i4004** · 04-05-2008, 08:50 AM

Re: square output UPS vs. bulk capacitor voltage

dave, i'm on the 220V grid.

as for 120V grids, well, somebody should measure it.
i feel this post by per is wrong
http://www.jonnyguru.com/forums/show...3&postcount=49
because he took 470V(on the caps) of a 220V ups system as a reference, but i don't think that's correct.
it should be measured.

but given everybody here is from australia or europe i dunno how can we do that?
hehe....

did kc8adu say he's doing some psu work?
if anybody from usa attemps it, be carefull upon measuring: this is not low voltage!

what we have concluded so far is that 220V "Stepped Sine Wave" ups' are overloading psus when they're powering the system from the battery.

**starfury1** · 04-05-2008, 10:01 AM

Re: square output UPS vs. bulk capacitor voltage

yeah I was looking at Per's post,

Now correct me if I am wrong on this people

With the older Bridge rectifier 2 cap supply setup, its as follows

The bridge rectifier with caps in series (across it) is the case for 240V
(now in Oz 230VAC@50Hz)
(Ideally 2 resistors are used as a voltage divider across the caps to force equal voltage sharing...well thats one purpose for them, thats if they bother to put them in..if you have a look at the guts of the El -cheapo jobs)

USA (or other 110 VAC) it is a "Voltage Doubler" when the switch is thrown to
that position.

Net result you got 340 VDC (aprox) to chop up and use in either case

Apparently there are some Auto switches jobs but not sure how they were generally set up I know chips for this do or did exist

Now I am assuming that the stop gap Passive Power Factor Correction method, was to basically throw large inductor in, in series with the above.

From cypherp post

Active Power Factor Correction is done with to quote Cypherp

"Rather than a simple rectifier, this uses a non-isolated "boost converter" as the input stage." also says "PFC boost preregulator" which might be a more accurate name for the circuit

(and yeah really lost on that setup)

So I just got to think about it before I go opening my big trap and sticking my foot in my mouth

and up setting Per

cheers

**starfury1** · 04-05-2008, 10:09 AM

Re: square output UPS vs. bulk capacitor voltage

See Pete's tech outline of how an SMPSU works Here
"Rectifier filter" section

**i4004** · 04-05-2008, 02:42 PM

Re: square output UPS vs. bulk capacitor voltage

i think per was wrong on 2 accounts:
1-why take 470V example and not others? you have ups' doing (the voltage on caps) from 408 to 497V on those examples.
2-per said only one capacitor is used, but i think that's not correct:
http://www.kronjaeger.com/hv/hv/src/mul/
it's visible both caps are used and each only need to do 1xUp.

the issue is the following: some ups will produce bigger overshoot than others.
those closer to peak of sine(at about 170V for 120V grids) will fare better than those that produce more.

either way, measuring voltage on the caps tells us if they're overloaded or not.
different ups=different results.

as for values it seems to me peak for sine is rated value x 1.41(sq.root of 2), so for 120 this is about 170V and for 230V it's about 320V.
my measured values on caps (which is half the peak) of 157,160 and 156V confirm this.
so as starfury wisely notices, voltage on the caps should be same in both cases.
at 230V grid you have peak(320V), and at 120V grid you have peak(170V) doubled, which brings us to 320V or 340V "to chop".

in both cases each cap is taking half the load so about 160V-170V.
(grid voltage is not precise science at all...some european grids are closer to 240, some to 220, 120 standard seems to mean "100-127V" etc.
http://en.wikipedia.org/wiki/Mains_e...Voltage_levels )

with that out of our way, we still can't say for sure what 120V ups are doing.
if they're doing same thing as these 230V models (most probable, i think) then you have same net-effect, and that is overloading the caps.

for 120V grid that would mean the following:
to compensate for shorter pulses (they employ in this square configuration) they're increasing the amplitude, nearing the 200V peak(or going over it) and therefore endangering the caps in psus.

one thought: this is easily solved by putting 300V caps.
for active pfc psus that use one cap only there's no easy solution on the psu side.
dunno if active pfc design could be accomplished by usage of two caps in series(?).

Announcement

square output UPS vs. bulk capacitor voltage

square output UPS vs. bulk capacitor voltage

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment