25 Apr
2009
25 Apr
'09
12:53 a.m.
Well, the Apple monitor is running fine now. Details below for anyone
who's interested.
The most basic test is to remembner that a transistor
has 2 diode
junctions, one between base and emitter, the other between base and
collector (no, you _can't_ make a transistor from 2 diodes!). So you can
test each of those junctions as a diode using either an analogue
ohmmeter, or more likely these days the diode-test range of a DMM.
I'll assume you're using the latter. Desolder the transistor form the
circuit and connect the meter, on the diode test range, between the base
and emitter. Then try it again with the proes the other way round. One
way should read about 0.7V (forward biased Si junction), the other should
be 'overrange'. Then do the same tests with the meter conencted both ways
between the base and collector terminals of the transistor. Again, one
way should be 0.7V, the other way 'overrange'.
If either junction measures 'overrange' both ways round, it's open. If it
reads 0V (or 0.7V) both ways round it's shorted.
This will actually find a lot of defective transistors. Most failures end
up with one of the junctions open-circuit.
First thing I did was remove the transistor I initially suspected and do
the test as above. My multimeter proved a bit strange - it read
approximately 1000 (unspecified units...) in one direction for both
'diodes', and out of range for the other. That seemed like it might be
reasonable, but I figured I'd try replacing it anyway. The replacement
showed exactly the same behaviour - no great surprise. So, I decided to
look further upstream, and actually take note of the waveform voltages.
There was something very odd; the input from the blanking pulse
generator was about 10V peak to peak, fed to the input of the amplifier
via a voltage divider, but the signal at the base of the first
transistor was only about 50mV peak to peak (not what I expected given
the divider's values). That made no sense - I checked the resistors, and
then pulled that transistor and checked it. It looked exactly the same
as the first. Next, I checked the waveform at the amplifier's input
without the transistor fitted, and it was much higher voltage. And more
surprisingly, I had what appeared to be a perfect picture. I figured I
just couldn't see the re-scan lines due to the fairly bright picture.
Loading different software to get a mostly black screen, the lines were
obvious. So I replaced that transistor, the lines disappeared, and the
monitor once again has a very nice picture. I'm very pleased with that
success - thanks to everyone who offered suggestions, especially Tony!
If a transistor passed this test, you need more
complex equipment to test
it. Basically you bpass a small current trhough the base-emitter juction
and see how it affects the collector current -- in other words you use
the transisorr as am amplifier. Real enthuisats have a thing called a
'curve tracer' which does this automatically and plots one of a range of
voltage-current or current-current traces on a CRT. But I find my Tekky
575 (such an instrument) rarely gets used when I'm faultfinding, the
defective transsitor can be found with simpler tests.
In my case, perhaps a tool like that would have found the bad transistor?
I don;t recognise that sort of number at all. US
bipolar transistors tend
to have numbers starting 2N, European ones are things like 'BCxxx or
BFxxx' and Japanese ones start 2SA..2SD (often the '2S' is omitted, but
you still start with a letter).
My suggestions is to look at the maximum voltage in this part of the
circuit (is it just 12V'? and to pick a small-signal transistor of the
right polarity (NPN or PNP). The well-known 2N3904 (NPN) and 2N3906 (PNP)
would proalby do.
In this case, I had some random BC547s and BC557s floating around -
these are what I used to replace the odd transistors in this monitor.
Now, if only I could get one of my three faulty Apple II Pluses running
again... I've tried that before with no success.
Mike.