dsi@unccvax.UUCP (Dataspan Inc) (12/14/85)
Folks, I've a real problem ...
I, for better or worse, inherited a Cetec-Sparta SS1000 MW transmitter
to be retuned from 1590 to 1210 kHz. This is an early all-solid-state
bipolar AM transmitter.
Well, I set the impedance at the input of the output filter to 130 ohms,
plugged in the crystal for 1210, turned it on, and promptly blew 48 output
transistors in the final amplifier. Yes, I had a 50 + j0 ohm dummy load
connected to the transmitter output.
The output transistors were made by TRW, type PT 9122. According to the
TRW RF Products division, this isn't even an "rf" transistor but a "power"
device sold by the Opto division. They were trying to get the data on
this power transistor, when I finally got fed up with this approach, I
called Elcom-Bauer (who supposedly bought Cetec-Sparta). They suggested
that TRW said that type SVT-350-5 would work.
A quick check of my Motorola book crosses this device over to another
with a storage time of 3 microseconds. Isn't this a wee bit long?
My question is:
The output stage consists of four identical bridged PA sections which
are transformer coupled through a wideband ferrite device. Each PA section
consists of two quasi-complementary NPN transistor stages, which is switched
in opposite phase by a driver transformer (square wave drive)... see below
C----------------------------C------------- 110 v
Q1 Q3
E -----mmmmmmmmmm---------E
C C
Q2 Q4
E E
------------------------------------ground
mmmmmm represents the transformer winding which couples that particular
PA section to the output transformer. The drive is arranged such that the
transistor Q1 and Q4 conduct, then Q3 and Q2 on the opposite phase. There
are various diodes in these leads to prevent undershoot, and to "remove
the effects of storage time."
What this transmitter really seems to amount to is a giant switching
power supply.
If you were selecting output stages for this, wouldn't one want to
get output transistors which had the fastest possible switching time? What
are my chances for converting this device to HEXFET's (tm)?
If you have the data for the PT9122, SV3-350-5, or any help at all,
let me know. I've got to get this thing on the air before the end of the
year (or face a quick rebuild of a 25 year old 10 kw Collins transmitter).
David Anthony
Chief Engineer
DataSpan, Inc
(704) 597-8317 days, 535-8292 nites