PC power supply: failed in weird way

Thread Starter

jeka616

Joined Jan 14, 2009
131
Good day all.

I've been using PC power supply to power up my electronics, and when today I added battery back up to the system, it failed.

Want to mention one stange fact: as we now, there are about 10 of +12V wires (yellow usually) coming out of the supply, and I used 3-4 of them to power up several parts of circuit. Today, I've connected one more yellow wire to Lead Acid Battery through the relay, so when mains supply fail, the circuit would've been connected to battery. The battery voltage was ~ 12.6V at the time of connection....

after 3-4 mins working, the one single yellow wire which I connected from power supply to battery FRIED, like fuse...the rest of yellow wires were ok...


so question is, can 12.6V cause failure? the supply now works about 50%, i.e. when I connect a load to it, 1st time it fails, 2nd time powers the load, 3rd time fails etc....looks like some protection circuit is functioning..

it is 11pm already, and I'm to tired to think deep....appreciate your feed back.
 

Thread Starter

jeka616

Joined Jan 14, 2009
131
Thanks SGt. Can you summarize in 4-5 words the reason? And should I decide to repair it, where to start? (actually it is about 19$, so I will probably wont spend time fixing it, just will buy new one....)

It was obvious that something went wrong after introducing the battery, and now after your words I've gained some confidence in my newly built project, which was functioning ok for the last 3 days.

on the other hand, I've seen on video, how a guy uses PSU for charging battery, however it was going through an electronic battery cycler. So thats why maybe his was succesful...

as always, I'm finding answer in this great forum....
 

SgtWookie

Joined Jul 17, 2007
22,230
If you want to use the battery as a back-up to the 12v, you'll need to isolate the battery from the PSU using a diode, with the anode towards the PSU's 12v supply.
You will also need a way to keep your battery charged. The float voltage on a Sealed Lead-Acid battery is typically around 13.6v @ 25°C.

The internal construction of PC power supplies varies greatly, from manufacturer to manufacturer and even between a mfgr's various models. I have an old Compaq 250W ATX form factor supply that I modified into a bench supply. The +12v regulator is a 7812 linear regulator! Normally, the regulator will drain the output cap when power is turned off, but if you're powering the output with a battery, it likely won't survive.

12v lead-acid batteries will rapidly die if you keep them at 12v or less. They need to be maintained at ~13.4-13.7v.
 

Thread Starter

jeka616

Joined Jan 14, 2009
131
I've been thinking to put LM317, and a separate mini transformer just to keep the battery as you mentioned, around 13.5V. And this will go through the relay, which will isolate the battery from PSU when mains fail, and power up the circuit.

And after reading your notes, I just thought should've done it, instead of trying to kill 2 birds with single shot, that is to feed the circuit and charge the battery.....

thanks for feedback. I will buy new power supply tomorrow and fix the circuit.
 

sheldons

Joined Oct 26, 2011
613
firstly a power supply of this type is not at all mains isolated as it it a switched mode supply,connecting your battery to it obviously has killed it.
anything you connect to the secondary side cannot be grounded in any way or there will be big bangs and lots of smoke invoved...if you are lucky all you may have done is simple diode damage/capacitor damage maybe....you may also have a problem with the feedback side of things too as the supply seems to be tripping when you start it up.lets see a pic of the supply and any model/type nrs there are....
 

Adjuster

Joined Dec 26, 2010
2,148
firstly a power supply of this type is not at all mains isolated as it it a switched mode supply,connecting your battery to it obviously has killed it.
anything you connect to the secondary side cannot be grounded in any way or there will be big bangs and lots of smoke invoved...if you are lucky all you may have done is simple diode damage/capacitor damage maybe....you may also have a problem with the feedback side of things too as the supply seems to be tripping when you start it up.lets see a pic of the supply and any model/type nrs there are....
Clearly the OPs arrangement connecting of a battery straight to a "12V" supply was unsatisfactory. That said, his power supply may or may not achieve mains isolation, but you are in no position to make categorical statements on the matter.

Computer SMPS do achieve effective mains isolation, and are generally required to do so in order to be legally sold for use in PCs. If it were otherwise, all computer peripherals and attachments such as headsets could become live to mains, which would be terribly dangerous.
 

bountyhunter

Joined Sep 7, 2009
2,512
Clearly the OPs arrangement connecting of a battery straight to a "12V" supply was unsatisfactory. That said, his power supply may or may not achieve mains isolation, but you are in no position to make categorical statements on the matter.

Computer SMPS do achieve effective mains isolation, and are generally required to do so in order to be legally sold for use in PCs. If it were otherwise, all computer peripherals and attachments such as headsets could become live to mains, which would be terribly dangerous.
Correct, AC line isolation is usually required in all consumer devices. They typically do it with a half bridge offline switcher. The transformer isolates AC side to secondary DC side. They typically use opto-coupler feedback across the isolation barrier.
 

Wendy

Joined Mar 24, 2008
23,415
Just curious, where is your sources on this? I've opened up more than one PC power supply, there are no transformers, only toroids, diodes, and FETs.
 

SgtWookie

Joined Jul 17, 2007
22,230
Bill,
Bountyhunter is an expert on this stuff. He worked in the power electronics section of a major electronic component manufacturer for many years.

I don't know a fraction of what he knows about this stuff, but I can tell you that the PC supplies DO have galvanic isolation from the mains.

Take a look at the datasheet for the UC2842; on page 8 you'll see an offline flyback current-mode regulator. Notice that there is no electrical connection from the mains side to the secondary side.

That schematic is lacking something, which is a feedback from either the 5v or the 12v back to the controller. For the purpose of isolation, feedback is usually via an optocoupler.

[eta]
Check out this site:
http://danyk.wz.cz/s_atx_en.html
LOTS of schematics for AT and ATX form factor PC supplies. It's a good site to bookmark.
 
Last edited:

Evil Lurker

Joined Aug 25, 2011
116
Just curious, where is your sources on this? I've opened up more than one PC power supply, there are no transformers, only toroids, diodes, and FETs.
All PC power supplies will have one to three transformers (usually right smack dab in the center of the PCB) depending on the topolgy used. In fact I would love to see a standard ATX supply that doesn't have a transformer.

My guess is what happened was the protection IC detected an overvoltage and shut down the SMPS and the battery began backfeeding into the PSU.

Simply put, there is no way you can hook a 12v lead acid battery to a standard computer power supply without first tweaking the resistor divider network on the secondary side to increase the duty cycle of the SMPS controller on the primary side, and even then 13.6 or whatever volts may be a stretch, especially if it has any zener diodes on the secondary used to detect over voltage conditions.
 

Adjuster

Joined Dec 26, 2010
2,148
Here is a spec. for a typical ATX: rather a heavy .pdf, I'm afraid. http://www.sunpower-uk.com/public/ranges/pdfs%5Cr1635%5Cr1635_1.pdf

Input-output leakage and isolation specs are given about half-way down page 1. The maximum leakage current on this one is actually a rather tingly 3.5mA @ 240V, probably mainly from input filtering.

This thing would therefore have to have its outputs firmly grounded - especially in my country with its doubly lethal 230V supply. Normally desktop PCs here are wired to grounded outlets, in contrast to laptops which normally do not. Very much lower leakage versions with enhanced isolation integrity are produced for critical applications, e.g. in medical equipment.

The isolation is not achieved by 50Hz / 60Hz mains transformers, but by double wound higher frequency transformers driven by switching circuits. I have to admit it's been a while since I took one of these apart, but the ones I have seen used transformers with ferrite C-cores rather than toroids, which are not so easy to wind. Maybe improved production methods make it worthwhile to harvest the potential gains e.g. in lower leakage inductance, but I wonder if they are actually as secure.

Maybe in the USA where the mains voltage is lower, it is considered acceptable to build PCs with no mains isolation???

In the words of John McEnroe...
 
Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Adjuster,
No, it's NOT acceptable to build PC power supplies without mains isolation here.

I posted this link in my prior reply; have a look:
http://danyk.wz.cz/s_atx_en.html
LOTS of PC AT and ATX form factor supply schematics.
All desktop/tower PC's in the States come with 3-prong plugs on their cords, so that the chassis of the supply (and therefore the case) is connected to earth ground.
 

tom66

Joined May 9, 2009
2,595
Just curious, where is your sources on this? I've opened up more than one PC power supply, there are no transformers, only toroids, diodes, and FETs.
Ferrite transformers - the yellow things.



If you look carefully, you'll see an isolation gap between the live side (with the mains filter caps and primary switching transistors) and the secondary side. Especially near the opto couplers; to achieve isolation as required, a cut has been made in the PCB between the hot and cold sides.

However, note that many modern power supplies attach a Y cap on live and neutral to earth. This is why the power supply must be earthed. If it is not, you will get ~115Vac (~62Vac for US) on the case of the PC, which makes for quite an unpleasant shock (although the current is low because the caps are usually only a few nanofarads each.)
 

Adjuster

Joined Dec 26, 2010
2,148
As you might have guessed, I did not seriously expect live chassis computers to be allowed, even in countries with lower (but still potentially fatal) supply voltages.

I was expecting to get some responses to end uncertainty, as has now happened.
 

Wendy

Joined Mar 24, 2008
23,415
Anytime my computer goes intermittent power supply is my first guess. I've had a fair share go out in my time.

I've never thought to take a picture of one, I usually look close at it before throwing it away. Great, now I have another ritual for my obsessive compulsive index.

I've heard horror stories about bad PSUs taking out everything else. My personal experience with this happening involves my Dad blowing me off about a bad fan in the PSU. I told him to stop using the computer if the noise stops, he tells me he thought it fixed itself. $200 and a new power supply later...

I pay close attention to fans, and don't hesitate to replace them if they go bad. I have a large stock just because they are so critical. This is probably why I've never had a bit of trouble using really cheap PSUs.
 

tom66

Joined May 9, 2009
2,595
The main problem with cheap supplies is that they will never meet their specifications, especially not over time. As Evil Lurker indicated, Fuhjyuu caps are some of the worst (next to Capxon, Chhsi and Sacon.) In fact, 80-90% of all power supplies I repair have bad caps.

There is a case of a certain series of Bestec power supplies incinerating themselves and the motherboard when they go bad. A small 47u cap goes bad, which causes the 5V standby (+5VSB) to rocket to 12V, quickly destroying many components on the motherboard.
 

Wendy

Joined Mar 24, 2008
23,415
Never is as strong word. :) I've never had a problem with them, until they fail. My usual service lasts around 3-5 years on the failures, I have ones that have worked from day one and still work. The ones where the fans fail get a new fan.
 

tom66

Joined May 9, 2009
2,595
Never is as strong word. :) I've never had a problem with them, until they fail. My usual service lasts around 3-5 years on the failures, I have ones that have worked from day one and still work. The ones where the fans fail get a new fan.
I suppose, for cheap systems, they might be okay. But I see a lot of people spending lots on a nice PC and then putting a cheap PSU in...

Me, sometimes I don't mind a cheap supply, as I usually can repair them. But some of the problems are not easy to correct, like the 5Vsb problem. (And that happens pretty much no matter how much you use it - the 5Vsb is usually very lightly loaded anyway.)
 
Top