Cycle dropout test in PSU reviews

[Travis]

Can we do some tricks on test equipments to simulate the sudden line drop and resuming and thus add a cycle dropout test to the testing methodology?

I mean, cut off (or step-down) the input electricity for precisely no more than 17ms(or 16ms), and see if the PSUs work properly, or just shut down, or die quickly.

[Spectre]

Yes you could test holdup time. Doubt anyone is going to include it because it is a yes/no thing for a being a bit of a PITA.

[Travis]

Maybe I'm just calling for a line-surge and line-drop test.

[Travis]

Yes you could test holdup time. Doubt anyone is going to include it because it is a yes/no thing for a being a bit of a PITA.
Not exactly the same as HOLD UP TIME test. Because in line-drop test, when the line voltage is cut off and then restored, there is an inrush current on the input side and ... some units may explode on the primary or secondary side, since the NTC resistor does no protection at that time and components aren't sufficiently rated.

By the way, what is "PITA" and what is "IIRC"?

[Spectre]

Maybe I'm just calling for a line-surge and line-drop test.

Maybe you should be clear what you are asking for because I already answered your question?

I already brown out during my tests.

[Travis]

Maybe you should be clear what you are asking for because I already answered your question?

I already brown out during my tests.

At first I thought I wanted to check if outputs were normal...
Then I realized the safety things are in fact much more important in this test.
So it's an off-on or low-high safety check, together with voltage readings.

It's neither a low-voltage input/line regulation test, nor a hold-up time test, but in fact a line drop test, which means the input voltage is cut off for 17ms, and then rush into the input of power supply, and a quality PSU must SURVIVE this test and also provide OUTPUTS WITHIN LIMITS.

[madmat]

Not exactly the same as HOLD UP TIME test. Because in line-drop test, when the line voltage is cut off and then restored, there is an inrush current on the input side and ... some units may explode on the primary or secondary side, since the NTC resistor does no protection at that time and components aren't sufficiently rated.

By the way, what is "PITA" and what is "IIRC"?

PITA = Pain In The Ass, IIRC = If I Remember Correctly. A few more; YMMV = Your Mileage May Vary, FTW = For Teh Win, ITMCBAYS = I Think My Cousin Bob Ate Your Snickers...

[Oklahoma Wolf]

Can we do some tricks on test equipments to simulate the line conditioner and thus add a cycle dropout test to the testing methodology?

I mean, cut off the input electricity for precisely no more than 17ms(or 16ms), and see if the PSUs work properly, or just shut down.

Can I? Probably. Will I? Nope.

Aside from the fact I'm not sure how to do it on the SM-268 without the PC interface and software, I get the feeling not a lot of people will care about it. Just like the PFC numbers - nobody really seems to be missing those.

[Travis]

Can I? Probably. Will I? Nope.

Aside from the fact I'm not sure how to do it on the SM-268 without the PC interface and software, I get the feeling not a lot of people will care about it. Just like the PFC numbers - nobody really seems to be missing those.
When it comes to the real-world usage of Active PFC, I think many people will get interested to see not only the line regulation numbers, but also if the unit w/ APFC works properly and don't explode on unstable lines. Imagine you bought a EA500, put it in your PC, played COD4 on it, and suddenly your friendly neighbourhood turned his air conditioner on, and then...
Your nice power supply didn't hold on as you expected, but shut off and died quietly, what would you think?

I think that's why I need to check this in reviews.

ps: A load tester can't carry out that test. A possible choice is the Chroma programmable AC source which Christoph is using. I'm thinking of other ways.

[Spectre]

At first I thought I wanted to check if outputs were normal...
Then I realized the safety things are in fact much more important in this test.
So it's not a brown out test, but an off-on safety check, together with voltage readings.

It's neither a low-voltage input/line regulation test, nor a hold-up time test, but in fact a line drop test, which means the input voltage is cut off for 17ms, and then rush into the input of power supply, and a quality PSU must SURVIVE this test and also provide OUTPUTS WITHIN LIMITS.

I am going to say this just one time. You need to STOP rapid fire posting and starting thinking and posting clearly. English may not be your first language but you are communicating in it and doing so poorly (or at least that has been you defense) as you continually change your questions when answers are provided and post completely off topic in threads.

Now, you asked:

I mean, cut off the input electricity for precisely no more than 17ms(or 16ms), and see if the PSUs work properly, or just shut down.

That is hold up. Your further expansion:

It's neither a low-voltage input/line regulation test, nor a hold-up time test, but in fact a line drop test, which means the input voltage is cut off for 17ms, and then rush into the input of power supply, and a quality PSU must SURVIVE this test and also provide OUTPUTS WITHIN LIMITS.

Still encompasses what hold up time and functionality is about. Given the spec is 16ms your explanition that you were meaning something else seems unfeasable at best.

[Spectre]

When it comes to the real-world usage of Active PFC, I think many people will get interested to see not only the line regulation numbers, but also if the unit w/ APFC works properly and don't explode on unstable lines. Imagine you bought a EA500, put it in your PC, played COD4 on it, and suddenly your friendly neighbourhood turned his air conditioner on, and then...
Your nice power supply didn't hold on as you expected, but shut off and died quietly, what would you think?

I think that's why I need to check this in reviews.

ps: A load tester can't carry out that test. A possible choise is the Chroma programmable AC source which Christoph is using. I'm thinking of other ways.


In the real world your neighbor turning on the AC doesnt cut your electricity.

In the real world the worst most people experience from large electrical draws would be a brown out. Which like I said is already covered.

[Travis]

Still encompasses what hold up time and functionality is about.

NOT EXACTLY hold-up time.

The hold-up time test is like this.

http://img.pconline.com.cn/images/bbs4/20085/6/1210073689906.png

No need to explain the graph, right?

The input is cut off and never turn on again, and the main capacitor keeps discharging. When the voltage on main capacitor starts to drop (not VERY fast), the PWM stage can increase the turn-on time and thus remain output voltage within -5% limit for a short period. When increasing turn-on time cannot help keep the output voltage, the output shows a sudden drop quickly to zero.

That's what happens when we encounter an electricity failure or pull out the input cord by mistake.


For the cycle dropout tests, things are a little different.

http://diy.yesky.com/cmsimagelist/2007/179/lcvmb5wk714a.jpg

The setup is entirely different from the hold-up time test or the line regulation test.
The line surge are suppressed by transcient suppress components like MOV and Zener Diodes and Gas Tubes, together with the EMI filter network.
The line drop however, is absorbed by the main capacitor. the cycle is 20ms for a 50Hz power line, so 17ms means the capacitor must remain its voltage during nearly a single cycle. Then the PSU is on again. When the voltage on main capacitor is significantly lower than line voltage, the capacitor is immediately charged and there may be a large amount of inrush current on the primary side, which may exceed the maximum rating of rectifying bridge. Because NTC resistor as the inrush limiter is still warm and not working on the input line, the components on the primary side must bear this sudden inrush. Also this inrush may influence the rectifier packs and caps on secondary side. With different inrush limiter like a power relay used on ET-850, things may be different, but I can't describe clearly right now.

So I'm emphasizing the "safety" more than the "hold-up time".

I've read thru your test method and I think you haven't covered that yet. You didn't mention the "brown out" test.
However maybe your line conditioner already has a sudden drop-out simulating function. I'm not sure.

I think I've described the technical reason and the difference clearly, otherwise please tell me why it's not important or how is it covered in other tests.

[Oklahoma Wolf]

A possible choice is the Chroma programmable AC source which Christoph is using.

That's not going to happen for this website - we're having enough trouble saving up for the SM-8800's. And really, I'm not about to go to the trouble of adding another test for just one or two people's benefit ;)

Just went through the SM-268 software - it can't do it even with the PC interface. IIRC the 8800 still can though. I don't have the software for it anymore since I figured I'd never need it until I got an 8800.

[Travis]

Yes there is difficulties describing theoritical things or expressing subtle meanings for a Chinese college student who has never been abroad or worked with foreign colleagues.

But I think I've made every effort to express my concerns with professional manner. I kept emphasis on not being a native speaker, only to remind you that there may be something missed in my post and if you pick it out I'll appreciate that. But why spend so much time on criticizing a newcomer not speaking well, instead of seriously giving your own view about the topic? If you say "A unit passing the holdup time test can surely pass this cycle dropout test so it's unnecessary", I'll be listening to see your explanations.

When first considering a cycle drop-out test I did think about voltage readings, that's in my first post. But then when I study the difference between an electricity failure, a brown-out and a low line voltage, and taking a dead unit cracked during a brown-out into consideration, I found there is relationship between this reliability and the ratings of components/the mechanic of inrush suppression. That's another question, but still not off that "cycle drop-out test" topic, right?

I know you hate flooding posts and you're sensitive to off-topic posts in a thread. But what I'm saying has certain meanings, though not described very clearly maybe. I don't write meaningless posts because I don't like that, and writing in English takes time. I believe I've shown enough respect to you and others so would you show your kindness and respect in turn?

[Travis]

That's not going to happen for this website - we're having enough trouble saving up for the SM-8800's. And really, I'm not about to go to the trouble of adding another test for just one or two people's benefit ;)

Just went through the SM-268 software - it can't do it even with the PC interface. IIRC the 8800 still can though. I don't have the software for it anymore since I figured I'd never need it until I got an 8800.

Chromas being able to do that doesn't mean there's no other way. There should be a technical solution, since we have MCUs and so many electrical components to make a transcient line-drop simulator.

A load tester alone cannot handle that. But if it's designed as an ATE, some tricks on AC controls may be done, cos an ATE can carry out a hold-up time test. So dig into the function descriptions of "hold-up time test" on SM-8800 ATE and see if there's a way to simulate a cycle drop-out.

Remember this is not for one or two's benefit. This is useful to all users who are using non-PFC units or APFC units. And we always do experiments before we carry out new test methods, am I right?

I do have a GreatWall ATX300-P4 unit died in a sudden line drop when it's pulling an AthlonXP platform and luckily mobo and HDDs weren't affected by that failure.

So I believe there is difference between quality and generic power supplies in this feature. Why don't we pick out disqualified or badly designed ones if we have this method available?

[Addition:
You just can't imagine how bad the electricity is in some rural area in China.
Maybe you don't have such problems so, just treat it as a check for "compatibility issues"]

[Oklahoma Wolf]

Remember this is not for one or two's benefit. This is useful to all users who are using non-PFC units or APFC units.

Not really. I'm here to review power supplies - not make sure they hold up to every single advertised spec. If I think a test is redundant or unnecessary, I'll leave it out unless many people ask for it. So far, you're the only one who's asked for this one since I started reviewing last year.

Why don't we pick out disqualified or badly designed ones if we have this method available?

The autopsy phase already performs this function. Again, I dislike needless redundancy. I have to review one unit a week if I have any chance of catching up on the backlog, and adding new tests isn't the best way to go about catching up. I barely have enough time to post on the forums anymore as it is between my review work and my novel writing.

[Travis]

Not really. I'm here to review power supplies - not make sure they hold up to every single advertised spec. If I think a test is redundant or unnecessary, I'll leave it out unless many people ask for it. So far, you're the only one who's asked for this one since I started reviewing last year.



The autopsy phase already performs this function. Again, I dislike needless redundancy. I have to review one unit a week if I have any chance of catching up on the backlog, and adding new tests isn't the best way to go about catching up. I barely have enough time to post on the forums anymore as it is between my review work and my novel writing.
I still think it's important, considering half the units I'll review are gonna be cheaper than Earthwatts and FSP GreenPS.

So let me clarify: I'm asking for ideas that may help me design this system, and may also be useful to other reviewers who are not that busy. I'm not requesting others to add redundancy test in their works.

I'm not saying "JonnyGuru site should include this test".

What I'm asking in the first post is "are we possible to do this using equipments at hand or with some modding?". If the answer is yes, I'll consider doing this myself.

I'll go out and check some equivalent ways. Autopsy phase do reveal bad builds, but still cannot check if components are overrated in the worst case. That's a challenge and I think I should figure out.
[I even misunderstood "autopsy", damn!]

[jonnyGURU]

I know you hate flooding posts and you're sensitive to off-topic posts in a thread. But what I'm saying has certain meanings, though not described very clearly maybe. I don't write meaningless posts because I don't like that, and writing in English takes time. I believe I've shown enough respect to you and others so would you show your kindness and respect in turn?

I know your posts are supposed to have direction and meaning and this is why we've let you bump so many old threads without much alarm. I do appreciate your eagerness and the knowledge you already posess. That said, I have to agree with Spectre in that if you slow down and stop "rapid-posting" (a new post every 5 minutes) you can work on better communicating what you're trying to convey.

There isn't a race here. Nobody gets bonus points for having the most number of posts. Just take a deep breath, take your time thinking about what you're trying to say in your post. Cool?

[Travis]

I know your posts are supposed to have direction and meaning and this is why we've let you bump so many old threads without much alarm. I do appreciate your eagerness and the knowledge you already posess. That said, I have to agree with Spectre in that if you slow down and stop "rapid-posting" (a new post every 5 minutes) you can work on better communicating what you're trying to convey.

There isn't a race here. Nobody gets bonus points for having the most number of posts. Just take a deep breath, take your time thinking about what you're trying to say in your post. Cool?
Yeah I appreciate your tolerance. I'll improve my expressions but... don't expect too much on me since language is a bit of problem to me right now :(

Next time I'll write my post in a txt or pdf file and paste it as an attachment. Hopefully this may at least make this forum look cleaner. If I have new ideas, I'll just add a new attachment and say it's updated.

[Spectre]

Yeah I appreciate your tolerance. I'll improve my expressions but... don't expect too much on me since language is a bit of problem to me right now :(

Next time I'll write my post in a txt or pdf file and paste it as an attachment. Hopefully this may at least make this forum look cleaner. If I have new ideas, I'll just add a new attachment and say it's updated.

Posting in an attachment is not going to help. What is going to help is not bumping random threads with off topic posts, and not asking a question and when that question is answered then saying that isn't what you were asking when it was. You see? Stay on topic and stay precise the first time.