Compaq PDP-121P - Baked MOV

[socketa]

Re: Compaq PDP-121P - Baked MOV

Looks like this PSU has a crowbar circuit that's getting triggered
But none of the output voltages are over spec
The PSU runs OK with just a hard drive, but a crowbar is activated if it's plugged into any motherboard (any 5v or 12v board)
What would be causing the crowbar to activate?

[socketa]

Re: Compaq PDP-121P - Baked MOV

Made an error on the schematic
Here is the correct version

After connecting the rail cables to ATX, CPU, and hard drive, and then plugging in at the wall:
Attaching a meter to the gate of the 12V MOSFET
There is initially no voltage
After a couple of seconds (i guess after the primary's charge up and the computer automatically turns on) the PSU fan clicks, and the gate voltage rises to 5.5V, and settles at 3.5V
If i, instead, attach a meter to where it says on my schematic "large toroid and heatsink-mounted component", the voltage rises to 4V and drops to 0V

So can we say for sure that there is some problem with the creation of gate voltage?

[socketa]

Re: Compaq PDP-121P - Baked MOV

I soldered the 12V rail MOSFET back onto the PCB and connected an analogue meter to it's gate terminal.
Then i plugged the rail cables into the motherboard and put the power cord into the wall socket
(i'm pretty sure that these Compaq computers start up straight away by themselves)
Then the voltage quickly rises to 5.5V and quickly drops to 3.5V, and stays there until i pull the wall plug out, and it quickly drops backs to zero after about 8 seconds (probably after the primarys discharge)

Pretty much the same result if i connect it to the other side of the 100ohm resistor

[socketa]

Re: Compaq PDP-121P - Baked MOV

Check your schematic, though - I don't think that circuit can work just like that, with the Gate only connected to the Source pin via a cap. There has to be something more.

- yep, thanks for pointing that out (makes sense, and i should have recognized that there was nothing, on my drawing, to turn the MOSFET on).
i missed a thin trace that was going from the Gate, to a 100ohm resistor, and then to pin 3 (cathode) of a KA431AZ Programmable shunt regulator, that's located on the add-on board - i'll update my drawing later
And i'll check the voltage at the cathode of this programmable shunt regulator.

So when you had the PSU power up OK with the other (working) PSU feeding into it, did you have that PSU's 12V rail connected to this PSU's 12.8V rail or just the 12V rail?

Just the 12V rail

[momaka]

Re: Compaq PDP-121P - Baked MOV

I tested the STP40NF03L MOSFET with the transistor tester, and it identifies it as an "N-E-MOS"
with addional information:
vt = 1.8V
Cg = 2.2nf
RDS = 0.2ohms
Uf = 632 mV +

Does that look OK?

Yes, looks pretty normal. If the TT tester was able to identify it, then that means the MOSFET is actually working.

I've traced out some of the circuitry and identified the components

Looks like the MOSFET between the 12.8V and 12V rail is used either as a linear regulator or just a simple pass-through device. Check your schematic, though - I don't think that circuit can work just like that, with the Gate only connected to the Source pin via a cap. There has to be something more. In fact, if that MOSFET really is to operate as a pass-through device (or a linear regulator for that matter), then the Gate must always be higher than the Source by at least as much as the Gate threshold voltage, V_th. And for that to happen, then the Gate has to be pulled up by an external voltage source that is higher than 12V.

with the theory that there is a problem with the 12V rail (that appears to be derived from 12.8V rail) ...
would it be OK (i.e. not cause any damage) to disconnect the MOSFET and then try and start it up, to see if 12.8V comes up?

Yes, you can try that test. The 12V rail will just not appear. That being said, I suspect the PSU will not start at all with the 12V rail missing (when you remove that MOSFET), as it probably is monitoring it, along with the 12.8V rail.

So when you had the PSU power up OK with the other (working) PSU feeding into it, did you have that PSU's 12V rail connected to this PSU's 12.8V rail or just the 12V rail?

[socketa]

Re: Compaq PDP-121P - Baked MOV

with the theory that there is a problem with the 12V rail (that appears to be derived from 12.8V rail) ...
would it be OK (i.e. not cause any damage) to disconnect the MOSFET and then try and start it up, to see if 12.8V comes up?

[socketa]

Re: Compaq PDP-121P - Baked MOV

I tested the STP40NF03L MOSFET with the transistor tester, and it identifies it as an "N-E-MOS"
with addional information:
vt = 1.8V
Cg = 2.2nf
RDS = 0.2ohms
Uf = 632 mV +

Does that look OK?

[socketa]

Re: Compaq PDP-121P - Baked MOV

I've traced out some of the circuitry and identified the components
The MOSFET isn't shorted, but maybe it's still faulty?
Could whatever it was that blew the primary side components also have damaged this secondary side MOSFET?

Just because i could, i removed the 4 pin CPU plug from the motherboard and powered it from another PSU, but it didn't make any difference.
So i get the impression that the problem is with the 12V rail, not the 12.8V rail

[socketa]

Re: Compaq PDP-121P - Baked MOV

...and perhaps it's useful to note that the resistance between the 12V, and the 12.8V, rails is 12.8kohms

[socketa]

Re: Compaq PDP-121P - Baked MOV

I replaced the coil with a jumper - but no change, as expected

So then i back-fed the 5V, and no change
Then I back-fed only the 12V, and the computer started
That was a nice surprise
With the computer running, I checked the back-fed 12V rail, and it's the same as the 12.8V CPU rail, at 11.65V
Could it be mere coincidence that both rails are the same voltage?

(Wasn't exactly sure about what you meant by connecting the 3.3V sense wire on both PSUs - did you mean that sometimes the manufacturers don't bother connecting the sense wire on some power supplies?)

So it looks like there is an issue with the 12V rail?
Perhaps something 'funny' going on between the 12V and the 12.8V rails?

[sam_sam_sam]

Re: Compaq PDP-121P - Baked MOV

That being said... and recently trying out an experiment with a non-working PSU that helped me find the problem... maybe this troubleshooting trick could work here too:
- Take a working ATX PSU and backfeed each rail into the non-working PSU, one at a time only, then power-up the non-working PSU too and see if it stays turned On.

So for example, let's start with the 3.3V rail: connect the 3.3V of both PSUs together and turn On the working PSU. Verify that 3.3V is back-feeding into the non-working PSU. Then turn On the non-working PSU with a load and see if it can stay turned On without shutting down. If not, repeat this experiment for the 5V rail (after disconnecting the 3.3V rail.) Then do the 12V rail and -12V rail. Since the 5VSB is derived from a different circuit, don't do this experiment for the 5VSB. Also don't connect any of the signal lines on the two PSUs, such as PS-ON or PG. However, for the specific test of the 3.3V rail, make sure the 3.3V "return" / "sense" (if there is one) is connected on both PSUs. Without 3.3V sense, 3.3V rail could be off-value on one or both PSUs.

If you do find that the non-working PSU stays turned On when one of these rails is connected, then there may be something wrong with the output of that rail on the PSU.

Can you please show a hook up diagram for how you hook up the two power supply board and give more details about how to do this test correctly

Thanks

[momaka]

Re: Compaq PDP-121P - Baked MOV

It's a long shot, but could broken stands that were going to the coil result in the power supply not getting, (and maintaining under load) enough voltage? Hence the computer shutting it down.

Probably not.
The DC voltages on your input caps looked good and didn't seem to drop out / dip when turning On the PSU... so I suspect this isn't the issue.
Nonetheless, it does need to be addressed just in case. Just because I don't *think* it is a *likely* issue doesn't mean that it can't be one.

Can i cut the yellow plastic tape, and reattach the wire to the coil? (hopefully there is no windings inside of it)

Yes.
Though there should be windings inside. This is the PPFC inductor, after all.

Would it be OK to remove it (i'm guessing that it's passive PFC) and replace it with a jumper?

Yes.

yes but the values less 1k ohm are an issue

Why would that be an issue?
I've seen a lot of PSUs with low-value load resistors... hence measuring much lower resistance on the 12V rail. So I don't really think the resistance alone can tell you anything.

Your reading should be very close or the same as the good one if both switching power supply are exactly the same model number and have the same exact ic chip on both boards

^ This.
If the PSUs were the same, I'd definitely try to see why there was such a wild mismatch between the resistances. But...

The power supplies are not the same

That pretty much invalidates the resistance comparison... or at least as far as showing any information that may suggest a problem.

i know that PSUs can have quite different values for minimum load resistors.

Exactly!
I suppose no harm in posting the resistances. Though when things measure above a few hundred Ohms on the output, then generally that means at least there is nothing wrong with the big power delivery components (i.e. rectifiers or their snubbers being shorted.) But it doesn't really amount to much of anything else. Open feedback resistors can essentially go completely unnoticed, due to their relatively high resistance value. So a resistance check on the voltage rail won't tell you if one went bad or not. This can only be found through careful component testing and perhaps a schematic or application diagram of the supervisor chip to understand what is going on.

That being said... and recently trying out an experiment with a non-working PSU that helped me find the problem... maybe this troubleshooting trick could work here too:
- Take a working ATX PSU and backfeed each rail into the non-working PSU, one at a time only, then power-up the non-working PSU too and see if it stays turned On.

So for example, let's start with the 3.3V rail: connect the 3.3V of both PSUs together and turn On the working PSU. Verify that 3.3V is back-feeding into the non-working PSU. Then turn On the non-working PSU with a load and see if it can stay turned On without shutting down. If not, repeat this experiment for the 5V rail (after disconnecting the 3.3V rail.) Then do the 12V rail and -12V rail. Since the 5VSB is derived from a different circuit, don't do this experiment for the 5VSB. Also don't connect any of the signal lines on the two PSUs, such as PS-ON or PG. However, for the specific test of the 3.3V rail, make sure the 3.3V "return" / "sense" (if there is one) is connected on both PSUs. Without 3.3V sense, 3.3V rail could be off-value on one or both PSUs.

If you do find that the non-working PSU stays turned On when one of these rails is connected, then there may be something wrong with the output of that rail on the PSU.

The only tricky part with the PSU from this thread is that it has a 12.8V rail... so not sure if we should be back-feeding this one or just the regular 12V rail. But I suppose you could try both. Nothing should really smoke or burn, so long as both PSUs are decent brands (i.e. have good working protections.) I would avoid using a "cheapie" gutless wonder for this test.

[socketa]

Re: Compaq PDP-121P - Baked MOV

Would it be OK to remove it (i'm guessing that it's passive PFC) and replace it with a jumper?

[socketa]

Re: Compaq PDP-121P - Baked MOV

I've now measured all of the rail capacitors that are on the the main board and they don't appear to be bad (forgot to put the other AUX rail measurement in, but it tested good as well)
While turning the PCB over, the black wire to that heavy (filter?) coil broke off (i assumed that they would be soldered on to the coil very well, and could handle the movement of me turning the PCB over and back over again)
This coil is between live and an AC terminal of the bridge rectifier
It's a long shot, but could broken stands that were going to the coil result in the power supply not getting, (and maintaining under load) enough voltage? Hence the computer shutting it down.

Can i cut the yellow plastic tape, and reattach the wire to the coil? (hopefully there is no windings inside of it)

[socketa]

Re: Compaq PDP-121P - Baked MOV

There is only two values below 1K
One is on the 5VSB and it's capacitor tests good (and i even changed it)
The problem arises when the computer grounds PS_ON and then (i suspect) that something on the computer motherboard shuts down the main rails as they are starting up - So i don't think that there is any problem with the standby rail.
The other value that is below 1K is on the 12.8V 4-pin CPU rail - which i'm not sure is acceptable or not; as a lower resistance might be normal for an independent CPU rail

[sam_sam_sam]

Re: Compaq PDP-121P - Baked MOV

yes but the values less 1k ohm are an issue

[socketa]

Re: Compaq PDP-121P - Baked MOV

The power supplies are not the same, hence N/A (non applicable) for the 12.8V (it doesn't have a separate rail for the CPU 4 pin connector)
i compared them just in case someone saw something that was terribly wrong with the faulty one
i know that PSUs can have quite different values for minimum load resistors.

[sam_sam_sam]

Re: Compaq PDP-121P - Baked MOV

Yes “ out-of-whack ”
12 volt ( is ?able )
5 volt SB***very low resistance not good
5 volt***something is very wrong with this resistance value on the bad power supply
3.3 volt***something is very wrong with this resistance value on the bad power supply
12.8 volt***very suspicious what this reading is about on the non working power supply
-12 volt***something is very wrong with this resistance value on the bad power supply
PSON***something is very suspicious what this reading is about on the non working power supply

PG to me might be close enough to the work power supply

All others are about are very low resistant and should be investigated very thoroughly

How were these resistance measurement made plus meter lead and negative meter leads were where on this switching power supply board

Your reading should be very close or the same as the good one if both switching power supply are exactly the same model number and have the same exact ic chip on both boards

[socketa]

Re: Compaq PDP-121P - Baked MOV

i compared the resistances of these two PSUs

_______Good PSU___Faulty PSU
12V_____0.85K_____12K
5VSB____9.7K______260ohms
PG_____ 1.35K_____ 1.25K
5V_____362ohms____1.2K
3.3V____27ohms____10K
-12V____3.65K_____11.86K
PSON___9K_________3.6K
12.8V___N/A_______234ohms

Anything look obviously out-of-whack?

[socketa]

Re: Compaq PDP-121P - Baked MOV

Am back home now,
Plugged the PSU into a well-working computer (dell dimension 2400, which has a 20 pin ATX, and 4 pin CPU connector), and also connected the HD, DVD, and Floppy (i.e swapped all of the connectors over)
This particular board normally starts as soon as the power is plugged in.
But all that i hear is a click in the PSU fan, and it shuts down pretty much instantly.
So, as suggested, i changed over the three optocouplers; but that didn't make any difference.
And there is a similar blinking front panel LED.
There is 4 diagnostic LEDs on the back panel, but none of them light up
So, it looks like there is some faulty condition within the PSU that is causing the motherboard to shut down the PSU (probably by pulling Power_Good low)
I watched the voltage of Power_Good when i plugged the PSU in, and it went from 0 to about 3V when the fan clicked and then back to zero, probably cutting out before it has a chance to reach it's normal value, (or maybe my meter's response time isn't fast enough)

Apparently the blinking amber light means:
Indicates system has power, but the POWER_GOOD signal is not yet active.

Maybe it is one of the other output caps
Is there some chip on motherboards that shuts down the PSU if ripple, or ESR, is too much?