Thanks to Tom for working on the keying waveform issue and finding a good
point to do a mod in the radio. I set up my 1000MP for keying waveshape
measurements to examine the effect of Toms. mod. Here are some numbers.
C added to C2148 Rise time Fall time Duty cycle Speed
0 2.0mS 0.8mS 45% 40WPM
0.05uF 5.0 1.5 36.7 40
0.10uF 6.0 2.5 33.3 40
The mod does indeed reduce the bandwidth of the CW signal substantially.
However, the keying waveform is distorted in the process. A perfect CW
waveform would have dots which measure 50% on and 50% off half way up the
waveform. The unmodified MP has 45% on and 55% off and is acceptable in
operation. After adding the 0.1 uF capacitor to C2148 the on time is
reduced to 33.3% at a speed of 40WPM Since the fall time is much shorter
than the rise time, it is primarily responsible for the excess bandwidth on
CW. A perfect situation would be equal rise and fall times and 50% on time.
The keyed waveform at C2148 swings from -4.5 to 0 volts and is used to bias
the transmitting mixer off at the -4.5 level and on at 0. The 2SK302 FETs
in the mixer are completely biased off at -2.5 volts and really only start
conducting around -1 volt. So the portion of the keyed wave which switches
the RF on and off is in the top 1/4. With equal rise and fall times of the
keyed waveform, the duty cycle will be less than 50% above the half way up
point, or 2.25 volts. Since the mixer is switching around -1 volt, the duty
cycle will be less than 50% at the RF output. Adding capacitance across
C2148 will only make the duty cycle worse. The solution is very simple.
It is not required to run the gates of the 2SK302s as low as -4.5 volts to
cut them off. Something less is quite adequate, say 3 volts or so. Raising
the off voltage will improve the duty cycle of the RF wave. This can be
done by putting a 91K resistor across C2148. A couple of measurement points
are listed below with the 91 K in place.
C added to C2148 Rise time Fall time Duty cycle Speed
0 1.5mS 1.2mS 50% 40WPM
0.05uF 2.5 2.2 48.3 40*****
0.1uF 3.0 2.5 46.7 40
****** Preferred mod. C added = 0.05uF and R shunt = 91K.
The preferred mod listed above reduces the CW bandwidth by approximately the
ratio of the fall time after mod to that before mod or 2.2/0.8 = 2.75x. Yet
it preseves and even improves the CW duty cycle over the unmodified radio
performance.
Note that this mixer is not the only keyed stage. Even with C2148 shorted
to ground the transmitter keys fairly well with a duty cycle of slightly
more than 50% on dots.
George, W2VJN
--
CQ-Contest on WWW: http://lists.contesting.com/_cq-contest/
Administrative requests: cq-contest-REQUEST@contesting.com
|