Date: Thu, 23 Sep 2004 00:46:58 -0400
From: Todd Doucet 
To: rnlucas@micxxx.com
Cc: pasha@kxxx.com, steven@luxxx.com
Subject: On the pulse of morning

Working my way up to measuring the group delay characteristics of
some crystals and filters.  I can use the frequency counter in
A/B timer mode to get nanosecond timings, but I need to be able
to pulse the rf signal for this to be feasible.

The rf signal will be around 14 MHz, say, and it can be pulsed
anywhere from a few Hz on up.

None of my test equipment will obviously let me generate pulsed,
unmodulated rf.  So I'm trying to rig something up.  Surprisingly
difficult, or I'm missing something obvious.

My first try was to set up a diode switching circuit, and turn that
off and on using a TTL pulse (provided my my handy function
generator).  It was tricky to get right.  (Maybe kind of the
difference between writing code and reading white papers on software
development.)

The result is not immediately useful, but kind of interesting.  As
kind of an extreme case, I am trying to pass about a half dozen cycles
of a 14 MHz signal, and then turn it off for a similar duration.  Here
is what I get:
The center horizontal line is ground, so note that this waveform has
the RF at a positive DC offset, and the "ground" at a negative DC
offset.  This is a consequence of the diode switching circuit being
capacitively coupled (input & output).  The 1-MHz squarewave, which
turns the switch on and off, simply gets AC'd, as expected.  This is
inconvenient, and as far as I can tell unavoidable with this kind
of setup.

You might be curious about the fuzziness at the leading edge and the
trailing edge of the RF envelope.  Took me a while to figure that out.
It is an artifact of the oscilloscope triggering, which is suggested
by the following:
That's the same waveform, but I adjusted the triggering controls to
trigger on the 1-MHz switching signal instead of the 14-MHz switched
signal.  This is a characteristic TTL risetime.  So, what's happening
is that when you trigger on a level within the RF envelope, there is
fuzz involved in when you'll get the trigger, because there is no
clear relationship between the switched and switching signal, and
the TTL risetime is fast, but not instant.  I thought that was kind
of neat.

---

I really need a good, pulsed RF signal, and I cannot tolerate any kind
of offsets.  And switch-on times might cause me grief, too.  I don't
think the diode switching idea can be made to work.

Right now, I have the signal, and a simple emitter-follower buffer.
I'm happy to modify that, or add a transistor stage or two.  You
wouldn't think it would be hard to turn the output on and off crisply.
And it probably isn't, but I've suffered all day and I don't know how
to do it.

Any ideas would be appreciated.

PS: Don't tell me to use a MOSFET switch.  They don't work well at RF.