7 May
2005
7 May
'05
7:01 p.m.
Tony Duell wrote:
When I said it results in a cooked computer, I didn't mean to say it
fried the whole thing, it just broke it. I believe the failure
mechanism is exactly as you say, the commutating transistor in the
switcher overheats and dies.
The compucolor also another failure mode; it is a rare example of how a
software bug can cause physical damage. The compucolor logic is stuffed
inside a normal TV case (there are even empty holes for
tint/brightness/contrast along the bottom and a convenient carrying
handle on the top) with the board driving baseband video to the normal
TV logic. The CRT timing is generated by a chip that has timing
registers loaded from an on-chip ROM. However, the CPU is free to write
new values to the timing registers, and if the wrong value is entered,
it causes something dire to happen in the TV section's logic (eg, if the
horizontal frequency is too low then the flyback transformer's impedence
drops and draws a lot of current and smokes).
Vintage
Computer Festival wrote:
I still wonder about the diodes/capacitors (assuming a conventional
circuit) in the raster correction area....
On Sat, 7 May 2005, Jim Battle wrote:
...
Mine produced smoke too :( Where did yours smoke
from? I want to get
mine working again. It seemed to have come from the CRT section. I'll
bet I have a smoked cap somewhere.
I didn't see any smoke. The previous owner reported it, so it was just
speculation on my part that it was dust cooking off the CRT neck. A very common problem reported is that the
switching power supply
sometimes wouldn't start oscillating and it would result in a cooked
machine.
If an SMPSU fails to oscillate, it gives no outputs (well, an isolating
one does anyway, and anyone who designs a computer that's not isolated
from the mains input is criminally insane!). It will not cook anything
else. But it'll probalby blow its own chopper transistor and assorted
other parts...