How does a Desktop PC ATX P/S make 5V Standby Power?

[martys]

I recently opened up a blown ATX P/S and I am fascinated to learn how it develops the 5V standby power.
The PCB inside has three yellow-taped transformers. The largest I can clearly see is used to create the 3.3V, 5V, -12V, 12V outputs.

I can safely assume one of the other two tall, skinny transformers is to deliver the base-drive to the half-bridge converter.

So the third identically-shaped skinny transformer might be used to provide 5V 1-Amp approx. of standby power when the PC is off.

But I don't see any control IC to create a SMPS 5V supply, except the the single 18-pin jungle-chip controller on the low-voltage side. But it can't control the 5V by itself without getting startup power.

Does anyone have any idea how the 5V standby power works?  Is the third skinny-tall transformer just a step-down transformer?

If it is just a passive transformer. then it baffles me; how "Green" is power like that? Every old heavy wall-wart now is now replaced by a small, light and efficient SMPS to save standby energy.
 
Is this how things are done in every new efficient high-powered PC ATX power supply?


I could easily find a use for a free and easily salvageable and reasonably small transformer to use in one of my 5V powered projects.

[SeanB]

The standby supply typically uses a self oscillating converter, using a single ( or perhaps 2 in some cases)transistor on the primary side, with a primary winding, a feedback winding to make it oscillate and then a secondary providing either a rough 8V rail that is regulated to 5V using a standard 3 terminal regulator, or a slightly more complex one with an optoisolator and a TL431 to provide feedback to keep this 5V rail constant.

You will trace out the circuit on this transformer and see 4 pins on one side, connected to a single TO220 power transistor there, that does the switching. The secondary side will have 2 pins, one being connected to ground and the other to a power diode and then a capacitor, typically 1000uF 16V, for smoothing.

[martys]

Aha! I found and unsoldered a TSF2N60 3-pin T0-220 device( a MOSFET) mounted on the hi-side bridge large heatsink and a T0-18 (yet-unknown) device very close to the tall-skinny Xformer

Problem solved. 

Wow! A lot o nice salvageable components to pull off this P/S board. Maybe only one of them is defective.

I now have a nice quiet fan from the P/S and it is(double-sided-taped to a flat piece of plastic) sitting on my work bench to create an adjustable-direction gentle breeze to blow soldering smoke away from me while I work and powered from a discarded 8-V 1-amp wall-wart.
 

[martys]

Truesemi TSF2N60M TSF2N60 TO-220F 600V N-Channel MOSFET

I Googled it and found one, but it is no-longer available. I wonder if this is the trouble with this P/S. (Update: Checks out ok)

Why do reasonably-new ATX power supplies suddenly just up and die? I have a few given to me by a guy who fixes PC's. He says, typically no smoke, just no turn-on dead. As far as he is concerned when this happens, he just pops in a new one and charges the customer. Repair of PCs at the PCB-component level is just not done!

Is there a most-common cause for ATX P/S to go bad? I opened them up and suprisingly, inside, everything looks good as new, with maybe a little dust for decorating.

[RGB255_0_0]

If it's a $20 Amazon special then no, not worth spending the time finding the fault and at $20 many of the components are going to be Shenzhen reject special  parts.

[james_s]

There are ICs available specifically designed as standby converters for power supplies.

The most common problem I see is worn out electrolytic capacitors, that applies to all switchmode power supplies, usually you switch them off and then then next time you turn them on they won't power up. The ESR of an electrolytic capacitor rises as the temperature falls, I don't know why this is, but it means that a tired cap will often work ok when warm, but once it cools off the ESR is too high for the thing to start up.

[Iceberg86300]

Heh. The only PS I had go bad failed EPICALLY. Sparks shooting out the back, me screaming NOOOOOO! epic.

Quite scary, not because of the sparks, but because it was a brand new computer build. I thought it toasted everything for sure.

But nope, just the PS. I pulled the cord before it was done sparking and exchanged it the next day.

Fry's idiotic return process made the entire ordeal worth going through. Even though I was exchanging for the same exact product, they insisted on testing it.

Both the store manager and the person manning the returns desk ended up on their ass.

The replacement is still going strong, powering my i7 ~10 years later. (PS was very nice, but was being sold at a super discount as it was last year's model and they had a ton of them)

[SeanB]

Most common power supply ( in the PC world) failures are failed capacitors ( bulging on most cases, but there are plenty that just dump everything out the bottom with no fanfare) and open circuit resistors on the primary side. The power supply relies on the standby 5V supply running, and this provides power for the main controller on the secondary side, and this then drives the main switching stage to drive the power outputs.

This standby supply needs a start up supply and bias, typically provided by a high value ( 470k, 1M or so) resistor from the high voltage rail to provide initial bias, and this then starts a single cycle of oscillation and then the feedback winding on the oscillator keeps it running. If this resistor goes high or open ( really the same thing) then the supply will keep on running with power applied, but disconnect it and it will not start up again.

Replace the resistor, replace 20 capacitors ( basically every electrolytic, even the small ones) and it will run again, but the cost of the capacitors is often more than double the cost of a new supply, showing you just how cheap the original ones were, plus the markup in the supply chain for the replacements.

[james_s]

Heh. The only PS I had go bad failed EPICALLY. Sparks shooting out the back, me screaming NOOOOOO! epic.

Quite scary, not because of the sparks, but because it was a brand new computer build. I thought it toasted everything for sure.

But nope, just the PS. I pulled the cord before it was done sparking and exchanged it the next day.

Fry's idiotic return process made the entire ordeal worth going through. Even though I was exchanging for the same exact product, they insisted on testing it.

Both the store manager and the person manning the returns desk ended up on their ass.

The replacement is still going strong, powering my i7 ~10 years later. (PS was very nice, but was being sold at a super discount as it was last year's model and they had a ton of them)

I've had two ATX PSUs fail with a bang and sparks shot out the fan vent on the back. One only killed the power supply, the other fried the motherboard, CPU chip, video card, RAM and one of the hard drives. The sound card, network card, CD-ROM and floppy drive survived.

[Iceberg86300]

I've had two ATX PSUs fail with a bang and sparks shot out the fan vent on the back. One only killed the power supply, the other fried the motherboard, CPU chip, video card, RAM and one of the hard drives. The sound card, network card, CD-ROM and floppy drive survived.

Ouch!! That had to hurt.

The parts that failed in your case were the only ones I had installed. In hindsight, I should have gone a different PS when I returned the original. Obviously the "Tested/OK" sticker from this company (possibly involving Fry's, given their purchasing habits) wasn't worth anything.

I think the PS was ~60 bucks, while everything else totaled ~1000. Spending another $100 on the PS shouldn't have been a worry. I let out one hell of a sigh when it finally powered up and posted.

What's the failure mode in this case? It looked like a HV transformer letting all the PCB's out.

[james_s]

PCBs in reference to transformers is insulating/cooling oil that was used in large utility transformers and as a dielectric in some types of oil filled capacitors found in fluorescent ballasts and such up into the 1970s, it's not something you're going to find in a PC power supply.

I didn't do an exhaustive investigation into the failures I had but from what I could tell they were typical of the cascading failures that tend to occur in these things. Something caused one of the mosfets in the half bridge to short which blew the other mosfet and the resulting direct short blew up the rectifier and other stuff until the fuse blew and opened the circuit. In the one that fried the motherboard I suspect something caused the feedback loop to open up and the regulator went balls out full throttle until it blew up, but there's no way to know for sure. In both cases I walked in, pushed the power button and BANG!

[martys]

When I gathered up the dead ATX P/S's to take home to tinker with, I only took the cleanest, no smoke-smell, newest looking ones.
So, I have the cream of the cr*p.
The electrolytics inside them look as good as new, mostly 105-C rated. no domed, bulging tops. I unsoldered a few and their ESR is like some  new caps.I compared them to.

Must be something else, maybe the Silent Supply Killer's work.

[amyk]

Lots of AT/ATX PSU schematics here: http://danyk.cz/s_atx_en.html

[Iceberg86300]

PCBs in reference to transformers is insulating/cooling oil that was used in large utility transformers and as a dielectric in some types of oil filled capacitors found in fluorescent ballasts and such up into the 1970s, it's not something you're going to find in a PC power supply.

I didn't do an exhaustive investigation into the failures I had but from what I could tell they were typical of the cascading failures that tend to occur in these things. Something caused one of the mosfets in the half bridge to short which blew the other mosfet and the resulting direct short blew up the rectifier and other stuff until the fuse blew and opened the circuit. In the one that fried the motherboard I suspect something caused the feedback loop to open up and the regulator went balls out full throttle until it blew up, but there's no way to know for sure. In both cases I walked in, pushed the power button and BANG!

Interesting. I suppose I got lucky with my failure. What was weird was that it was a prolonged failure which stopped shooting colorful sparks when unplugged, but resumed when the idiots refused to take my word that it failed and proceeded to plug it back in.

That was part of the reason that I mentioned HV transformers & PCB's. I know PCB's all too well, as I grew up in Pittsfield, MA. Was A HUGE GE boom town that housed much of their power transformer business along with r&d until they pulled up stakes, leaving a huge mess of PCB's and other contaminants behind. I believe it was an EPA super fund site. Still not cleaned up completely, and I went to elementary school on a site in which they had to remove a few feet of soil, put a barrier down, then back fill with clean soil. Shit is still in the water table, just at "acceptable" levels now.

My PS failure just reminded me of a few large transformer failures that cascaded somewhere in the plant/lab. Those were some really awesome light shows.

I couldn't dissect mine as they definitely wouldn't have taken it back with broken seals.

Still amazed at what seemed like a very miniature version of those cascading power transformer failures I witnessed at GE though. And no big bangs either. Plenty of what sounded like "arc popping" though, also similar to those large transformer failures.

But I guess I'll never know. Probably just junk shorting and burning various materials.

I shouldn't make inside jokes that only include me, due to bitterness of what GE did to my hometown. LOL.

It's been an interesting discussion though. Just shows how things can go terribly wrong with any scale of power transformation gear. Only difference is the monetary scale. Our PS failures ranged from a couple hundred bucks to probably a couple grand, whereas I could have built a super computer cluster a few times over with the about of money that went up in smoke with what I saw at GE.

Hopefully we won't be adding any more personal experience to PC PS failures!!

[Iceberg86300]

When I gathered up the dead ATX P/S's to take home to tinker with, I only took the cleanest, no smoke-smell, newest looking ones.
So, I have the cream of the cr*p.
The electrolytics inside them look as good as new, mostly 105-C rated. no domed, bulging tops. I unsoldered a few and their ESR is like some  new caps.I compared them to.

Must be something else, maybe the Silent Supply Killer's work.

Always good stuff to have in the parts bin!

Unfortunately, with my experience working with TV's anyway, it's been about 50/50 for bad caps; they've either failed in a glorious physical fashion, or they look brand new.