VOGONS


Reply 60 of 73, by kotel

User metadata
Rank Member
Rank
Member

Okay with an 1k in the voltages are down to 5.5v. Still to high though.... And besides now there's an funny smell inside. I cannot describe it but it's like plastic melting or something like that but veery light. Maybe it's from the outside but I can only smell it when there's power to the PSU.

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 61 of 73, by momaka

User metadata
Rank Oldbie
Rank
Oldbie
kotel wrote on 2025-03-06, 13:35:

As for the opto coupler I'll just replace it with a new one. Have like 20 of them I forgot about😆

The big question here is what was the old optocoupler and what was the new one that replace it.
There's standard "817", then there's the "817c", and then there are many others too.
The LED inside these may react a little different and require either more or less current to drive... which if not selected right will do exactly what yo saw: output too high.

kotel wrote on 2025-03-06, 16:58:

Actually, on a second thought this doesn't sit right with me...
The PSU after replacing all the bad parts for the first time had spot on 5vsb (~5.17v). Then something screwed up fully and now its 5.8v after replacing the PS106R (mine started with FR10 and the rest was burnt away) with FR107. I don't think its gonna be the voltage increase by 100v.

The change in the voltage rating of the snubber diode is irrelevant. All the snubber diode does is it allows the 5VSB transformer's primary winding to "push out" current "somewhere" when the MOSFET closes up. So when this current is pushed through D5, it then gets stored in C8, and parallel resistor R61 "slowly" dissipates it over the time between the next switching cycle. The whole purpose of the snubber circuit is to protect the MOSFET from the inductive kickback of the 5VSB transformer's primary side, which can generate very high voltages and kill components. Other than that, it play no role in the voltage regulation of the circuit.

Tiido wrote on 2025-03-06, 17:06:

It is possible there's a damaged voltage reference component in the feedback loop or maybe there is a need to adjust one of the feedback resistors due to different replacement components etc.

Agreed.
And I suspect it's the new optocoupler that might be the smoking gun here. Probably needs a higher driving current to turn ON its LED in order to drive the MOSFET on the primary (via small transistor Q7) harder into the OFF state to reduce voltage.
Resistor R13 (looks like 390 Ohms on my PSU) is responsible for limiting the current going to the optocoupler LED. If you compare this to the circuit of the HEC Orion HP585d 5VSB circuit I put several posts above, you will see that one uses a 150 Ohm resistor.

So with that said, you can try replacing R13 with something like 330 Ohms first (orange orange brown gold), and if that still keeps the 5VSB too high, try 270 Ohms (red purple brown gold) or 220 Ohms (red red brown gold) or even 200 Ohms (red black brown gold). See how the 5VSB voltage changes with these and report back.

Tiido wrote on 2025-03-06, 14:53:

You need some big enough load on the output for voltage to get in spec, i.e a few LEDs or something like a 100 ohm resistor. Unloaded SMPS outputs are rarely if ever in spec.

PSU 5VSB circuits doesn't require any external loading (i.e. anything extra that's not already inside the PSU). That said, most include a load resistor on the output to have proper regulation. This resistor is usually in the range of 10 to 470 Ohms... with 10 Ohms being really waaay too low and wasting too much power / producing too much heat. But 47 Ohm to 100 Ohm is acceptable (and what I usually put on my PSUs with 2-transistor circuits.)

kotel wrote on 2025-03-06, 17:21:

Okay with an 1k in the voltages are down to 5.5v. Still to high though.... And besides now there's an funny smell inside. I cannot describe it but it's like plastic melting or something like that but veery light. Maybe it's from the outside but I can only smell it when there's power to the PSU.

OK, here's a simple-ish way to find if anything is overheating.
First, solder wires to the primary side 200V caps so that you can measure the voltage across them without having to poke at them with your MM.
Next, power up the PSU and let run for a few seconds or longer... preferably until you can start to smell the burned plastic smell again.
Now remove AC power from the PSU (via unplugging from the wall) and look at the DCV reading of the primary caps. As soon as the voltage falls below 50V total (i.e. under 25V on each cap), feel free to start probing around with your finger at various parts in the 5VSB circuit to see what's running hot. Only do this with one hand while the other is not touching anything on the PSU, just in case. With the primary caps discharge, there won't be any component to hold energy and shock you, but nevertheless, it's a good safety practice. Now see if you can see what component(s) are running hot.

Reply 62 of 73, by kotel

User metadata
Rank Member
Rank
Member

So I've replaced the R13 (only one near opto coupler which was 390 ohms) with an 330ohm and the voltages are bellow spec (4.56-ish volts). Which is a good and bad thing 😀 I'll try an 360 ohm next when the PSU discharges. If that doesn't help then idk what will. It didn't help, only made this worse (4.2v now)... It seems this PSU "fixed" itself overnight and now the 5VSB generates proper? Or maybe it's because I'm doing cold starts and a lot less than yesterday. Probably latter.
As for the burning smell, one of the fuse wires started to melt near the fuse box I've soldered on... It weren't some cheap thin ones, an thick 22 gauge wire IIRC. I think this is the reason for that smell, but why and how could it start melting? I have melted some of the insulation but not this much....

Oh and I didn't replace the 817 with "C" or anything at the end. I just dropped in an normal PC817 from the LC which "seems" to work fine-ish.

Last edited by kotel on 2025-03-07, 17:01. Edited 1 time in total.

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 63 of 73, by momaka

User metadata
Rank Oldbie
Rank
Oldbie

🤣 that ain't right either. 390 Ohms --> 330 Ohms is about 15% reduction in resistance, but 5.8V to 4.6V is a little more than 20% reduction, which doesn't make much sense. R13 value shouldn't be that sensitive at all, once there's enough current to drive the LED in the optocoupler. So might need make sure the new opto matches the old one (if the old one is a "C" version, then look for another opto with "C" as well), or I suspect the PSU won't regulate properly with any kind of load on the 5VSB.

That said, you can test this theory by putting something like a 10 Ohm 2W or 3W resistor load on the 5VSB to pull about 0.5 Amps (not soldered, just attached to the 5VSB output wires). With that, see what happens to the 5VSB voltage. Again, I suspect you will find that the no load and load (10 Ohm resistor) voltages change too much between each other. FWIW, the 5VSB should be close to 5V anywhere between no load and all the way down to 2 Amps. Most of the PSU's I've tested tend to sag to no less than 4.9V usually (with 2 Amp load), and pretty close to spec with 1 Amp or less.

Anyways, keep working at it, we are getting there. 😉
As I mentioned before, 2-transistor self-oscillating circuits are quite challenging to get fixed/working once something has gone wrong, especially if you're new to analog electronics / PSU repair.

Last edited by momaka on 2025-03-07, 17:05. Edited 1 time in total.

Reply 64 of 73, by kotel

User metadata
Rank Member
Rank
Member
momaka wrote on 2025-03-07, 16:59:

🤣 that ain't right either. 390 Ohms --> 330 Ohms is about 15% reduction in resistance, but 5.8V to 4.6V is a little more than 20% reduction, which doesn't make much sense. R13 value shouldn't be that sensitive at all, once there's enough current to drive the LED in the optocoupler. So might need make sure the new opto matches the old one (if the old one is a "C" version, then look for another opto with "C" as well), or I suspect the PSU won't regulate properly with any kind of load on the 5VSB.

That said, you can test this theory by putting something like a 10 Ohm 2W or 3W resistor load on the 5VSB to pull about 0.5 Amps (not soldered, just attached to the 5VSB output wires). With that, see what happens to the 5VSB voltage. Again, I suspect you will find that the no load and load voltages change too much.

For the opto I have no idea what mine was. There was nothing I could tell from the cracked in half with scorch marks all over it, so I depend on DAVE86's PSU (which is most definitely an older rev). I guess 360 ohms would be next in line then?

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 66 of 73, by kotel

User metadata
Rank Member
Rank
Member

Okay I have tried with the 360 and 390ohm and no change at all..... Still 4.2v. Turns out on the old 390ohm resistor was damaged by the 5vsb carnage and now it measures 50k ohms laughing I have no idea what is worng now. Go higher with the resistance?

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 67 of 73, by kotel

User metadata
Rank Member
Rank
Member

Hmm, it looks like my 817 might be a C variant. There's the producer, date of manufacturing, 817, F43 and lastly C. Anybody here that knows if this one is an 817C or is it an normal 817?

Okay from TRW server I got the info that it should be an normal pc817.
I have replaced it with a new SHARP PC817 but same results. Maybe even worse cause now the 5vsb starts dropping quite fast (more or less .01v per 2-3 seconds). That might also be due to me leaving it on for longer than before (I also saw it drop with the previous 817 but I cut the power very early due to the voltages being under speced).

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 68 of 73, by kotel

User metadata
Rank Member
Rank
Member

While messing around with this one I noticed that after the voltage drops around .30v there's an weird funny smell inside (not the one before!). If I go over the 5VSB arena I can feel with my face the heat dissipation from that spot.
My idea is, before poking inside with my finger when the caps go bellow 50VDC, to desolder the 5VSB transformator and see if the heat dissipation (more like heat radiation(it's that powerful)) is still present.

EDIT: I have applied power to the PSU without the 5VSB transformator and there's no funny smells and no heat dissipation from that arena after around 1-2 minutes, so fairly confident the transformator is at fault.
The closest I could find was an EE-16-2005 half the size of the original transformator (Casecom PSU IIRC) with 1 extra pin and another one in the wrong spot. I could also try to wire an EEL-16-2005 (almost 1:1 replacement) from the solder side but I don't think the primary side from LC b350w will match the one on my FSP....

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 69 of 73, by kotel

User metadata
Rank Member
Rank
Member

Question, on the original FSP 5VSB transformator is it normal for the pin connected with the yellow wire to show 0.3-5ohms to the one near it out of circuit? It doesn't have any passthrough to the 2 pins right next to each other (which both have ~1.8-2ohms). I have found that the smaller EE-19-2005 also has close to the same readings.

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 70 of 73, by kotel

User metadata
Rank Member
Rank
Member

I have soldered in the EEL-19-2005 transistor and the EE-19-2005 one using wires and double checking the pinouts and I still get absolutely nothing on the 5vsb rail and sec. aux. rail.
I have no idea what is wrong here...

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 71 of 73, by kotel

User metadata
Rank Member
Rank
Member

Since it looks like I am nowhere near repairing the 5VSB circuit I decided to go with an 10V psu for sec aux rail and an 5V regulator using TL431 from the datasheet using an BD135 as the NPN transistor and potentiometers for the resistors.
Sadly I get 1.71V on the output.... I have double checked the connections, all go where they should. What's weird is when I touch all the pins on the transistor with my finger the voltage spikes up to 5.35v!
Any ideas?

Scratch that, I forgot I could use an 5V VRM module.
With an hahn 230-10v transformator, bridge rectifier (with it's filter cap) and an 7805 VRM I now get both 10VDC and 5VDC. The next step would be to solder it temporarily inside the PSU just to test. And if everything will work I'll make room in the 5VSB arena to fit the new PSU.

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 72 of 73, by kotel

User metadata
Rank Member
Rank
Member

And scratch that, again.... The transformator is rated for max 200mA (2VA) which is not enough for 5vsb. I do have an 450mA 12v transformator, but that one is on the brink of minimum amperage rating on 5VSB....

"All my efforts were in vain...
Let that be my disappointment."
-Kotel

Reply 73 of 73, by kotel

User metadata
Rank Member
Rank
Member

Okay I have found an 1A toroid that generates 11.5V on output and made an 5V regulator for 5VSB and 10V regulator for sec aux rail using the 7805's.
After putting everything together (including bridge rectifier) I can now power the PSU on no problem. Sure, there's an toroid case sticking out of the back but who cares, the PSU lives!
Before I summarize the repair I'll have to try it without dim bulb and toaster protection. Right now the voltages are stable with 4 3.5" HDD's, 2 fans and an nf610 board (without CPU power).

Everything seems to be working so far, so this is another PSU saved from the trash 😀
Now onto the excellent power...

Attachments

"All my efforts were in vain...
Let that be my disappointment."
-Kotel