Sun 3/80 Magic Smoke

[Curtis H. Wilbar Jr.]

Tony Duell wrote:
>>
>> Tried to power up a Sun 3/80 mainboard...
>>
>> Inductor L0500 (right near the power input) (I assume this is an 
>> inductor by the L0500
>> marking on the PCB) smoked....  (pink smoke no less)
>>     
>
> This sseems to be the day for smoking indcutors :-)
>
>   
>> Anyone familiar with the 3/80 to give an idea of why this might have smoked.
>>     
>
> I've never seen this machine, but from the location of the component, and 
> the fact that it's an inductor, I would asseum it was part of a DC power 
> line filter, and it smoked because of a short (to logic ground, most 
> likely) 'downstream' of said component.
>
> My first suggestion is to desolder the indcutor from the PCB. Then trace 
> one of the connections back to the PSU input connector, which will tell 
> you which power line it's associtated with. If you're lucky it'll be 
> something like a +12V rail, not used by many components. If you're 
> unlucky, it'll be the 5V rail that goes everywhere...
>   
I'm unlucky... it is 5V.

> Msot likely you'll find the 'other' side of the inductor postion on the 
> PCB -- the one not conencted to the power connector -- has a low 
> resistnce to ground.
>
> There are many 'techniques' for finding such shorts. Pulling all socketed 
> devices is a good first step, if the short goes away, you know it's one 
> of the chips you've pulled. Suspecting tantalum capacitors is another 
> good idea :-). More logical is the idea of applying a low voltage between 
> the output side of hte inductor and ground, so that a mdoerate current 
> flows (I would guess areound 1A would be OK here), and using a sensitive 
> millivoltmeter to measure the voltage drop across components. The shorted 
> one will test lower than all the others (due to the drop along the PCB 
> tracks), but you do need a good millivoltmeter for this test.
>
>   
I've pulled all socketed components (NVRAM, CPU, FPU, SIMMS, LSI ? chip,
BOOT ROM)... still smokes.

I've identified 13 surface mount capacitors (not counting teeny ones) on 
the top
side of the board that tie between 5V and ground.  On a good board, I 
get two
slightly different ohms readings (39.1 and 39.4 depending on which cap) 
between
their + side and ground.  (I'm guessing two different 5V 'rails')

On the bad board, I get the same reading on all of the 5V caps.

What are the chances that one of these caps are bad ?  I know tantalums 
tend to
short, but since I'm familiar with tantalums from working on 80s arcade 
games,
I'm used to the 'bulb' with two legs variety of tantalums.  I wouldn't 
know how
to tell if these surface mount caps are tantalum or not.

Short of pulling these caps one at a time and metering them out of 
circuit (I have
soldering equipment, but it isn't intended for surface mount... not that 
I haven't
pressed it into that duty before... but because it isnt the 'right' 
tool... it is tedious
and frustrating to make it work for the task... although it isn't too 
bad on components
with just two connections).

So far despite the 'smoke' tests.... the inductor has not burned open 
yet.  I doubt it's
value is particularly correct at this point... but I can worry about 
that if I can find
and clear the short.  It is like a 1watt ? resistor with bands:

silver, black, brown, tan/gold ?, and then a very wide silver band 
(which must be a
multiplier, tolerance, etc)

Any further suggestions, or should I start pulling caps ?

-- Curt

> -tony
>
>