Cisco 877 router - doorstop?

[pedro]

A customer presented this out-of-warranty router as a "can you have a look at this ... ?". Symptoms: when power applied, all LEDs blink in unison about twice a second, no other functionality. History: had worked great until a thunderstorm, was on a APC UPS at the time.

We opened it up and - apart from lots of nice Nichicon caps - there were two evil CapXon 6800/6.3V with characteristic bulging. ESR measured 1.4 and 1.5 ohms, so replaced them with Panasonic FM's. (All Nichicons were close enough to ESR spec.) No change in behaviour.

The board has a 5A fuse on the 5V input, which is intact. Stuck the CRO on the 5V rail and found it wasn't getting up. It would get to about 2.7V and then collapse, repeating as per the LEDs. This router is supplied from a dual voltage brick, rated 5.2V @ 4.4A and 12V @ 650mA. Independently loaded up the brick and it was smooth as, no signs of distress. The inference is that the board is overloading the SMPS brick.

So somewhere on this board is effectively a dead short. But with schematics of Cisco gear being unobtainium sheltered behind NDA's in their service depots, it's doubtful I can progress this.

(I have ensured that the owner knows which is his left shoulder, so he knows where this eventually goes.)

Have attached a couple of pics of the power area of the board. Any comments?

[Radio Fox]

Re: Cisco 877 router - doorstop?

Have you actually buzzed out the 5V line to confirm that you have a dead short?

If there is, then the first component I would look at is D12. I suspect it is a transient voltage suppresor connected across the 5V line.

I have seen a couple of computer monitors which use external PSU's, where the suppresor has gone s/c.

[pedro]

Re: Cisco 877 router - doorstop?

Have you actually buzzed out the 5V line to confirm that you have a dead short?

If there is, then the first component I would look at is D12. I suspect it is a transient voltage suppresor connected across the 5V line.

I have seen a couple of computer monitors which use external PSU's, where the suppresor has gone s/c.

{When I said dead short I referred to a fault sufficient to overload the 4.4A output of the brick.}

D12 is before the switch, the fault is after it.

I had briefly suspected D11 (up near the Pana FM's) until I traced out the tracks. It is the (schottky?) after L10 driven off the switcher U37(LTC3414). There's no other likely suspects that I can identify as potential rail shunts except a zillion bypass caps ...

The 12V input rail going into the 5V reg U18 (left rear of FM's) is echoing what the rest of the identifiable rail is doing - it rises to 10V and collapses. U18's output is a clean 5V when the input is high enough to allow, and its output feed the 3.3V reg which shows a similar waveform. There is no heat detectable at U18 or the 3.3V reg U32 (right rear of FM's).

Doing a thermal check, nothing topside gets hot after several minutes running BUT the pcb under U28 (ST chip marked TS616 EB603 afaict) gets decidedly uncomfortable to touch. This device is in the midst of the TDK and UMEC transformers.

[pedro]

Re: Cisco 877 router - doorstop?

Doing a thermal check, nothing topside gets hot after several minutes running BUT the pcb under U28 (ST chip marked TS616 EB603 afaict) gets decidedly uncomfortable to touch. This device is in the midst of the TDK and UMEC transformers.

TS616 is a dual wideband op amp with high output current:

https://cdn.badcaps-static.com/pdfs/...549de35714.pdf

[smason]

Re: Cisco 877 router - doorstop?

What if you fed it with a different power supply, like a bench supply that you could current limit, without it shutting down? You could get measurements more easily, and perhaps find a hot component.

Or go for broke, give it full current and watch for magic smoke

[pedro]

Re: Cisco 877 router - doorstop?

What if you fed it with a different power supply, like a bench supply that you could current limit, without it shutting down? You could get measurements more easily, and perhaps find a hot component.

Fair thought. Unfortunately my bench supply with current limit is single output, and I'd need to feed the +5 and +12 to the DUT. So one rail wouldn't be "smoke-free" ...

Or go for broke, give it full current and watch for magic smoke

Make that "expensive magic smoke". Nah, I'd rather return it to the client without any added damage.

Incidentally, the heat under the TS616 is because of the exposed thermal pad chip package.