DIY Benchtop Power Supply Failure?

[Renegade]

I've been working diligently on building a benchtop power supply from a PC ATX power supply as one of my first projects.

I'm looking for any advise about potential trouble shooting steps that I should take. Did I make a beginner blunder?

I got to the end where all I needed to do was glue in the displays and bolt on the enclosure lid.

I tested it at this point and... nothing. The standby LED that I wired up from the +5VSB turned on, but that's all. Voltage in the various binding posts registered 0V or 0.003V (or very low). The fans don't turn on either.

I used these additional parts:

3-digit voltage meters:

http://www.aliexpress.com/item/0-28-Inch-2-5V-30V-Mini-Digital-Voltmeter-Voltage-Tester-Meter-LED-Screen-Electronic-Parts/32550332566.html

Volt/amp/power meter:

http://www.aliexpress.com/item/50A-DC-Digital-Power-Panel-Meter-Monitor-Power-Energy-Voltmeter-Ammeter-Shunt-free-shipping/32432078283.html

Buck boost voltage converter:

http://www.aliexpress.com/item/Z116-LM2577S-LM2596S-DC-DC-Step-Up-Down-Boost-Buck-Voltage-Power-Converter-Module/32243179400.html

For that, I desoldered the trim pots and soldered in 4 pots: 2 for voltage (fine and coarse), and 1 each for the current pots. I tested this independently and it worked.



+5V and ground are connected to a 30 ohm 10W resistor for a phantom load.

POWER_GOOD

--- sorry -- at this point it posted by accident... I'm continuing... ---

POWER_GOOD is connected to a blue LED and does not turn on (has a 100 ohm resistor in there just like the +5VSB standby rail). 

+5V is connected to one of the 3-digit voltmeters. It doesn't turn on.

+3.3V is connected to one of the 3-digit voltmeters. It doesn't turn on.

+12V is connected to one of the 3-digit voltmeters. It doesn't turn on.

I also wired up -5V and -12V as well as an additional +12V that is wired to the volt/amp/power meter.

When I wire up a 12V battery pack to the inputs of the buck/boost converter, the PSU fans (both) turn on, the display lights up, and the +12V voltmeter display lights up (all these are on the +12V rail).

I'm a bit mystified as I walked over each and every tedious step at least 3x each to recheck that I was wiring things up properly.

Is the PSU dead? Do I need to rip it out and stick in a new one?




Any pointers are appreciated.

[Skimask]

Did you ground PS_ON# (green wire)?

https://www.google.com/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=atx%20psu%20green%20wire

[Renegade]

Did you ground PS_ON# (green wire)?

https://www.google.com/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=atx%20psu%20green%20wire

Yes. Sorry. I forgot to mention that. I grounded it through an On/Off switch.

When I flip the switch, the large volt/amp/power meter display flickers on for a brief moment then goes off.

[Skimask]

2 things...

On a lot of the newer ATX PSU's, if a hard short is detected, they'll shut down...and that's that...until power is cycled.

And on most, each of the rails needs a minimum load to function correctly...something like a few % of full load.

[Renegade]

On a lot of the newer ATX PSU's, if a hard short is detected, they'll shut down...and that's that...until power is cycled.

It's an older PSU. I took it out of an old Windows XP computer.

And on most, each of the rails needs a minimum load to function correctly...something like a few % of full load.

I used a 30 ohm 10W resistor on the +5V rail for that. Did I make a mistake there?

EDIT: The +5V rail has 25A on it.

[Ian.M]

So you put time and effort into modding a PSU that was at least 8 years old to start with, more likely 10, and had >80% odds of being a cheap Chinese P.O.S. even when it was brand new?

The standby supply is still working, so you know power is gettig as far as the HV DC bus, but you are seeing nothing out of the main supply, haven't heard any loud bangs or smelt any smoke, so either the control chip's startup circuit has failed or its gone into protection mode.   You don't have the skills and experience to fix it, and even if you did it wouldn't be economic.

Where to go from here?  Well, I'd probably start over with a good quality new PSU if I was going to have to mod it extensively, or if I was using used PSUs, with the power connector salvaged from a scrap motherboard so I could easily plug in a replacement whenever one fails.

[Skimask]

EDIT: The +5V rail has 25A on it.

EDIT: EDIT:  The +5V rail apparently has 0A on it.

[Renegade]

So you put time and effort into modding a PSU that was at least 8 years old to start with, more likely 10, and had >80% odds of being a cheap Chinese P.O.S. even when it was brand new?

Yep. What isn't made in China?

The standby supply is still working, so you know power is gettig as far as the HV DC bus, but you are seeing nothing out of the main supply, haven't heard any loud bangs or smelt any smoke, so either the control chip's startup circuit has failed or its gone into protection mode.   You don't have the skills and experience to fix it, and even if you did it wouldn't be economic.

No need to rub it in. It's not about the economics of it. I'd have gone out and spent the money on a new lab power supply if it were that important. This is about learning, and I've hit a hiccup here.

My aim is to plow through the problems in the real world and get some experience to go along with the theory.

Where to go from here?  Well, I'd probably start over with a good quality new PSU if I was going to have to mod it extensively, or if I was using used PSUs, with the power connector salvaged from a scrap motherboard so I could easily plug in a replacement whenever one fails.

I considered going the connector route, but was just impatient and didn't want to bother with it. In retrospect, that was a bad decision. So... lesson learned.

[Renegade]

EDIT: The +5V rail has 25A on it.

EDIT: EDIT:  The +5V rail apparently has 0A on it.

Yes. It's supposed to have 25A there, but... not much ATM.

[Ian.M]

I've fixed a few PC PSUs in my time, but it was only ever worth doing if it was a custom form-factor or motherboard connector. 

Most of the time you are working 'blind' without reference to a schematic.  At best, you have the schematic of a reference design using the same controller chip.   Unless your PSU is 120V only, it will have a voltage doubler circuit that charges the reservoir caps to approx 340V DC, and there is a *LOT* of energy stored in them.  The primary side of a SMPSU with a high voltage DC bus is no place for a novice.  Make a mistake, and the likely outcome is a blown oscilloscope or a serious electric shock. 
If you've got the experience and equipment to work on it, you should be well past the stage at which converting one to a 'bench' supply seems worthwhile.  Therefore your investigation should be restricted to 'cold' checks after safely discharging the reservoir caps.
If you are lucky, you may find a small failed electrolytic or an open circuit high value startup resistor, which will go from the +HV DC rail to one of the pins of the control chip.

If you've got a scope, you could check the various output rails unloaded to see if any of them are pulsing slightly.  If they are, look for the one that isn't and check for shorts on it.  However the odds are not good at all.

[Skimask]

Yes. It's supposed to have 25A there, but... not much ATM.

No, it does NOT have 25Amps on it.
It will NEVER have 25 Amps on it.  Unless you put a load worthy of 25 Amps on it.
Do you know exactly what that means?

[wblock]

Was the XP computer still working when the power supply was removed?  Does visual inspection reveal any bulging capacitors or other problems?

Were the modifications made directly to the power supply, or does the additional circuitry all plug into the 20- or 24-pin connector?  (If unmodified, the power supply can be tested on the original motherboard.)

Was this from a Dell?  Just to be evil, Dell used to use non-standard pinouts on some of their power supplies.

Pictures might help.

[Renegade]

@Ian.M Thanks. I think I'll just get another PSU.

Yes. It's supposed to have 25A there, but... not much ATM.

No, it does NOT have 25Amps on it.
It will NEVER have 25 Amps on it.  Unless you put a load worthy of 25 Amps on it.
Do you know exactly what that means?

I thought you were simply making a joke above. Yes, I know what it means. Linguistic shortcut there. has vs. is rated for.

[Skimask]

I thought you were simply making a joke above. Yes, I know what it means. Linguistic shortcut there. has vs. is rated for.

Ok, just checking 

[Renegade]

I thought you were simply making a joke above. Yes, I know what it means. Linguistic shortcut there. has vs. is rated for.

Ok, just checking 

Checking can't hurt.

In other news, I've ordered another PSU that will be arriving tomorrow.

I'll chop off the wires, then solder them all again. Yay. Fun. Oh well... This will be my first PSU to benchtop power supply hack. I want to do another from (mostly) scratch later on, but first things first. Baby steps and all. Perhaps some day I may be giving advise in these forums.

[rickselectricalprojects]

I thought you were simply making a joke above. Yes, I know what it means. Linguistic shortcut there. has vs. is rated for.

Ok, just checking 

Checking can't hurt.

In other news, I've ordered another PSU that will be arriving tomorrow.

I'll chop off the wires, then solder them all again. Yay. Fun. Oh well... This will be my first PSU to benchtop power supply hack. I want to do another from (mostly) scratch later on, but first things first. Baby steps and all. Perhaps some day I may be giving advise in these forums.

You said you have already ordered a new PSU but in the future i would recommend getting a Dell PSU. I have two Dell PSU's and they both are built very well. One of them has Nippon-chemicon capacitors!

[Ian.M]

Its definitely preferable *NOT* to chop the wires.  Get an ATX power extension cable and hack that, or if you are really cheap, hacksaw the connector section off a scrap motherboard , just outside the connector footprint, clean up the edges with a file then solder your own wiring direct to the pads.

[Renegade]

You said you have already ordered a new PSU but in the future i would recommend getting a Dell PSU. I have two Dell PSU's and they both are built very well. One of them has Nippon-chemicon capacitors!

I have a tendency to go apeshit crazy when spending, so I'm trying to limit myself there by only buying used PSUs. I will keep that in mind though for the next time if I can find a Dell PSU.

I manged to find a Compaq PSU for $5, which I figured would be decent quality. However, when I measure the voltage on the 12V rail, it comes in at about 10V. I don't know why the voltage is so low, so I decided not to use it and got another one instead (the one I mentioned above).

I may revisit the Compaq PSU that I got later, but for the moment, I've got another PSU that measures the correct voltages.

Its definitely preferable *NOT* to chop the wires.  Get an ATX power extension cable and hack that, or if you are really cheap, hacksaw the connector section off a scrap motherboard , just outside the connector footprint, clean up the edges with a file then solder your own wiring direct to the pads.

Heh! I actually prepared to go both routes -- I hacked the connector out of a motherboard and I also bought a 24-pin/20-pin connector as well as a couple 20-pin male connectors hacked off of PSUs. So, I have those ready for when I have time to redo all the work that I'd already done.

Its definitely preferable *NOT* to chop the wires. 

I thought about that before after seeing this video (How to NOT convert ATX power supply to benchtop or lab power supply):



Given the pain I have now with a dead PSU, I'm really thinking that leaving the PSU cables intact is the right way to go. I knew that beforehand, but was just impatient.