Compaq PDP-121P - Baked MOV

[socketa]

Re: Compaq PDP-121P - Baked MOV

Thanks for that explanation
Yes, the resistor colour looks more brown than red when it's not in the sunlight, and i found another of the same value, but 14mm, which i heat-shrinked and installed.

Still, run the PSU through the series bulb / current-limiting device for a bit to verify the new IC didn't get damaged from the open Source resistor... and on that note, do check if the resistor between the Source resistor and CS pin is still good.

It's covered in glue, and it measures 2Kohms


Turned on the PSU, with PS-ON grounded, and the fan twitched slightly.
Looks like some protection is kicking in?
I have to wait about 10s, in between removing the paper clip and reinserting it, to get a response
The rail voltges come up, but then immediately drop off.
My maximum observed reading are:
3.3V rail : 3.7V
5V rail : 2.3V
12V rail : 3.9V

It wouldn't be because of the replacement MOSFET that i used in post #16, would it?

[momaka]

Re: Compaq PDP-121P - Baked MOV

V_ref pin is 5V - so that's nice to see, thanks
and i get 28K ohms from MOFET source to ground, and there is a suspect looking resistor in between, that looks like should be 0.28 hms

Looks like a 1.8 Ohm to me (and that's a standard value for E12/E24 series 4-band resistors.)

There is 0V on the voltage feedback pin - I'm guessing that that is because the MOSFET is not being turned on, and therefore there is no feedback from the secondary to regulate the UC3845 chip?

No, the feedback pin (2) goes inside the 384x to an error amplifier. So V_fb should be around 2.5V when the IC/PSU is regulating normally. When the voltage on the secondary side goes slightly higher than what it should be, the supervisor IC sends a signal through a 2nd optocoupler, which then (attempts to) increase the voltage on the V_fb pin above 2.5V. This tell the UC384x to decrease its PWM output, which in turn lowers the power going into the primary and reduces the voltage on the secondary side. If the reverse happens - voltage are too low on the secondary - then the supervisor does not send any signal through the 2nd optocoupler. This (attempts to) decrease the voltage on the V_fb pin below 2.5V... which tells the UC384x IC to switch "harder" (i.e. increase PWM duty output.) When that happens, more power is sent through the primary side MOSFET and transformer, thus increasing the voltages on the secondary side too. So in essence, the supervisor is constantly bouncing around between telling the UC384x IC to increase and decrease PWM output... but it all happens so quickly, that it looks as if the voltages on the secondary side are constant.

And that's how regulation is achieved in these PSUs, really.

Is that resistor a fusible resistor?

Likely yes.

The resistor is likely a metal film flameproof.

^ This.
Source resistors are always metal film / metal oxide and flameproof/resistant. DO NOT use carbon film type. Ceramic wirewound may also be a problem due to higher inductance.

There is usually another resistor between the mosfet source and the drive ic's CS pin. When the source resistor goes open it is very likely the ic has been damaged.

Could be.
Though damage most frequently comes from when the MOSFET's Gate shorts to either Source or Drain and then backfeeds high voltage into the driver output pin (6).

i could only find a 0.15 ohm heat-shrinked, grey-bodied, non-glossy, 14mm, resistor - is it OK to use that? (the 328 transistor tester shows it to be about 0.20 ohms, after subtracting 0.17 ohms when shorting the test leads)
The original looks like it's a 0.28 ohms and is about 11mm.

Yeah, that should work fine.
Again, I suspect your old resistor is 0.18 Ohm (brown, gray, silver, gold) and not 0.28 Ohms (red, gray, silver, gold.) Perhaps the brown/red band on your resistor is a bit lighter, hence possibly why you think it's red.

But even if it is 0.28 Ohms... the 0.15-Ohm should work OK too. Only difference will be during an overload / over-power condition: lower resistance will allow more power to go through before the IC detects the overload condition (which may or may not be detrimental to the rest of the circuit, depending on how over-built it is.) That said, if your resistor really is 0.18 Ohms and you put in 0.15 Ohms... that's about 15% difference roughly... so you can expect the overload limit allowed by the IC to be increased by roughly just as much.

Given that we are dealing with a HiPro PSU here, I don't think that's going to be a problem, as they tend to be built quite well.

Is there any significant difference between the pink-bodied and the grey-bodied resistors that you find on ATX PSU boards?

None.
I've seen them used interchangeably. Both are metal film type / metal oxide - you can tell by the non-shiny "eggshell" finish. On the other hand, Carbon film resistors are usually shiny.

Is it good to permanently replace the pink resistor with the larger sized grey resistor?

Yes, should be OK.
The bigger replacement one means it's also rated for a higher power, so it should run a tiny bit cooler... though not that this should matter. I suspect the original one (11 mm) is rated for 2 Watts, and your replacement (14 mm) is rated for 3 Watts. Again, not that it should matter, but just stating this for information completeness.

Yeah, i've replaced the controller chip and the MOSFET, they were, as mentioned previously, faulty

Still, run the PSU through the series bulb / current-limiting device for a bit to verify the new IC didn't get damaged from the open Source resistor... and on that note, do check if the resistor between the Source resistor and CS pin is still good.

[socketa]

Re: Compaq PDP-121P - Baked MOV

Yeah, i've replaced the controller chip and the MOSFET, they were, as mentioned previously, faulty
Is it good to permanently replace the pink resistor with the larger sized grey resistor?

[PeteS in CA]

Re: Compaq PDP-121P - Baked MOV

A low value current sense resistor in a MOSFET Source circuit probably won't open up or go high value on its own. That resistor probably received really high current, such as the MOSFET shorting or the transformer core saturating. I'd check the health of the MOSFET. If the MOSFET is shorted D-G-S, then high voltage got into the Gate circuit.

[socketa]

Re: Compaq PDP-121P - Baked MOV

i could only find a 0.15 ohm heat-shrinked, grey-bodied, non-glossy, 14mm, resistor - is it OK to use that? (the 328 transistor tester shows it to be about 0.20 ohms, after subtracting 0.17 ohms when shorting the test leads)
The original looks like it's a 0.28 ohms and is about 11mm.
Is there any significant difference between the pink-bodied and the grey-bodied resistors that you find on ATX PSU boards?

[R_J]

Re: Compaq PDP-121P - Baked MOV

and i get 28K ohms from MOFET source to ground

This is way too high for a source resistor, they are usually less than one ohm. The resistor is likely a metal film flameproof, There is usually another resistor between the mosfet source and the drive ic's CS pin. When the source resistor goes open it is very likely the ic has been damaged.

[socketa]

Re: Compaq PDP-121P - Baked MOV

i wouldnt be so sure.
see if the pints on the board where the wires go are labelled.

i don't think that drinking ale and working on PCBs go well together.
But yeah the labels on the board confirm that the connector pinout is the usual one.

Is that resistor a fusible resistor?

[stj]

Re: Compaq PDP-121P - Baked MOV

(i'm farly sure that the ATX pinout is normal, as they have just used different colours

i wouldnt be so sure.
see if the pints on the board where the wires go are labelled.

[socketa]

Re: Compaq PDP-121P - Baked MOV

V_ref pin is 5V - so that's nice to see, thanks
and i get 28K ohms from MOFET source to ground, and there is a suspect looking resistor in between, that looks like should be 0.28 hms

There is 0V on the voltage feedback pin - I'm guessing that that is because the MOSFET is not being turned on, and therefore there is no feedback from the secondary to regulate the UC3845 chip?

[momaka]

Re: Compaq PDP-121P - Baked MOV

The voltage values in the last post, are the pins of optocoupler as they appear in the photo, and the dotted lines is the gap between the primary and the secondary sides.
There's 130mV at Vcc of the UC3845 when the power is turned on and the 5VSB (green wire) has the correct voltage,
and when i short the white wire to ground there is 13.5V on Vcc.

So the UC3845 should be working / sending out pulses then, as according to the datasheet, minimum startup voltage is 8.4V, and minimum running voltage is 7.6V for Vcc for UC3845 and 3843 models. Check voltage on V_ref pin of UC3845, just to confirm - should be 5V. If it is, check low-resistance Source feedback resistor connected between primary ground and Source of the MOSFET - is it open? If not, check back voltage on V_fb pin on UC3845 too so we can see if the UC3845 is told to switch or just held down to not output anything.

[socketa]

Re: Compaq PDP-121P - Baked MOV

The voltage values in the last post, are the pins of optocoupler as they appear in the photo, and the dotted lines is the gap between the primary and the secondary sides.
There's 130mV at Vcc of the UC3845 when the power is turned on and the 5VSB (green wire) has the correct voltage,
and when i short the white wire to ground there is 13.5V on Vcc.

[socketa]

Re: Compaq PDP-121P - Baked MOV

Ignore my last post - i somehow that wrong, including the wrong transistor

The optocoupler is not shorted.

The values in normal font are when the PSU is powered on
The values in bold font are when the paper clip shorts the white wire to ground (i'm farly sure that the ATX pinout is normal, as they have just used different colours)

Secondary side

4.96V 5.28V
0.037V 1.18V
-----------------
-----------------
0V 15.4V
12.92V 13.18V

Primary side

[R_J]

Re: Compaq PDP-121P - Baked MOV

If there is 2.5 volts on each pin of the optocoupler on the secondary side, the difference in voltage is 0 so the internal diode is not lit and the primary side will act open circuit so no vcc
The secondary side should be supplied by the standby circuit, 2.5v seems low so check if the optocoupler is shorted, the sec should check like a diode. The secondary of the optocoupler will be controlled by the power supply ON/OFF signal

[socketa]

Re: Compaq PDP-121P - Baked MOV

I powered it up but have no power to Vcc on the UC3845
Vcc goes to an optocoupler via a BJT transistor.
The optocoupler (U3) has 16V on one pin of the primary, and nothing on the other, and 2.5V on both pins that are on the secondary side.
The transistor (222A NPN) is just above the UC3845 chip in the photo.

[socketa]

Re: Compaq PDP-121P - Baked MOV

if it's shorted to the gate then you need to check the gate-drive circuit too.

That's only a couple of resistors, 20 and 200K - they test good.
I replaced the MOSFET with the other one that i mentioned
Good to to power it up now with a bulb?

[stj]

Re: Compaq PDP-121P - Baked MOV

if it's shorted to the gate then you need to check the gate-drive circuit too.

[socketa]

Re: Compaq PDP-121P - Baked MOV

Thanks for the confirmation.
That Vcc cap tests good
So i desoldered the main mosfet and get 14 ohms across gate and source - so looks like that's baked as well
This mosfet is rated at 800V 1.5Ω 8A -Is it OK to replace it with a STW8NC80Z N-CHANNEL 800V 1.3 Ω 6.7A Zener-Protected PowerMESH™III MOSFET?

[R_J]

Re: Compaq PDP-121P - Baked MOV

The ic will be the same, the only difference is the N suffex which refers to the package (DIP). I suspect there will be a small electrolytic on the vcc line, I would replace it as well.

[socketa]

Re: Compaq PDP-121P - Baked MOV

Put the primarys back in (a small bit of the sleeve came off when removing the tough white glue, so i just put plastic masking tape over it), and the replaced the MOVs
60W lightbulb 'fuse' briefly lights, then goes out
5VSB auxiliary rail (green wire) is working good
But it doesn't jump start it with a paper clip between the black and white ATX wires (no voltages on the other three rails)
There is a 5H0265RC Fairchild power switch (for the 5VSB?) and a 2SK2648 N-Channel MOSFET (for the main PSU?). There is no shorts on the fairchild power switch, and i get 13ohms across the MOSFET Gate/Source, so that's not a short either
Tried connecting a hard drive, but it still wont jump-start
There is 158V and 155V across the primarys and 313V across the bridge rectifier DC legs
This was all done with the lightbulb in place of the fuse
The main controller is the 8 pin UC3845B IC
I removed it, and there is 3 ohms between pins 5 ground and 7 Vcc
I luckily found a UC3845BN and checked the same pins and there is no contact between them
Should i replace it with that one, since i can't perceive any difference on the datasheet between the two ICs?
Anything else that i should check before replacing it, so that i don't damage the replacement controller?

[PeteS in CA]

Re: Compaq PDP-121P - Baked MOV

I've seen ceramic cased fuses that blew apart, though the fuses were SMD types. They blew apart due to the current briefly continuing to conduct by arcing. A longer ceramic fuse might not have arced quite as long, and with less dramatic results.

A higher rated energy MOV will handle higher energy single hits without catastrophic failure, though the same kind deterioration will still occur.