Author Topic: Computer power supplies - current limiting  (Read 9561 times)

0 Members and 1 Guest are viewing this topic.

Offline GeoffSTopic starter

  • Supporter
  • ****
  • Posts: 1272
  • Country: au
Computer power supplies - current limiting
« on: June 06, 2010, 01:27:01 am »
Yesterday I had the power supply in my server failed - smoke, blown mains breaker etc. Put in a spare supply I had and same thing.
The first supply was about 6 years old and ran 24/7. The second one was similar vintage but has sat in the cupboard for 12 months unused.

My question is about current limiting in modern switching supplies. Assuming the motherboard has a fault that presents as a short to one or more of the outputs of the supply, what should happen? Will the current limiting shutdown the supply and limit any damage? I don't want to put in a nw supply and have it die as well.

I'm ready to believe that given the age of the supplies, having them both give up was just one of those things.
Sorry if this question is only marginally engineering related.  :)
 

Offline Kiriakos-GR

  • Super Contributor
  • ***
  • !
  • Posts: 3525
  • Country: gr
  • User is banned.
    • Honda AX-1 rebuild
Re: Computer power supplies - current limiting
« Reply #1 on: June 06, 2010, 01:34:41 am »
At Computer power supplies , brand and specs walks together ..

Unbranded designs , usually have fake labels ..

My rule are to always get PSU's with active PFC + cost over 120$ or EUR  .

At this price range , you can be demanding about stability - protection - longevity.
 

Offline GeoffSTopic starter

  • Supporter
  • ****
  • Posts: 1272
  • Country: au
Re: Computer power supplies - current limiting
« Reply #2 on: June 06, 2010, 01:49:23 am »
The 2 power supplies in question were not cheap no name brands. ( Tagan and Enermax )
I just ordered a new Tagen 700 watt supply based on the fact the the last one kept going for nearly 6 years which I'm sure is well beyond the designed life.
 

Offline charliex

  • Frequent Contributor
  • **
  • Posts: 336
  • Country: 00
  • Car Hacker
Re: Computer power supplies - current limiting
« Reply #3 on: June 06, 2010, 04:30:32 am »
even our $2000+ servers with fail safe psu's will fry the motherboard on a fault, since they can deliver 1600W each.
 

Offline GeoffSTopic starter

  • Supporter
  • ****
  • Posts: 1272
  • Country: au
Re: Computer power supplies - current limiting
« Reply #4 on: June 06, 2010, 04:47:17 am »
The servers at work are replaced on a 3 year cycle so I can only remember a handful of power supply failures.
I guess the clean, airconditioned environment helps. When I opened the power supplies up, it was impossible to identify any component for the greasy, black dust that covers everything.

If it turns out that the motherboard has died it's a good excuse to upgrade. Who doesn't like a new toy?  ;)
A 5 years old server should be retired anyway.
 

Offline charliex

  • Frequent Contributor
  • **
  • Posts: 336
  • Country: 00
  • Car Hacker
Re: Computer power supplies - current limiting
« Reply #5 on: June 06, 2010, 05:31:34 am »
I had one fry OOB last week.. not fun, big old smoking hole left behind :)
 

Offline GeoffSTopic starter

  • Supporter
  • ****
  • Posts: 1272
  • Country: au
Re: Computer power supplies - current limiting
« Reply #6 on: June 06, 2010, 05:40:10 am »
I had one fry OOB last week.. not fun, big old smoking hole left behind :)

You're just trying to cheer me up now :'(
 

Offline cybergibbons

  • Frequent Contributor
  • **
  • Posts: 400
Re: Computer power supplies - current limiting
« Reply #7 on: June 06, 2010, 06:36:20 am »
I used to use PC power supplies as general purpose power supplies. They used to have some protection in them at least - if you shorted a rail it would shut down rather than blow up.

You could draw stupid amounts of current from the 3.3V rails though, way over the nameplate rating. The regulation got pretty bad though.
 

Offline Kiriakos-GR

  • Super Contributor
  • ***
  • !
  • Posts: 3525
  • Country: gr
  • User is banned.
    • Honda AX-1 rebuild
Re: Computer power supplies - current limiting
« Reply #8 on: June 06, 2010, 07:58:37 am »
I had one fry OOB last week.. not fun, big old smoking hole left behind :)

You're just trying to cheer me up now :'(

No I will cheer you up ..  You will get  one motorized air-compressor ,
and you will start  to schedule  maintenance cycles every 6 months ... about keeping clean,
the PSU - any fan on the box , and any heatsink .. 

There is no foolproof devices out there , only smart people !!     
 

alm

  • Guest
Re: Computer power supplies - current limiting
« Reply #9 on: June 06, 2010, 09:00:55 am »
Most quality power supplies have over current protection, but it will usually only kick in with a dead short. Computer supplies are designed for high current draw, so if there's just a little resistance there, it might just fulfill its secondary duty as arc welder :). The over current protection is usually set significantly above the rated current, since you don't want it to kick in on some peak (over current protection triggered = server down).

No I will cheer you up ..  You will get  one motorized air-compressor ,
and you will start  to schedule  maintenance cycles every 6 months ... about keeping clean,
the PSU - any fan on the box , and any heatsink ..
Too much dust will prevent proper cooling, but no need to get obsessive about it, a proper design should tolerate some dust.

There is no foolproof devices out there , only smart people !!     
How true.
 

Offline Kiriakos-GR

  • Super Contributor
  • ***
  • !
  • Posts: 3525
  • Country: gr
  • User is banned.
    • Honda AX-1 rebuild
Re: Computer power supplies - current limiting
« Reply #10 on: June 06, 2010, 09:29:16 am »

No I will cheer you up ..  You will get  one motorized air-compressor ,
and you will start  to schedule  maintenance cycles every 6 months ... about keeping clean,
the PSU - any fan on the box , and any heatsink ..
Too much dust will prevent proper cooling, but no need to get obsessive about it, a proper design should tolerate some dust.


I do this task every six months ,  and common dust , its not a major enemy , but we forget that we have pets , or in the shops like my , I can not keep the door closed .. open door = open shop .
Plus room ventilation ..

At industrial environments , I have see the worst ,  some that does wood work had stuffed computers with particles of wood ..... the ones who works by handling iron had  black particles from the plasma cutter all over the place ..

Even so , the computers should be inspected every six months , at list ..

Because of this thread , I got motivated to do maintenance in my own system.
I had it on schedule  , but I did it now for the peace of the mind ..   ;)

   
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8998
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: Computer power supplies - current limiting
« Reply #11 on: June 06, 2010, 01:19:04 pm »
Quote
My rule are to always get PSU's with active PFC + cost over 120$ or EUR  .
Not sure about Australia but in the US, active PFC is a waste of money for residential use and some commercial use.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline charliex

  • Frequent Contributor
  • **
  • Posts: 336
  • Country: 00
  • Car Hacker
Re: Computer power supplies - current limiting
« Reply #12 on: June 06, 2010, 03:39:11 pm »
We'll we install our servers in climate controlled clean rooms, they don't see much in the way of pet hair and dust, or vacuum cleaners.

Talk about cheering up, that was the replacement server for one that went bad a few days before hand (Which had suffered a bit of abuse since we were swapping lots of video cards in  and out ) https://www.eevblog.com/forum/index.php?topic=507.0 that was not a fun week, but the new replacements are all up and running again :)

but yeah given the current even a desktop psu can shift there's little you can do to save it.
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 17802
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Computer power supplies - current limiting
« Reply #13 on: June 06, 2010, 06:55:23 pm »
I take my pc to work roughly every 6 months and put the compressor to it, does a wonderful job, every month i may open is and show it the vacuum cleaner but only for superficial stuff

now I'm moving i intend not putting it on the floor again that might help. one thing to remember is that most stuff is temp controlled now on the fans so the hotter it runs (more dust) the faster and noisier the fans will run and i hate fan noise to the point that in very hot weather I underclock and under volt my processor to reduce consumption and noise
 

Offline cybergibbons

  • Frequent Contributor
  • **
  • Posts: 400
Re: Computer power supplies - current limiting
« Reply #14 on: June 06, 2010, 08:41:14 pm »
I take my pc to work roughly every 6 months and put the compressor to it, does a wonderful job, every month i may open is and show it the vacuum cleaner but only for superficial stuff

One thing - always be careful with compressed air and electronics, the two sometimes don't mix...

I used to work on ships as an engineer, and we always had several PCs in the control room for looking up manuals and service guides. But then we ended up printing out the relevant pages and taking them down to the workshop. Massive waste of paper, annoying if you forgot a page, and a lot of the diagrams just didn't have the required resolution on A4 paper.

The PCs were normal Dells and replaced every 18 months as a matter of course as they rarely made it to 2 years with the levels of vibration.

So we took one of them down to the workshop to use for manuals.... we fitted a couple of bits of filter wool over the fans to keep most of the crud out. But one day the machine kept on restarting... so we opened it up, and it was covered in a fine mist of probably lube, diesel and hydraulic oil. Normal method to clean this kind of mess up would be the blow gun, which someone decided to use, despite the rest of us advising caution. It may or may not have been the chief engineer.

Needless to say, the PC didn't make it out alive. The first thing that signalled an issue was the "wheeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee" noise as the CPU fan was brought up to about 100,000rpm before the bearing failed. Next, some of the electrolytics were peeled off the board. Best yet, and we only found this afterwards, is that some of the oil mist ended up in the hard drive casing - I think the solvents and air pressure were enough to get into the casing past one of the sticker.

But better still was on another ship. The RPM and torque output (and hence power) of the main engine is measured with a pair of optical tachometers on the prop shaft - it detects how much the 1.2m diameter prop shaft distorts. We were woken in the middle of the night by an alarm which was a fairly worrying "main engine slowdown" (these are normally caused by an oil mist detector, warning of the possibility of a crankcase explosion).

Power meter:


Prop shaft:


On heading down, we found the issue was that one of the two tachometers had failed, and this had caused the indicated power output to jump to several hundred MW, way above the normal maximum.

They are similar to the break beam detectors on many tachometers - a disk with many cuts in it, and a pair of IR break beams, set up so the direction of rotation can be detected. We figured it was just crud blocking the beam, but couldn't work out a safe way of removing and cleaning the sensor with the engine running.

So some bright spark, who may or may not have been the first engineer, decided to use compressed air. Now, I don't know exactly what happened here, but I think that the very thin slit disk was deformed by the air, causing it to impact with the sensor. Needless to say, the 80MW output of the engine was more than a match for the sensor and disc, which ended up wrapped around the prop shaft, ripping about 15m of assorted cabling from a loom before something finally snapped. Quite a few cables were severed or shorted.

The first thing to go out was the engine room lighting were we were, which I think was our first indication that things were about to get interesting. Then practically every single alarm was indicated (bar the CO2 flooding system, which is on an entirely different circuit it would seem).


It was about then that we felt the first shudder of the main engine going into an emergency shutdown (i.e. stopping as quickly as it can). It turns out that this was because the same loom had several of the main crank bearing temperature sensors on it.

Anyway, long story short... we also managed to short half of the PLC power supply system, which meant that not only did the main engine shut down, but none of the generators started up. We suffered a total blackout and spent the next 45 minutes trying to get one of the five generators back on line. It was a good few hours before we had the main engine back up, and had to work watches because we'd lost so many engine room alarms.

Different life as a software engineer now, it has to be said. Sorry for going a bit off-topic, but just needs to be said, always think before using compressed air to clean something!

Some pics of the ships:
http://www.flickr.com/photos/cybergibbons/sets/72157594324779107/
http://www.flickr.com/photos/cybergibbons/sets/72157600000178587/



« Last Edit: June 06, 2010, 08:50:07 pm by cybergibbons »
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 17802
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Computer power supplies - current limiting
« Reply #15 on: June 06, 2010, 08:51:09 pm »
well our system is about 70 psi and i stand a few feet away and let the "wind" do the job, I never get too close, the second story is interesting
 

Offline Kiriakos-GR

  • Super Contributor
  • ***
  • !
  • Posts: 3525
  • Country: gr
  • User is banned.
    • Honda AX-1 rebuild
Re: Computer power supplies - current limiting
« Reply #16 on: June 07, 2010, 12:03:08 am »
I have this Italian hero ( Bologna) .... makes lots of noise when loads with air , but never stopped working since 1980.
12 Bar =174 PSI    140 liters ..

(needs a bit of cleaning , but I do check the oil level in it (lubrication) , more often than the cleaning)  ;D
« Last Edit: June 07, 2010, 12:07:41 am by Kiriakos-GR »
 

Offline tecman

  • Frequent Contributor
  • **
  • Posts: 444
  • Country: us
Re: Computer power supplies - current limiting
« Reply #17 on: June 07, 2010, 12:29:04 am »
cybergibbons... very cool pix

paul
 

Offline GeoffSTopic starter

  • Supporter
  • ****
  • Posts: 1272
  • Country: au
Re: Computer power supplies - current limiting
« Reply #18 on: June 07, 2010, 06:19:33 am »
cybergibbons... very cool pix

paul

I agree! Bigger than anything I've ever worked on (aircraft)

FWIW, . My new power supply is fitted and ticking along. Looks as if the two failures were just due to age.

One other dust/age problem is the fans. After being turned off for 24 hours they made a hell of a racket for the first 10 minutes of operations.
I know how they feel, I complain loudly the first time I have to work after a break.
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 17802
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Computer power supplies - current limiting
« Reply #19 on: June 07, 2010, 06:47:43 am »
yea my work pc fan does that on a cold winters morning
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf