Simple switching power supply repair help, for those that have some time.

[ezanolin]

Greetings all... I am now in my mid forties and a few years ago I decided to start up my childhood hobby of electronics again. This was mostly spurred on by digital electronics, something that I have always focused on, and by the rise of all these easy to use micros. Micro's are very close to devices like my early "PCs" of back in the day (Apple ][, ZX-Spectrum, BBC, etc), only much smaller, and more affordable, and not to mention faster. I always enjoyed working on digital electronics and have built and repaired a fair amount of digital stuff to date. I have however never been much good at the analog side of things and have pretty much used boiler plate designs where I needed to in the past. Which brings me to today...

I have decided to try my hand at repairing a PSU from an old Cisco catalyst switch that I had lying around. Traditionally, in this situation, I would check the fuses, diode bridges, MOV's, capacitors and any obvious component that has exploded, failing this I would usually just give up. So at this point, usually, If I really needed to repair it, I would buy an off the shelf PSU unit and butcher it into the chassis. However the switch is way too old and I don't actually have any use for it, but it does present me with an opportunity to fiddle. I have gotten part way into diagnosing it but have come stuck and am hoping some kind soul can point me in the right direction, and hopefully I will learn some things on that journey.

The meat of it...
The power supply ouputs 12V and is rated at 8.5A. The output connector leads mostly to the 12V and GND rail of the switch PCB(3 black wires and 3 white wires). There a 2 additional wires of interest that are orange and purple which are used for "PG" and "RPS-pres". This seems to be a "Power Good" signal to the switch, and the "RPS-pres" seems to be some kind of presence detection for the switch, it is just a 3.3K resistor to ground. This is probably to facilitate detection of a backup DC power unit that can also be plugged in.

The circuit seems to have 4 stages that I can identify.

• AC-DC rectification and smoothing
• Power factor correction
• Switch mode DC-DC regulation
• AC-DC rectification smoothing and feedback to 3

The story so far...
There does not seem to be any physical damage to the PSU (all the smoke is still inside). I have removed and tested the mosfets that I could find, and they seem to be OK. I am not sure how good my tests are, but mosfets tend to fail catastrophically when they go faulty in any case. The DC conversion portion of the PSU seems to be fine in so much that the rectifier bridge checks out, and I get about a 300VDC reading on my multimeter on V+.

I checked and the PFC IC in the first stage seems to be getting 10V on its VCC pin and I also read about 300V after the second stage choke/inductor/transformer(L010) , so I am assuming that everything in the second sage is OK too (assumptions assumptions). In the 3rd section I am only reading 1.2V on the DC-DC converter IC's VCC pin 7(IC300) and my thinking is that this is a problem.

The 1.2V seems to come as from the 300V rail through a 100K 2W resistor(R335), through a mosfet(Q121) and then a silicon diode(D121). Its actually 1.4V after the mosfet but the diode forward voltage drop seems to be about 0.2V.

I have tried to replace the above mentioned mosfet(Q121) to no avail, and I did not have much hope to begin with as the drain voltage was only 1.5V in any case. At this stage I do not think that the VCC for the DC-DC IC(IC300) is actually being generated by this section, but possibly from the winding pin 5 of the transformer(T120). This however does not sit well with me as I thought that the DC-DC chip would generate the switching frequency to run the transformer so how can its supply come from the transformer? I must be missing something.

I am not even sure where the 10V feeding the VCC of IC001 in the second stage is coming from either. It looks to be from the 300V DC rail through a 220K 1W resistor(R004) but I would expect a second resistor to ground to form a voltage divider circuit, and the second trace that could possibly form this part is diagrammed in the third stage through a 68K resistor and to the gate of the mosfet(Q121) that I was suspicious of. it also branches from there to a transistor(Q122) that goes to ground through a 7.5K resistor(R133) but I am not sure how this section works at all.

Luckily I was able to find a schematic for the unit online. I feel that it's not very well laid out but its a godsend regardless. I have attached it with some of my annotations scrawled on it. After a lifetime of staying away from this stuff I would really like a win, so if anyone can help me on this matter it would be much appreciated.

(excuse the GIF format but it compresses 3 bit indexed gray images better than PNG)

[duak]

I don't see any attachments, perhaps try again?

[MagicSmoker]

Yeah, I was going to say the same as duak - retry attaching that schematic.

In the meantime, it sounds like the trickle-charge startup circuit is being loaded down way too much, preventing startup. Likely the controller IC is fried, perhaps as a result of other components failures making their way back to it.

[ezanolin]

I have amended the post. Sorry about that. Initially I tried to attach 3 files larger than 5M, and after that, even though I removed them, the file size detection was all over the show. In the end I managed to merge everything into one file and attach that and posted. I even saw the browser uploading during the process, but something must have gone whacked after the first time I exceeded the file size limit. I guess I should have reloaded the page from scratch.

[MagicSmoker]

I stand by my earlier suggestion. Is R335 getting fairly toasty? If so, then the PFC circuit is working and Q121 is on, but being too heavily loaded (>3mA) for R335 to charge up C330 enough for IC300 to kick on, probably because IC300 is toast.

That said, IC300 is unlikely to fail all by itself, but if the PFC stage is working that rules out a shorted Q120...

[ezanolin]

 I had checked R330 before because there was browning around the PCB and the track had peeled off the subtrate a bit. I had already done some minor surgery to fix it up so the contact is no longer an issue but was not sure of the heating status. So I checked now and it is indeed getting toasty with a 300V drop across it.

[ezanolin]

I have in the past de-soldered IC300 and tested the VCC pin without the IC in place and had a similar VCC reading, in fact that's when I first noticed the low VCC.  If the IC is the faulty component, and it is drawing to much current would removing it not have allowed C330 to charge up properly?

[MagicSmoker]

I have in the past de-soldered IC300 and tested the VCC pin without the IC in place and had a similar VCC reading, in fact that's when I first noticed the low VCC.  If the IC is the faulty component, and it is drawing to much current would removing it not have allowed C330 to charge up properly?

Well, the only other possible culprits are a shorted C330 (very unlikely if an elko; they tend to fail open/high resistance), a shorted D330 (possible and most likely), or R133 being way too low in resistance and/or shorted (highly unlikely unless someone replaced it with the wrong value).

[xavier60]

I have in the past de-soldered IC300 and tested the VCC pin without the IC in place and had a similar VCC reading, in fact that's when I first noticed the low VCC.  If the IC is the faulty component, and it is drawing to much current would removing it not have allowed C330 to charge up properly?

With the IC out of circuit, the Gate can float up and blow the MOSFET, This can also happen with the IC in circuit if a fault prevents it from starting,
I fit a 10K Gate bleed resistor to the MOSFET if there isnt one already fitted.

[MagicSmoker]

There is a 1k resistor from gate to source. Though that reminds me, another possible culprit is if gate protection Zener ZD120 is shorted, though that typically only happens if the switch fails catastrophically.

[ezanolin]

Your advice is yielding some results, D330 is shorted. The schematic calls for a 1n4148 but the installed diode is a much beefier UG2D. I only have some 1n4001 lying around which I though I would try out as a temporary replacement. So now, when I power up the unit It trips my RCD(earth leakage switch) on my main board. I don't think the choice of diode is the cause of the trip (though I am happy to be wrong about it), but it does indicate an issue further down the line in the PSU.

[ezanolin]

Crikey, ok, full disclosure. I forgot to replace the mosfet(Q121) I was suspicious of . But worse than that, I had a short between pin 1 and 2 which I created when I last removed it. Gonna clean up and see but I fear I may have caused more damage

[ezanolin]

Doing the dance of joy over here!!! Replaced the mosfet (and undid the short obviously) and I am getting 12V on the output!!! Thank you so much everyone for the feedback and patience. I must admit I am still a bit in the dark about how this whole switching section works exactly, but I have definitely picked up some crumbs along the way. If anyone feels gracious enough to drag me through an explanation I would love it but I don't expect it. Thank you  MagicSmoker for your patience and thanks Xavier for highlighting mosfet protection

[MagicSmoker]

If you want to read up on it, power factor correction is done by a boost converter and the actual power supply is a conventional forward converter with reset winding. You can think of the latter as an isolated buck converter limited to a maximum of 50% duty cycle.

[ezanolin]

If I am not mistaken the diode we replaced was ensuring we only got half the wave out of the transformer and then that was smoothed out by C330 and used for the regulator VCC? The fact that it was shorted allowed the full wave through and meant the net voltage over time was 0?