Message-ID: <3A33B48A.643C1EA7@mlecmn.net>
Date: Sun, 10 Dec 2000 10:51:22 -0600
From: "Lyle Koehler" <lyle@mlecmn.net>
MIME-Version: 1.0
To: "Stewart Nelson" <sn@scgroup.com>
Cc: rsgb_lf_group@blacksheep.org,  lowfer@qth.net
Subject: LF: Re: [Lowfer] Detection of QRSS ("LEK" raw data)
References: <200012081559.KAA6713991@indyweb.cgocable.ca> <001b01c062a5$e30f98f0$0700000a@parissn2>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

In simple terms, yes, LEK chirps. Now for the long version: My PLL
circuit exhibits a slight amount of frequency pulling during keying for
several reasons. Probably the biggest factor is power supply
instability, which is always present even though separate regulators are
used for the PLL and the rest of the circuit. The box which houses my
exciter includes the synthesizer, keyer (although an external keyer is
presently being used for QRSS) and a complementary pair "final". The
real final amplifier is at the other end of 100 feet of coax, so the
final amplifier in the exciter box is just acting as a line driver. But
the driver does take some current, which pulls down the supply line to
the PLL circuit regulator a tiny bit, which introduces a little bit of
frequency shift. I also have an LED on the front panel of the exciter
that blinks on and off with the keying, and that puts an additional load
of 10 mA or so on the main supply during key-down conditions. Everything
adds up.

There are at least two other sources of frequency shift. The switch that
I use to select the internal or external keyer is on the front panel,
very close to the loop filter components. Even though the 5-volt keying
signal is nearly DC, it does of course have AC components during the
rise and fall time, and these couple electrostatically into the loop
filter. Adding a bit of shielding a couple of years ago reduced this
source of frequency shift to a fairly low value.

Finally, any oscillator (mechanical, electronic or laser) will tend to
pull when a signal at the same frequency is fed back with a varying
phase and/or amplitude. In fact, any oscillator will "lock" to an
external source over a small but finite frequency range. Some small
amount of signal from the keyed output of the driver stage inevitably
finds its way back into the oscillator, so there will always be some
phase pulling because of the RF feedback effect. In BPSK mode, the
frequency shift due to power supply fluctuations is minimal because the
load is essentially constant, but the RF feedback problem is worse
because of the phase reversals.

In addition to the frequency shift associated with keying, there is the
plain old phase noise of the oscillator in my PLL circuit itself. LEK's
homebrew synthesizer uses an L-C oscillator in the HF range that gets
divided down to the output frequency. PLL circuits based on the R-C
oscillator in the CD4046 or 74HC4046 chip are much noisier. I use the
trick of listening to the harmonics to see how bad the phase noise of a
particular synthesizer really is. Another piece of "test equipment" that
is useful for this purpose is the vector scope in some of the PSK31
decoder programs. In trying to determine which manufacturer's 4046 chips
were the least noisy in my "all-in-one" transmitter design, I used the
vector scope while tuned to the 10th harmonic of the LF signal. None of
them looked very good, although the National (no longer available) and
Phillips CD4046 chips were the best. On 10m, they all sound awful, but
fortunately there is little apparent degradation in copy on LF using
QRSS or C-BPSK at MS100.

I am actually surprised that LEK's chirp isn't greater than what you
measured. As for fixing it, throwing away the entire existing circuit
and replacing it with a DDS might do the trick. There is somewhere
between zero and no chance that I'll do that this season, but it's
interesting to speculate on what coherent detection could do for us.

73

Lyle, K0LR

http://www.computerpro.com/~lyle