On Fri, 18 May 2012 16:12:40 +0100, "N_Cook" put
finger to keyboard and composed:

A bit further forward I suppose. Exactly the same, for 2153 and original,
audible low level oscillator noise , 10 seconds or so after switch off
presumably coming down from ultrasonic and dropping in pitch over a couple
of minutes while there is HV in the main caps. And +/-0.5V on the main rail
caps transfered across the Tx from the startup pulses, instead of the +/-60V
or so . But no sustained drive. Next is some monitor of the Tx primary
feedback route to the driver and somehow monitor the osc f, perhaps change
the SM cap is easier, the SMRs measure ok. Any tricks for fooling the driver
into thinking there is more feedback than actual, a battery wired-in?

I'm having some difficulty understanding your plan, but first allow me
to explain how I think the circuit works. If I'm wrong, someone will
hopefully correct me.

Firstly, I would not attempt to substitute a battery for the
"feedback" voltage. This will defeat the IC's overload protection. If
you are going to do this, at least disconnect the 1uF capacitor (C68).
Also, instead of a battery, you could add several 270K feed resistors
between Vdc and Zener3. Then monitor the IC's Hi and Lo outputs. That
should at least give you some confidence in the IC.

As for how the IC works, I believe it would have a start mode and a
run mode. In the start mode it would draw a very small current,
probably less than 1mA. This would allow the Vcc capacitor (C52) to
charge. When Vcc exceeds the undervoltage lockout threshold, the IC
switches to run mode. It then attempts to kickstart the supply by
pulsing the MOSFETs. In run mode the IC would draw more current than
can be provided by R78, in which case C52 would begin to discharge. If
a regenerated supply cannot be developed from a secondary transformer
winding via D34, C69, D21, and C70, then C52 will discharge to below
the lockout threshold, causing the IC to revert to start mode. C52
then charges once again, and so on.

Therefore, if there is an overload on the secondaries, the regenerated
supply never eventuates, and the IC hiccups. Some IC's will count the
number of hiccups and shut down completely if there are too many
attempts.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.