A quick inspection for the Dell's power supply

[eagle1109]

Hello,

I got this PSU from one of my colleagues, and did a quick inspection for the faults in this PSU.


I recorded a video on YouTube and looking for opinions and ideas about this unit before I do any desoldering.

 

[Akanimo]

Hi!

Can you tell us what exactly is the problem with the PSU? You haven't told us what fault you're troubleshooting. Any observations you can tell us about its operation?

Can you check and tell us the part number for the transistors in question? Are they IGBTs, MOSFETS, BJTs...? Can you take a closer and clearer picture of it and post it here? That way we might begin to have a direction of thoughta.

 

[betwixt]

You are looking in the wrong place.

The 'Transistors' with shorts across the outside pins are dual-diodes. The outside pins are connected through the transformer secondary to ground and to each other. It is normal for a very low resistance between them (< 1 Ohm). The middle pin is the DC output and it connects to the reservoir capacitors and then to the output wires. The short beep you hear is caused by the current flowing into the reservoir capacitors, it is quite normal.

One of the two transformers is for the main power, the other is for producing the 5V standby output. Your first check should be to feed AC in to it and see if 5V is being produced across the output wires, usually it is on the purple colored wire to any of the black wires.

Be extremely careful, even if it isn't working, there could be very high voltages present (> 300V) and part of the circuit is not isolated from the AC power wires so they present a serious electrocution risk. The line of black opto-couplers marks the division between the mains and the isolated side of the supply, nothing must ever bridge one side to the other.

Brian.

 

[eagle1109]

Hi!
Can you tell us what exactly is the problem with the PSU? You haven't told us what fault you're troubleshooting. Any observations you can tell us about its operation?

It gives 0V on all output rails.

Can you check and tell us the part number for the transistors in question? Are they IGBTs, MOSFETS, BJTs...? Can you take a closer and clearer picture of it and post it here? That way we might begin to have a direction of thoughta.

Yes I know that is essential to the troubleshooting process is to know the part number, but I thought to record a video just before desoldering any parts. But you're right this video is like %10 explaining the problem of the PSU. I guess I have to take them out and take each one's number and get its datasheet.

You are looking in the wrong place.

The 'Transistors' with shorts across the outside pins are dual-diodes. The outside pins are connected through the transformer secondary to ground and to each other. It is normal for a very low resistance between them (< 1 Ohm). The middle pin is the DC output and it connects to the reservoir capacitors and then to the output wires. The short beep you hear is caused by the current flowing into the reservoir capacitors, it is quite normal.

OK, I'm not fully understanding the whole picture about the output transistors. I remember one time I worked on a CRT TV which has rectification diodes after the high frequency transformer. But the transistors is just a new topic to me.

I fixed my gaming rig PSU which had the 12V rail power MOSFET broken, I replaced it and the problem was fixed. This is why I think these semiconductors are power transistors and mostly MOSFETs.

One of the two transformers is for the main power, the other is for producing the 5V standby output. Your first check should be to feed AC in to it and see if 5V is being produced across the output wires, usually it is on the purple colored wire to any of the black wires.

Yes I guess the bigger one is for main DC switching and the smaller one for standby power.

Yes I tried to measure the voltage across all power rails but I get 0V.

Be extremely careful, even if it isn't working, there could be very high voltages present (> 300V) and part of the circuit is not isolated from the AC power wires so they present a serious electrocution risk. The line of black opto-couplers marks the division between the mains and the isolated side of the supply, nothing must ever bridge one side to the other.

Brian.

Yes I know. I actually after each power up to the PSU, I discharge the voltage in the main capacitor which is 320V with a resistor until the voltage is very low.

 

[betwixt]

OK, I'm not fully understanding the whole picture about the output transistors. I remember one time I worked on a CRT TV which has rectification diodes after the high frequency transformer. But the transistors is just a new topic to me.

It's a completely different principle - what look like transistors are actually two power diodes sharing a common connection on the center pin. Because they are wired across a transformer secondary with just a few turns of wire, they will measure shorted with a test meter. They are not the problem.

Almost all faults on that kind of PSU are on the primary (high voltage) side, it is unlikely a fault on the low voltage side would completely shut it down. You would still get some output but at reduced voltage or fluctuating voltage.

As the standby supply is also missing, it indicates a fault near to the AC input. After the rectifier and high voltage reservoir capacitors, the DC splits two ways, one to the standby supply and one to the main supply. The main supply requires the standby supply to be working before it will operate. In a normal PC, the standby supply keeps the power switch monitoring circuit on the motherboard active so when you press it, the main supply is turned on. There is a way to 'cheat' the main supply to turn it on by linking the green output wire to one of the black wires but if the standby supply is missing there is little point in trying that.

Most PSU start-up problems are caused by the smaller electrolytic capacitors dropping in value. Before taking anything apart, look for low value capacitors near the transformers and input side heat sink. Typical values will be 1uF to 47uF, don't waste time testing them, they start to degrade from the moment power is applied and they cost very little, just change them for new ones. I would guess they fix 75% of PSU faults.

Brian.

 

[eagle1109]

It's a completely different principle - what look like transistors are actually two power diodes sharing a common connection on the center pin. Because they are wired across a transformer secondary with just a few turns of wire, they will measure shorted with a test meter. They are not the problem.

OK, but one is for sure a broken one. I actually couldn't distinguish the numbers of the rest three transistors, just got the first one.

Almost all faults on that kind of PSU are on the primary (high voltage) side, it is unlikely a fault on the low voltage side would completely shut it down. You would still get some output but at reduced voltage or fluctuating voltage.

Yes, OK I'm working on this now, I did another test today and recorded a video.

As the standby supply is also missing, it indicates a fault near to the AC input. After the rectifier and high voltage reservoir capacitors, the DC splits two ways, one to the standby supply and one to the main supply. The main supply requires the standby supply to be working before it will operate. In a normal PC, the standby supply keeps the power switch monitoring circuit on the motherboard active so when you press it, the main supply is turned on. There is a way to 'cheat' the main supply to turn it on by linking the green output wire to one of the black wires but if the standby supply is missing there is little point in trying that.

OK, so the PSU won't work without the standby voltage in this unit because it has one, OK got that.

Most PSU start-up problems are caused by the smaller electrolytic capacitors dropping in value. Before taking anything apart, look for low value capacitors near the transformers and input side heat sink. Typical values will be 1uF to 47uF, don't waste time testing them, they start to degrade from the moment power is applied and they cost very little, just change them for new ones. I would guess they fix 75% of PSU faults.

Brian.

This is the video, I didn't which capacitors you mean

I took out these two but they are 10 and 100uF. Are they the ones?


Here:
I actually took out the second heatsink with all the 7 transistors.
**broken link removed**


What a surprise, non tested a single short across it's pins! WOW that's really interesting.

So the shorts are coming from elsewhere! Hmmm.

- - - Updated - - -

After taking out the transistors I did some measurements to guess the fault.

https://www.youtube.com/watch?v=n6WeYeBdTYI

 

[eagle1109]

Most PSU start-up problems are caused by the smaller electrolytic capacitors dropping in value. Before taking anything apart, look for low value capacitors near the transformers and input side heat sink. Typical values will be 1uF to 47uF, don't waste time testing them, they start to degrade from the moment power is applied and they cost very little, just change them for new ones. I would guess they fix 75% of PSU faults.

Brian.

Hello Mr Brian,

Could you confirm to me the location of the capacitors that I have to replace? Can you watch the videos and tell me if I got their location correct?

Thank you,

 

[betwixt]

Yes, those are the two I would suspect first. There may be another one on the small PCB under the heat sink as well. There is a slot cut through the PCB to isolate the low voltage side from the high voltage side so for now I would ignore any components to the right side (in the video) of it.

From experience, if one of the transistors goes faulty, it usually explodes or visibly burns the PCB traces around it. That is why I think the fault is more likely to be one of the other components. The small capacitors are used to provide a 'start-up' supply (not the same as the standby supply) to 'kick' the high voltage side into operation. When running, the supply for the primary side regulators comes from the transformers themselves but when the power is first applied the regulator cannot yet produce the signal to the transformer. Usually, a capacitor like the ones you identified is used to conduct a pulse of voltage to give enough for the regulator to start up, from then on it runs by itself.

The four 'transistors' you pictured are more likely to be the three leg diodes I mentioned. There will be a winding on the yellow transformer between the outer pins which is why you see almost a short circuit. It is quite normal and nothing to worry about.

Brian.

 

[eagle1109]

Yes, those are the two I would suspect first. There may be another one on the small PCB under the heat sink as well. There is a slot cut through the PCB to isolate the low voltage side from the high voltage side so for now I would ignore any components to the right side (in the video) of it.

Yes you mean the output side. OK, yes me too I think the low voltage side is less suspect to have a fault.

From experience, if one of the transistors goes faulty, it usually explodes or visibly burns the PCB traces around it. That is why I think the fault is more likely to be one of the other components. The small capacitors are used to provide a 'start-up' supply (not the same as the standby supply) to 'kick' the high voltage side into operation. When running, the supply for the primary side regulators comes from the transformers themselves but when the power is first applied the regulator cannot yet produce the signal to the transformer. Usually, a capacitor like the ones you identified is used to conduct a pulse of voltage to give enough for the regulator to start up, from then on it runs by itself.

Yeah that's the part I would suspect the fault to be from.

I changed one of the capacitors today but I forgot the other one at home. I changed the 100uF and returned the 10uF to its place.

I switched the PSU ON but the same thing. Nothing I get. I think the switcher IC could be broken, I'm ordering a lot of this IC from Aliexpress and of course I have to wait for it.

But for now I may change all the suspected capacitors near the transformers.

The four 'transistors' you pictured are more likely to be the three leg diodes I mentioned. There will be a winding on the yellow transformer between the outer pins which is why you see almost a short circuit. It is quite normal and nothing to worry about.

Brian.

OK, got that and you're right they are OK and no one is broken. There's little difference in the diode testing for each one, but I guess that's OK unless one is really broken and has a short circuit between the anode and cathode.

 

[eagle1109]

Is there a safe way to measure the HIGH voltage side with an oscilloscope?

 

[betwixt]

You can do it but only with extreme care. If in doubt - don't try it.

There are two ways:

1. use an isolating transformer at the AC input. This allows the whole PSU to 'float' independently of the mains AC supply so you can ground the negative line and measure as in a conventional circuit. The transformer could be quite expensive,

2. use differential measurement. You need a two channel oscilloscope. For this method, disconnect the scope ground probes completely, they must never touch anything else! Use the tips of the probes in 'x10' mode across the points to be measured and set the scope to 'add and invert' mode. It will then measure the difference between probe tips instead of between tips and ground.

Either way, it can be very dangerous if you don't follow the rules and it is quite easy for the whole oscilloscope to become 'live', including the chassis and front panel. Beware of the voltage rating of the probes as well, the components that do the 'x10' division will have a voltage rating which in some cases can be quite low.

Brian.

 

[eagle1109]

You can do it but only with extreme care. If in doubt - don't try it.
There are two ways:

1. use an isolating transformer at the AC input. This allows the whole PSU to 'float' independently of the mains AC supply so you can ground the negative line and measure as in a conventional circuit. The transformer could be quite expensive,

Where and how to connect this transformer? What I understood from isolating transformer, is that I take a 220V which we have a lot in the college laboratories convert 220V to 6 or 12V.

So I connect the primary wires which handles the HIGH AC voltage and connect them to the AC input rails then I connect the oscilloscope probe to the secondary rails, it that correct?

2. use differential measurement. You need a two channel oscilloscope. For this method, disconnect the scope ground probes completely, they must never touch anything else! Use the tips of the probes in 'x10' mode across the points to be measured and set the scope to 'add and invert' mode. It will then measure the difference between probe tips instead of between tips and ground.

Yes I saw this method on YouTube, I have 2-CH digital oscilloscope. I'm thinking of this method and maybe I may try it.

According to the YouTube video, as soon as I connect the +ve oscilloscope probe to any part in the HIGH voltage side I get half of that signal then I connect the second +ve probe to the ground and get the difference of that voltage. Is that correct?

Either way, it can be very dangerous if you don't follow the rules and it is quite easy for the whole oscilloscope to become 'live', including the chassis and front panel. Beware of the voltage rating of the probes as well, the components that do the 'x10' division will have a voltage rating which in some cases can be quite low.

Brian.

Yes, I really trying to find other ways to measure the signal at the HIGH voltage side because I don't own the oscilloscope and I way destroy something



But when I think about the PSU. There are no voltages on the output side. And obviously there's no power transmission from the HIGH voltage side.

So it's either the IC switcher or the POWER MOSFET.

 

[betwixt]

The transformer I was thinking of takes 220V from the wall socket and produces 220V at it's output. It is a 1:1 ratio transformer. The advantage it gives is that you can ground (like through a normal scope probe) anything on the output side without risk of a direct high current path from the wall supply. They tend to be big and heavy and expensive. If you use one, make sure it is safety rated as an isolator and not an auto-transformer for stepping up or stepping down mains voltage.

Even with a transformer in line with the AC, there will still be very high voltages in the PSU. Typically 1.4 times the AC voltage (220 * 1.4 = 308V) on the big reservoir capacitor and spikes of about twice that amount on the MOSFETS.

It could be the IC or the MOSFETs but in my experience they are usually OK. MOSFETS almost always fail as short circuited and blow fuses violently so dead ones are fairly easy to identify, a dead PSU usually means the MOSFETs have no drive signal to them.

Brian.

 

[eagle1109]

I purchased TNY280PN switcher IC, I desoldered the one in the PSU.

The one was in the PSU has a short between the BP/M and S. The new one doesn't.

But after the replacement, it still not working. About 254V on the power MOSFET and that's it.

 

[betwixt]

The TNY280 is only a low power device so I doubt it has anything to do with the main power output but it could be used to generate the standby supply.

Try this:
Link the green and black wires at the output plug.
Read the voltage between any black wire and the purple wire, let me know what it is.
Read the voltage between any black wire and any orange wire, let me know what it is.

Brian.

 

[eagle1109]

I purchased TNY280PN switcher IC, I desoldered the one in the PSU.

The one was in the PSU has a short between the BP/M and S. The new one doesn't.
Moderator action: removed link to external server

But after the replacement, it still not working.

The two MOSFETs are:
1. TK15J60U, 600V, 30A
2. STW9NK95Z, 950V, 7A

They both have the same pin arrangement, from left to right:

1 gate 2 drain 3 source
Moderator action: removed link to external server

Moderator action: removed link to external server

I tested the voltages on the two transistors, they are both the same:
gate = 0V
drain = 320V
source = 0V

=======================================================================================

I found something on the SMDs, one of the diodes is shorted.
Moderator action: removed link to external server

Could it be the cause for the whole problem? Can an SMD diode stop everything! I think it's in parallel with the R55, they are on the gate rail.

But the cathode of the diode is not towards the MOSFET gate! Does it work as a protection to the gate?

- - - Updated - - -

The TNY280 is only a low power device so I doubt it has anything to do with the main power output but it could be used to generate the standby supply.

Try this:
Link the green and black wires at the output plug.
Read the voltage between any black wire and the purple wire, let me know what it is.
Read the voltage between any black wire and any orange wire, let me know what it is.

Brian.

OK, I linked the green/black

Then tested:
purple/black
orange/black

All 0V

 

[KlausST]

Hi,

Instead of inline inserting links to large - not very informative - photos you better should give:
* the exact type and manufacturer, of the device
* or (a link to) the datasheet

Klaus

 

[betwixt]

Thanks Klaus.

Diodes in the gates of power MOSFETS are often used to speed up the gate discharge and they have low value resistors across them so the only way to be sure is to remove the diode from the board before measuring it.

There is still no standby supply and it has to be working before the main supply will start up. The purple wire should have 5V on it even if the power is turned off so that's the area to concentrate on. Please consider that we do not have a schematic to refer to so comments like "R55" do not mean anything, the advice any of us give is generic to any PC power supply but may not be specific to the model you have there.

Next check - I'm assuming the TNY280 is supposed to be generating the standby 5V, measure the voltage between pins 4 and 5, it should be >300V. Be careful because it is on the AC line side of the circuit!

Brian.

 

[eagle1109]

1. You're right, I took out the diode and tested it, it's OK and returned it back. Still the short when it's soldered on the board suspicious I guess.

2. The TNY280 actually doesn't have voltage between drain and source!

 

[betwixt]

No voltage means it can't operate so something nearer the AC input is causing the problem. It may be that there is a 'PFC' (Power Factor Correction) circuit upstream that has failed but usually they still let enough power through that something shows, even if it is incorrect.

Some detective work is needed and I'm afraid without a schematic I can only give you general directions. Pin 4 (pin 3 is missing) should be connected to a transformer, you need to find the other transformer pin that connects to the same winding. Basically there are a few turns of wire around the transformer core and you need to find which pins they connect to. One is the TNY280 drain, the other is the power input side. WITH THE PSU DISCONNECTED FROM THE AC and after leaving it for several minutes to discharge, measure the resistance from the pin connected to the other transformer pins, you should find two pins that have a just a few Ohms between them. If you can't it means the transformer is faulty, if you can, the fault lies between the other pin and the big reservoir capacitor. I'm not sure what is between them but if you can send me a photograph of the underside of the board and mark where those two transformer pins are, I might be able work it out.

Brian.