Atx 500 watt psu problem ?

[tron87]

I got one of these a while ago to power some 5vdc electrical projects of mine and has been good as gold for a few months then all of sudden the thing wont power up the fan wont spin.Ive disconnected and reconnected the lead the fuse is good.Any suggestions to get the thing working again ?


http://ecx.images-amazon.com/images/I/41aebWJNo6L._AA300_.jpg

 

[MrAl]

There are so many possible reasons this wont work.

Does the fan turn even a tiny fraction of one revolution when first plugged in (or turned on) ? That could mean it ties to power up but then detects a fault.

One of my power supplies went bad when one or more of the electrolytic capacitors developed a very high ESR. Replacing most of the caps fixed the problem.

 

[tron87]

There are so many possible reasons this wont work.

Does the fan turn even a tiny fraction of one revolution when first plugged in (or turned on) ? That could mean it ties to power up but then detects a fault.

One of my power supplies went bad when one or more of the electrolytic capacitors developed a very high ESR. Replacing most of the caps fixed the problem.

When i connect ps_on(green wire) to common(black) the fan blade twitches a tiny bit not even quarter of a rev and the psu makes a high frequency buzz noise for less then a second then nothing happens.Nothing is shorted across the power rails and ive got a 6 watt 6.8ohm resistor across the 3.3v rail to 0v to help mountain regulation.Tried cycling the switch on the back of the unit aswell to no luck.

 

[Sceadwian]

The high frequency buzz you heard is an internal short, the supply is likely dead. It would cost you more time in diagnostics and parts to fix than the supply is worth. Verify that none of your alterations have caused an internal short, and if you can't find anything wrong after a few more looks, and possibly replacing an obviously faulty cap, scrap it and start over.

By the way as far as I've read the load resistor for a PSU you should be applied to the 5V rail not the 3.3V rail.

Could you explain better what exactly was done to the supply as far as alterations go and what projects you've had that were powered by it?

 

[tron87]

The only alterations to the atx psu were the original molex connector being snipped off and the corresponding power rail wires grouped up like for example all the red for +5v and yellow for +12v etc and attached to a screw terminal block for easy access.It was used to power a non inductive resistive load of 200ma placed across the +12v and -12v rails at the same time it powered another non inductive resistive load of 8 amps across the +5v rail and 0v.It did this 24/7 for 2 months and out of the blue just quit on me,there was no dead shorts when i inspected it.

 

[Sceadwian]

Sounds like a cheap power supply. I'm not referring to it's possible cost but it's construction.

 

[MrAl]

When i connect ps_on(green wire) to common(black) the fan blade twitches a tiny bit not even quarter of a rev and the psu makes a high frequency buzz noise for less then a second then nothing happens.Nothing is shorted across the power rails and ive got a 6 watt 6.8ohm resistor across the 3.3v rail to 0v to help mountain regulation.Tried cycling the switch on the back of the unit aswell to no luck.

Hi,

That's what happens when it tries to turn on and the internal protection circuits detect some sort of fault and kills the power. That 'probably' means that the basic circuit still works but something on the secondary side is bad.

This can happen with several kinds of faults, such as over current which can happen with a shorted output cap, but this can also happen with a cap with high ESR because there is less filtering action and so the voltage peaks go up higher and that triggers the over voltage section which turns the power off.

What you would do is look for the section that detects over current and see if that is what is triggering it, and look for the section that detects over voltage and see if that is what triggers it. You might not find a schematic for your exact power supply but many computer power supplies are designed in a similar manner so you could use a schematic for a different power supply to get a rough idea where to look. It's not easy to do but it is possible.

It could still be a number of problems though that need fixing. You could check the input transistor with an ohm meter perhaps, check output caps for high ESR. I removed some of my caps and used a frequency generator and scope to check the ESR and that's how i found out that several of the caps had gone high in ESR. That was the only problem and it was triggering the over voltage shut down circuit so it would not start up properly. As yours, the fan would move slightly and then stop as the power supply shut down.
Caps that leak show signs of stuff on the PC board, and often the tops show leaking material, either wet or dry.

At the very least you'd have to remove the circuit board. You have to be careful though to make sure the two input caps are not charged. Probably wait overnight before opening the case, then remove the circuit board carefully. Take several pictures of the component side and the foil side and we can take a look right here. Use the best camera you have. Take side view pics too.

Do you have any test equipment like a scope?

 

[tron87]

Nothing seems a miss on the foil side no burning short etc.None of the caps look bulged from side of top view.

I can find schematics for 250,450watt atx psu but not the 500 watt model.
Any idea where i might find one ?

Thats the only pic i can provide hope its enough to help.

Ive got a multimeter and a 20mhz analog scope.

**broken link removed**

 

[large_ghostman]

Hi,

T check output caps for high ESR. I removed some of my caps and used a frequency generator and scope to check the ESR and that's how i found out that several of the caps had gone high in ESR.

?

sorry for asking but can you explain how you check ESR like that?? i would be very interestead as i cant afford a ESR meter yet
many thanks LG

 

[MrAl]

Hi again,

Tron:
Im afraid that picture is a little too blurry, maybe you could try to get a better pic. Alternately you could read the number off of that chip at the upper left corner of the picture and we might find a data sheet on it.

Little Ghostman:
Sure no problem. I included a diagram that shows what to look for.
The cap is driven by a frequency generator and a resistor, but you could use a 555 timer chip with approximately 50 percent duty cycle output and a resistor in series with the output.
Normally it takes a cap a little while to charge up with a voltage source and resistor in series, but when there is ESR the ESR allows the voltage to 'jump' up or down suddenly without taking any time because the resistance part is not reactive. This shows up in the voltage wave on the scope as shown in the attachment. The voltage due to the capacitance ramps up or down while the voltage due to the ESR shoots up or down without any ramping. Knowing the current and change in voltage we can calculate the ESR, but high ESR shows up as a large, quick change, and when compared to a 'good' cap the difference is quite obvious so we dont even have to calculate anything if we have a good cap laying around we can compare it to.

Also, i may have forgotten to mention that when testing the power supply out of the case without a motherboard, a load resistor is required. I was able to get mine up and running with a 10 ohm load from the 3.3v supply to ground. Yours might need 5 ohms or perhaps a load on the 5v supply. Without that it might shut down right away just like that.

 

[large_ghostman]

great information thank you very much!!! i will get the function generator out later (when dad isnt around) and my scope i have loads of old caps i can test and see if i can spot any bad ones i also have a few that are leaking that i havnt thrown away yet so they are sure to give a good result (a good result will be a bad cap ) i can stop saveing for a esr meter now thank you for your help lg

 

[large_ghostman]

by the way tron we use ATX power suplies alot dad runs the incubators from them so we have many of them and a few always pack up now and then i think out of the last 11 that gave up 10 of them were bad caps that dad replaced and 1 was a fuse we used to get the PSU's free from the computer repair shop because they gave dad all the broken ones they got alot of them it was just a fuse that had gone in the plug or PSU and others were bulging caps from what i remember the shop has gone now

 

[tron87]

Hi again,

Tron:
Im afraid that picture is a little too blurry, maybe you could try to get a better pic. Alternately you could read the number off of that chip at the upper left corner of the picture and we might find a data sheet on it.

Little Ghostman:
Sure no problem. I included a diagram that shows what to look for.
The cap is driven by a frequency generator and a resistor, but you could use a 555 timer chip with approximately 50 percent duty cycle output and a resistor in series with the output.
Normally it takes a cap a little while to charge up with a voltage source and resistor in series, but when there is ESR the ESR allows the voltage to 'jump' up or down suddenly without taking any time because the resistance part is not reactive. This shows up in the voltage wave on the scope as shown in the attachment. The voltage due to the capacitance ramps up or down while the voltage due to the ESR shoots up or down without any ramping. Knowing the current and change in voltage we can calculate the ESR, but high ESR shows up as a large, quick change, and when compared to a 'good' cap the difference is quite obvious so we dont even have to calculate anything if we have a good cap laying around we can compare it to.

Also, i may have forgotten to mention that when testing the power supply out of the case without a motherboard, a load resistor is required. I was able to get mine up and running with a 10 ohm load from the 3.3v supply to ground. Yours might need 5 ohms or perhaps a load on the 5v supply. Without that it might shut down right away just like that.

The number on that chip is first line : hs8108b then on second line : a6ahb

 

[MrAl]

great information thank you very much!!! i will get the function generator out later (when dad isnt around) and my scope i have loads of old caps i can test and see if i can spot any bad ones i also have a few that are leaking that i havnt thrown away yet so they are sure to give a good result (a good result will be a bad cap ) i can stop saveing for a esr meter now thank you for your help lg

Hi LG,

You're welcome i hope you can make use of it and pass the information on so others can use it too.

 

[MrAl]

The number on that chip is first line : hs8108b then on second line : a6ahb

Hi again Tron,


Any chance you can get a more clear picture of the board? I cant make out any components too well. Also, you should take several pictures from different angles before you take any parts off. If you remove caps for checking, you'd want to get them right back where they belong or new replacements, and you definitely dont want to mix up the locations for them or it's history. This is very important.

Also, i cant find that chip on the web except for just a picture of it. What would help here is a data sheet for that exact part number.

If we cant find the data sheet then the next best thing would be to just perform a basic test on all the parts, one by one. There are not that many parts that usually go bad so it wont take all day to do. Check the transistors for shorts (you know how to do this?) and it looks like you might have to check the caps for high ESR too. It's going to take a little work on your part.
You might be able to get a quick glimps of the output with the scope on the 12v line before the supply shuts down.

 

[tron87]

Hi again Tron,


Any chance you can get a more clear picture of the board? I cant make out any components too well. Also, you should take several pictures from different angles before you take any parts off. If you remove caps for checking, you'd want to get them right back where they belong or new replacements, and you definitely dont want to mix up the locations for them or it's history. This is very important.

Also, i cant find that chip on the web except for just a picture of it. What would help here is a data sheet for that exact part number.

If we cant find the data sheet then the next best thing would be to just perform a basic test on all the parts, one by one. There are not that many parts that usually go bad so it wont take all day to do. Check the transistors for shorts (you know how to do this?) and it looks like you might have to check the caps for high ESR too. It's going to take a little work on your part.
You might be able to get a quick glimps of the output with the scope on the 12v line before the supply shuts down.

You know what its a lot less trouble just getting another psu ill probably do that and hope it was just a one off.Thanks very much for taking time to give some very helpful advice most appreciated .

 

[Sceadwian]

Around 25 dollars online for cheepies, 50-150 for higher quality units. The 150 range is because they're dead silent fanless or ultra low noise fan's with minimal airflow.

 

[MrAl]

You know what its a lot less trouble just getting another psu ill probably do that and hope it was just a one off.Thanks very much for taking time to give some very helpful advice most appreciated .

Hi,


Yes that's right Power supply prices have come down a lot and on sale you can get one for a pretty good price these days.
I fix mine mainly for the experience of it all and i end up knowing a little bit more about them after that.