On 10/5/2014 9:42 PM, Michael Black wrote:
On Sun, 5 Oct 2014, gareth wrote:

Being somewhat of a polymath (just spent all day fence judging
at a horse trial) I find that I have a string of ideas faster than I
could ever implement them (rather unkindly described in one area
as vapourware), but I think it to be useful to punt them for a wider
discussion.

Musing upon the Huff-and-Puff technique, I wondered if there
was a better way to improve the frequency stability of older RXs,
because the Huff-and-Puff necessarily brings about a punctuated
frequency span (eg, multiples of 32 Hz), and this is what I came up
with ...

It's worth pointing out that the "Huff and Puff" seems to have been done
originally without a frequency counter. The circuitry is similar, but
it's just done with counters and gates, no inherent need for the latches
and decoders and readouts. I'm pretty sure that came first, unless the
concept came from two different places about the same.

So you could take any old receiver, so long as it had a decent dial, and
add some stability.

it was later that the same concept appeared (or so I remember) as an
extension of a frequency counter. That makes sense, since there was a
transition to frequency counters, and the "lock" circuit was just a bit
more.

Using the ubiquitous timers to be found en masse in most micros
that seem to sell for only a few pence / cents these days, implement
a frequency counter to measure the local oscillator. Then, when
the user presses a Lock button (yet to be provided) the same micro
can program an si570 to generate the same frequency indefinitely
and to switch the mixer stage from the original to this new oscillator.

As a standalone, the odd lock points are because of the way the counters
are arranged, I don't think it was anything to do with some design need
(other than simplicity). Once you are adapting a frequency counter, you
end up with decade type steps.

But (and I've never tried the circuitry) I don't think these tune the
way you are expecting. It takes time to count, so you aren't snapping
between steps like with an FM broadcast receiver with AFC or a
synthesizer of the type in the HRO-500. The notion of the circuit is to
keep it locked, with in the nearest step. So only if it drifts beyond
the specified step does it jump back (and "jump" is probably the wrong
word here).

I don't think what you are describing and what the OP is describing are
the same thing. His approach is to let the receiver tune normally, but
when a button is pushed, the receiver's VFO is replaced with a digital
VFO automatically tuned to the same frequency.

I don't see any problems with doing that. A frequency counter takes
some time to make a measurement which is related to the resolution
desired. But it is not in any way limited to decades or even Hertz.

But there are other ways of measuring frequency. The Si570 can be tuned
in 10 ms, if I read the data sheet correctly when I scanned it. A
relative measurement of frequency can be done nearly instantaneously
(relatively speaking). So doing a binary search will let you tune to 1
ppm accuracy in 19 steps or about 190 ms. Actually the Si570 can tune
faster than that once you get within 3500 ppm so every bit after the
first 9 is much faster. At some point taking the measurement takes more
time than the tuning as finer resolutions take more and more time.

Hmmm, I was thinking as I typed this and I think I realize the binary
search doesn't help you much if at all. The measurements are fast in
the beginning, but even to determine if your synthesized frequency is
high or low for the final bit takes as long as a complete measurement by
frequency counter... maybe. If a phase detector with a proportional
output is used, it can speed up the measurement. In essence use the
Si570 as the VFO in a PLL until it is locked and then hold the settings
and make the switch so the receiver is using the synthesized reference.

More thinking on the fly... a two stage approach may give the fastest
lock. A coarse measurement of frequency to get the VFO in the right
ballpark would be fairly fast. Then use a phase detector to lock the
Si570 to the VFO through the MCU. It could be used as a digital
frequency readout when you are manually tuning.

If anyone thinks there could be a market for this device I would be
interested in making it. I can see it being sold for as little as $50
in a case with a button and a couple of BNC connectors. This unit would
need to be spliced into the VFO signal path somehow. Do receivers
typically have inputs and outputs for the VFO?

--

Rick