EE "Brightbox" router. Not so bright now.....

[max_torque]

Came downstairs this morning, and immediately i smelt that familiar smell of toasted electronics  ;-(

Turns out my EE Brightbox router has let the "Magic smoke" out........

So, as i couldn't turn it on, i took it apart!





Looks like one of the dc-dc convertors has cremated itself.  There look to be two in parrallel, possibly one for 3.3v and one for 1.8v i'd guess.

More worrying, it looks like there has been a "costdown" carried out on the board, and a lot of footprints are unpopulated, including, what looks like to be the dc-dc freewheel diode!!!

In that pic, i'd expect D113 to be the free wheeling diode (it's a large footprint for a large diode) and there also seems to be a marked lack of filter caps around the dc-dc as well.


I wonder if EE are going to be getting a whole lot of these things back in the near future? (they are sending me out a new one btw, so good service from them).  Mine has been on and running pretty much continuously for  a couple of years now, so there may be a LOT of these devices approaching a similar failure  (wonder if it is an over voltage high cycle failure in the switching FET, as without a freewheel diode, that is going to potentially be seeing some voltage excursions etc?)



I've also been trying to work out what the IC is that has gone bang, it's labelled:

8958BG
C16931A

it's a soic8 package.  No sign of a seperate FET, so must have an integral power switch

Any guess anyone?
 

[McBryce]

It may have been very bright for a very short time, you just happen to have missed the "bright phase"

Will you be opening the new one to see if they've backtracked on the costdown?

McBryce.

[max_torque]

Will you be opening the new one to see if they've backtracked on the costdown?

McBryce.

heck yeah, and fitting those missing diodes if they aren't there!  (and possibly even a bit of extra filter capacitance too......)


If i could work out what the dc-dc IC is, then it'd be a simple repair to get the old one up and running again, as the board tracks look fine

[McBryce]

You'll be able to see exactly what it is when the new one arrives.

McBryce.

[madires]

My guess would be that the diode went first. A quick search found just https://www.fairchildsemi.com/datasheets/FD/FDS8958A_F085.pdf which is a N and P channel FET combo in a SO8.

[McBryce]

Acoording to Max (and his picture seems to confirm it), the diode was never fitted at all. It didn't fail, it was never there!

McBryce.

[amyk]

The MOSFET has its own body diode. Perhaps that's why they decided to omit the external one?

[max_torque]

The MOSFET has its own body diode. Perhaps that's why they decided to omit the external one?

Yup^^ looks like this is the case.

New router turned up, diode also left unpopulated.  Stuck my scope on it and it doesn't look too bad, with approx a -2v negative spike, then abput 80nS at -0.9v during the switchover deadtime.  Once the lower Fet is switched on, as you would expect, the input to the inductor climbs to just about -0.1V.

In terms of deadtime, the low side body diode is conducting for about 20% of the low time, so there will be significant amount of energy going through that diode at the "High loss" condition.


Turns out the dc-dc that went is the 3.3 to 1.2V convertor, so i can probably hack something into the device to provide that 1.2v rail and i'd have a "spare" router  (for when the new one goes bang......    )

[Circlotron]

In terms of deadtime, the low side body diode is conducting for about 20% of the low time,

Having a body diode conduct can be bad news in a switcher. When the lower side mosfet switches on, the thousands(?) of individual parallel diodes stay in forward conduction mode for a certain amount of time because of stored charges in their pn junctions. When the lower mosfet goes off and the upper ones on, these diodes start conduction again in the reverse direction as the charges get swept out and they all switch off at slightly different intervals, forcing less and less diodes to conduct the now shoot-through current. Pop just one of those tiny, tiny diodes and  it's all over, red rover. Much better to have a schottky in parallel and avoid this situation altogether.

[Kjelt]

Just another thought i see switches and routers getting problems due to the external powersupply.
It is usually cheap crap with elcos to last two years (warranty period).
So be sure to also check the external powersupply