11 Apr
2012
11 Apr
'12
12:56 p.m.
Begin forwarded message:
From: Brent Hilpert <hilpert at cs.ubc.ca>
Date: 2012 April 11 12:46:40 PM PDT (CA)
To: "General Discussion: On-Topic and Off-Topic Posts"
<cctalk at classiccmp.org>
Subject: Re: Equivalent of an A55 Transistor
On 2012 Apr 11, at 12:00 PM, Rob Jarratt wrote:
--whoops-> *with
Q22*
Thanks for that. I am left wondering about one thing though. I
believe the
c-e resistance of the transistor will have a material effect on the
frequency of the oscillation of the 555, which is controlling the
chopper
transistor of a switched mode PSU. I don't know how this
resistance might
vary between different transistors and how this might affect the
chopper and
the PSU as a whole.
The reason I think it affects the oscillation is as follows: The
transistor
switches in a resistor that affects the 555's timing. Before the
failure of
the transistor in question the 555 oscillated at about 28KHz.
After the
failure it oscillates at about 14KHz. As a test, using just a 9V
battery, I
shorted the C-E terminals of the transistor with a piece of wire
just to
bring the switched resistor back into the circuit; when I do this
the 555
oscillates at 33KHz, so the transistor must be adding resistance that
influences the timing.
It appears Q21 (the A55 at issue) is working in conjunction with
Q21 to form a constant-current supply for the timing capacitor.
(That should
mostly linearise the charge ramp on the cap, but I'm not clear yet
as to what the objective is there as the output from the 555 is
still off-on and I haven't examined the entire power supply
schematic enough to figure out where this fits in the big picture
of the whole power supply. Stability perhaps, or, is that 300V
through the 3.9M R a kickstart or is it regulation feedback?)
The critical components in the Q21/Q22 circuit are D34, R41, R42 &
R45, the transistors in operation will 'adapt' around (as
determined by) the values of those components. Transistor specs
vary widely (between units and with temp) and (again, generally)
one doesn't design around a specific value of transistor spec - one
chooses a transistor for various mins and maxes and the 'design'
proper is done with more precise and stable components (R/C).
(Flippantly, that's also an explanation of why digital has won: so
you no longer have to sit down with a slide rule to accomplish
something.)