[Question] UPS compatibiliy with Active PFC PSU

[Black Ninja]

Since price difference between pure sin and non sin UPSes is not that big , and there is no trade of when you go to pure sin...

Why don't we just assume we all have pure sine UPSes, and concentrate on the next issues like hold-up time of power supplies or any other issues that could come up so we don't have to go next step and buy Online UPS which is 8 time more expansive and very inefficient compared to non online.

 

[Bidule0hm]

The switching time doesn't change (at least on a correctly designed UPS...) as it's simply the switching time of the relay (usually 5 to 15 ms) :)

Yep, sadly "made in china" well sums up the actual world I think...

Yeah, but I bought this UPS used for almost nothing so I'll not complain (too much) :P

 

[Black Ninja]

The switching time doesn't change (at least on a correctly designed UPS...) as it's simply the switching time of the relay (usually 5 to 15 ms) :)

Yep, sadly "made in china" well sums up the actual world I think...

Yeah, but I bought this UPS used for almost nothing so I'll not complain (too much) :p

This time I am afraid I am going to have to disagree on 2 of 3 in the same order::)

1. If APC says in their white paper that switching time change or varies base on voltage variations, I will take THEIR word for it. You know APC is the biggest(best) manufacturer of UPSes and who knows better than the people who made it.:p

2.There is nothing wrong that is "made in china". People complaint about low quality , but in the same time don't want to pay dime for quality. China only makes what most people want - cheap, It's not that they can't make the best products, they just can't make something great out of nothing.

3. Here I agree with you: When you pay almost nothing you can't complain.:D

 

[Bidule0hm]

1) Ok, so now I'm curious to know the reason behind the variation :P

2) Most people, not everyone :) With time I learned (and I'm not old...) that it's better to buy one good thing than 10 crappy things that are 10 times cheaper (and it's valid for pretty much everything) because in the end a cheap crappy thing can damage other things, if it's a tool you'll have a poorer result even if you do everything else correctly, etc... It's why if I can make the thing myself I'll make it (even if it's more expensive) because I'll be sure of the quality, I can make exactly what I want for my needs, I can easily repair or modify it because I know exactly how it works, etc. ;)

 

[Tywin]

With time I learned (and I'm not old...) that it's better to buy one good thing than 10 crappy things that are 10 times cheaper (and it's valid for pretty much everything) because in the end a cheap crappy thing can damage other things, if it's a tool you'll have a poorer result even if you do everything else correctly, etc...

Getting off-topic now, but 100% agree with this; I am frequently annoyed by the "race to the bottom" to make the cheapest, lowest-common-denominator products. Be thankful that when it comes to computer components there is still good-quality stuff to be had. There are other areas of consumer goods where the only two levels are bargain-basement crud and gold-plated diamond-encrusted luxuries.

The way I typically phrase this effect is "you really do get what you pay for, and it's usually not even linear".

 

[Bidule0hm]

Yep, and don't get me started on the planned obsolescence thing...

I think most people don't "want" cheap things or things who broke 2 years later, they just buy whatever there is right in front of them because they don't even bother searching a bit before if there is a better product (even at the same price...). They are just assisted and lazy. After all it's their choice, but the problem is that because of that it gets harder and harder for those who want good quality things to find them.

 

[Black Ninja]

Getting off-topic now, but 100% agree with this; I am frequently annoyed by the "race to the bottom" to make the cheapest, lowest-common-denominator products. Be thankful that when it comes to computer components there is still good-quality stuff to be had. There are other areas of consumer goods where the only two levels are bargain-basement crud and gold-plated diamond-encrusted luxuries.

The way I typically phrase this effect is "you really do get what you pay for, and it's usually not even linear".

Even in computers the same rules applies unfortunately. Look what happened to the most reliable Intel SSDs. I bought tons of them ?only intel" now they when to the sewer and disappear.. for now. I am talking about consumer grade.

 

[Black Ninja]

1) Ok, so now I'm curious to know the reason behind the variation :p

2) Most people, not everyone :) With time I learned (and I'm not old...) that it's better to buy one good thing than 10 crappy things that are 10 times cheaper (and it's valid for pretty much everything) because in the end a cheap crappy thing can damage other things, if it's a tool you'll have a poorer result even if you do everything else correctly, etc... It's why if I can make the thing myself I'll make it (even if it's more expensive) because I'll be sure of the quality, I can make exactly what I want for my needs, I can easily repair or modify it because I know exactly how it works, etc. ;)

1) I wish I knew the exact reason. It seems when it comes to electronics you more knowledgeable than me. You are the guy with the oscilloscope , I am the guy with a multimeter. )))) What I remember was: example if voltage is 126V-106V it UPS will switch in 4 ms , if it's 96V-132V range it will take 8ms, something like that on a particular APC model. I think this was FAQ with a with number that apc representative gave me to read about switch times. I'll get you the link if i can came across.

2) That's exactly what I said. Not everyone , but most people. I always try to buy the best if I can afford it , not matter if it's CPU, car coolant, or toothpaste. I can't buy best , house , car because ... sky is the limit there, but anything I can afford - I can't settle for anything less than the best.
I am afraid most people in this case is almost all people.They were not enough people to keep Pioneer to keep making the best TVs, there were more expensive but the people who like quality will not hesitate to buy them , it seem still not enough of them to even keep the doors open.

 

[Black Ninja]

Yep, and don't get me started on the planned obsolescence thing...

I think most people don't "want" cheap things or things who broke 2 years later, they just buy whatever there is right in front of them because they don't even bother searching a bit before if there is a better product (even at the same price...). They are just assisted and lazy. After all it's their choice, but the problem is that because of that it gets harder and harder for those who want good quality things to find them.

You absolutely right. I think they just buy always the cheapest because of ignorance. They can't make the difference and just buying crap thinking they get great deal - the same thing for less. Not researching enough like you said , and many times end up buy overpraised junk instead of better product they can get for less if they knew: what and where.

 

[Bidule0hm]

1) This is very weird, I'll read some doc on that if I can found some, thanks ;)

Yeah, it's more and more often I'm surprised by the lack of quality of things and/or the design efforts that are made to render a thing non-repairable by yourself (apple anyone? ...).

The trick I use is to buy old products from the 90's when it's possible (mainly tools for electronics), generally it's more durable even if it's older, and even if it breaks at least it's repairable. The other trick is to buy used things (even recent ones) that are very expensive when new but not when used, and that are far better quality than the new things at the same price :)

Yep, before people would know (at least partly) how the thing they buy works and what is the real value of the thing, now pretty much everything is sold thanks to marketing bullshit (A/V systems anyone?) because people doesn't take the time to just learn, compare, think, ... "You've been told this thing is great, then you buy it without checking if it's true or not".

 

[Black Ninja]

1) This is very weird, I'll read some doc on that if I can found some, thanks ;)

Yeah, it's more and more often I'm surprised by the lack of quality of things and/or the design efforts that are made to render a thing non-repairable by yourself (apple anyone? ...).

The trick I use is to buy old products from the 90's when it's possible (mainly tools for electronics), generally it's more durable even if it's older, and even if it breaks at least it's repairable. The other trick is to buy used things (even recent ones) that are very expensive when new but not when used, and that are far better quality than the new things at the same price :)

Yep, before people would know (at least partly) how the thing they buy works and what is the real value of the thing, now pretty much everything is sold thanks to marketing bullshit (A/V systems anyone?) because people doesn't take the time to just learn, compare, think, ... "You've been told this thing is great, then you buy it without checking if it's true or not".

1) It took me awhile but I knew person like you will appreciate the material. Here is what you looking for: go to this link www.apc.com/site/support/index.cfm/faq/ and then put this ID in search FAQ bar FA156514
This explains the switch time in different situations.

2) You are absolutely right. I think this also applies for old computer . Good old sever with server grade components will run better and longer that latest cute little consumer desktop PC from HP for example. Only draw back is power consumption and noise.

A/V systems ? Well marketing and advertisement is what sell them , not quality of the product. Pretty much what you said.:)

 

[Bidule0hm]

1) Many thanks for the search ;) but unfortunately it doesn't explain why the switch time is longer. But I can guess that they don't use a comparator, they must be doing this in software and use some averaging at some point, so the bigger the change the longer it takes to be detected.

2) Yep, but sadly the majority of today's developers don't write (even barely) optimized code, mainly because of economy reasons (writing good code takes more time and time is money... Yeah, I know, it's stupid because on the long-term you ends spending more). It's like "bwarf, I can use 500 MB of RAM to display one tab and 3 buttons, today's computers have many GB of RAM, no big deal...", well, just a bit exaggerated, but take this case: "well, no need to free those 2 KB, it's nothing..." that's a leaky program, let it run one week (let alone one month if it's for a server) and you'll see the result... You'll say "that's why we invented the garbage collector" well, java has a GC...

Programs take more and more RAM to do the same things than before, that's why old computer hardware is less and less useful. Don't trust me? well, look at the Amiga: it takes 8 seconds to boot (from an old IDE drive, only 3 or 4 secs more from a floppy disk...), it's graphically equivalent to Windows 95, it use preemptive multitasking (like Windows and other modern OS), there is just no lag when you click on something, and guess what? it runs on a 14 MHz CPU (and I'm generous, the Amiga 500 use only a 7 MHz CPU...) and uses less than 300 KB of RAM. Ah, and it still boot 20+ years later... :D

 

[rogerh]

Even if they are using a simple comparator, with a standard 50/60Hz sine wave voltage as the reference, if the limits are set wider it will take longer during each quarter cycle before the limit is reached.

 

[Bidule0hm]

No because the cycle will be the same length, it's the dV/dt who will be higher.

 

[rogerh]

No because the cycle will be the same length, it's the dV/dt who will be higher.

The wider range won't affect the dV/dt of the actual signal, but if the limits of what is acceptable are narrower the signal will cross the limit line sooner. The 2-4 ms time quoted for the most sensitive setting is within a quarter of a cycle. They have not mentioned it, but it is probable that switching time is quicker if the mains voltage vanishes, rather than just falling below the set lower limit, but then it has a quicker impact on the PSU if it goes to zero rather than just dropping too low. I imagine there is some integration involved rather than a straightforward voltage comparator, let alone a rate of change response.

 

[Bidule0hm]

"The wider range won't affect the dV/dt of the actual signal" No, but a higher voltage signal will be at the same frequency, therefore the same maximum time to reach the 1/4 cycle peak (and a higher dV/dt) :)

 

[rogerh]

I'm not going to prolong this OT discussion beyond this one post! The point I am making is that the range of acceptable values does NOT affect the signal, which in my case chugs along at about 251V 50Hz regardless of limits set by my UPS. What a wider acceptable range does is prolong the time before an anomaly in the signal (for instance suddenly going to 0V which is the commonest anomaly) is recognised, because, whether dV/dt, V, or integral VxI over a half cycle is measured, it takes longer to reach the set limit because the set limits are wider.


Also note that V is 0V twice in cycle, dV/dt is 0V/s twice in a cycle, so I vote for using some integral function.

 

[Bidule0hm]

My point is: without averaging, integral, whatever other things, the time taken by a 251 V signal to be detected over a 250 V threshold is the same as a 301 V signal and a 300 V threshold. So my conclusion is: they must be using software and they're adding some averaging on a rectified signal, or doing some integral on the AC signal, or whatever formula, and this add some non-constant delay, when a simple comparator on the rectified signal would be better but probably $0.05 more expensive... :)

 

[rogerh]

My point is: without averaging, integral, whatever other things, the time taken by a 251 V signal to be detected over a 250 V threshold is the same as a 301 V signal and a 300 V threshold. So my conclusion is: they must be using software and they're adding some averaging on a rectified signal, or doing some integral on the AC signal, or whatever formula, and this add some non-constant delay, when a simple comparator on the rectified signal would be better but probably $0.05 more expensive... :)

True. But normally the signal won't creep up to 300v before going over the threshold. If that was a common case you would be right. But in real life it goes along at 251V, occasionally shooting up to 295V and back to normal, then up to 300V and back to normal, and the UPS has to catch it when next time it spikes up to 301V. If the common case was a gradual rise from an excessively high (or low) baseline then there would be no reason why the UPS should not be just as fast at wide settings as at narrow ones. The UPS manufacturers specify for the 90% of users where their power is generally running in the correct government specified range but has spikes and brownouts.

This fact accounts entirely for our difference of opinion!

Edit: putting it more clearly and looking at loss of power, the main hazard for a UPS; the UPS has to detect a drop to 0V as quickly as possible. Assume the voltage is normally 251V and it is on narrow range, it can detect a fall to 210V as definite loss of power. But on wide range it has to wait until the voltage drops to say 175V. The point of the wider range is to avoid cutting out on routine brownouts and spikes if they are prevalent, but still detect a real loss of power ASAP. And a wider range makes the common case (normal voltage goes instantly to zero) harder to detect.

 

[Bidule0hm]

Yeah, I think the point is: power companies sell us poor quality power. In the last three months my UPS has been used 5 times (AFAIK, it's probably more than that) --> one time because of user tripping the main breaker (so that doesn't count), one time because of power outage and 3 times because of under/over voltage.

And when I mesured the mains with an oscilloscope I couldn't believe my eyes and I though it was an user error; look at this waveform, I always though the mains was a pretty much perfect sinewave...:

Voltage is in yellow, current in blue (just replace mV by mA) and power in purple (replace V² by W), this is on a purely resistive load (the current isn't perfectly in sync because of the current probe I quickly built with a transformer, I was calibrating it). This is not a correct sinewave, I don't know what is the max allowed THD in my country but it seems to be pretty high...