Devolo fault

[throbscottle]

One of our ethernet over mains Devolo dLAN 200 AVdesk MT:2091 has developed a fault which looks like a low psu output. The lights on the front come on, but dim, and it doesn't show up on the network.

Mains plugs straight into the box for power and network to other devolo's, rj45 for connected device.

I'm kind of investigating and writing as I go along, so sorry if it's a bit disjointed...

The voltages coming out of the psu are: 1.2, 3.3, 12.2. I'm wondering if that 1.2v is supposed to be 5v, since it's smoothed by a 1000uF 10v cap. What does anyone think?

Theres no sign of swelling electrolytics and the semiconductors I can get to seem ok.

One huge symptom is a resistor on the primary side has been getting very hot, pcb is quite badly scorched. However the resistor is ok. It is connected in series with a capacitor not sure what type - large value ceramic perhaps, which measures ok. I don't understand what could cause this overheating though.

Switching chip gets warm. I notice the resistor doesn't get especially hot now, implying something was failing when it got hot, has now failed.

I lifted the schematic from the chip's datasheet, it's mostly similar to what's on the board. I added in the overheating resistor and series capacitor. Snubber perhaps? The point where they are connected doesn't appear to go anywhere else. The resistor is 470R, capacitor is unmarked but gives a slight kick on my meter.

The schematic is only to show the primary side - can't see what's going on with the secondary except it's a tapped winding with some hefty rectifiers 1N5822, and a smaller one, SB160 I think, though it's partly the same - I think the extras on the board are just voltage taps.

There's what appears to be a 6 leg NPN - ZXT10N20DE6T, in series with the 1.2v line, which is also fine. C~1.4v, E~1.2v, B~1.9v - looks like a classic series pass element. The collector voltage should be higher though.

If you look at the recommended layout on the TNY266PN's datasheet, it's very similar.

So should I start pulling electrolytics, even though they seem ok? Or should I be looking for something else drawing extra current?

Plan A so far - cut the lead of the rectifier (with 1.4v coming out) and connect a separate capacitor, see what the voltage is. Sound like a good plan to anyone? If it's 5v at least I'll know it's being pulled down.

(still looking up chips to try and identify what voltages are definitely required.)

Oh and it's a 6 layer board - it's got layer numbers in it.

Datasheet for smp chip: https://www.datasheetarchive.com/TNY266PN-datasheet.html

 

[alec_t]

However the resistor is ok. It is connected in series with a capacitor not sure what type - large value ceramic perhaps, which measures ok. I don't understand what could cause this overheating though.

Cap could be breaking down at high voltage? That would add DC current through the resistor. Was the cap tested at high voltage?

Edit: If two of the three power rails are ok then I'd be looking on the secondary side. The cooked resistor may just be collateral damage.

 

[throbscottle]

Not had time to look at it since posting. Yes, cooked resistor is collateral damage - actually there is nothing wrong with it and it doesn't get hot now - it is just symptomatic. So I doubt if the cap is passing DC at high voltage else the resistor would still be overheating.

I have been wondering if it could actually be a shorted winding on the transformer (hence the cooked resistor), perhaps that has since burnt out the shorted part and reconnected itself to another layer of the winding....

Maybe able to have another look this weekend...

 

[throbscottle]

Had the bright idea of testing voltages in the good one to compare, since they were bought as a pair.

Definitely not a psu fault, the voltage is supposed to be 1.2v. Compared voltages on pins of the chips for analogue and digital supplies, anything else I could find that wasn't a signal - no difference there. Compared voltages on a few transistors on the other side of the board, very little difference there. Voltages on the test points, same.

The Devlo contains an ethernet over mains controller, an AD9865 broadband modem, an IP101 ethernet transceiver, 2M of ram, OPA2674 wideband opamp, 12F5081 microcontroller

I noticed that the link light comes on when the ethernet cable is plugged in, so I guess the controller and ethernet transceiver are ok. The dLAN light comes on briefly when the unit is plugged in - I guess that's when it's looking for a data connection over the mains.

So I'm thinking somewhere in the area of the opamp which passes the signal on to the mains.

Any suggestions for my next line of attack? I'm pretty limited to just using a multimeter at the moment.

Cheers