EMRFD Message Archive 244
Message Date From Subject 244 2006-12-18 12:08:09 Wes Hayward VFO Drift
This is in response to posting #238 regarding drift in a VFO. The
title on the posting had nothing to do with the content, so I'll
start a new one.
You were not specific regarding the circuit in SSD that you elected
to duplicate, so it is hard to do any analysis. But there are two
things that could be a problem.
1. You mention a current of 4 mA with only 4 volts or so across the
resonator. It could be that the voltage swing on the resonator is
so high that the transistor goes into saturation during part of the
cycle. This produces an oscillator with a stable operating level, but
the limiting mechanism is the saturation resistance of the
transistor. This really trashes the Q and can do terrible things to
the thermal drift. There are two answers. First, you can drop
the standing current by increasing the emitter resistor value. Try
a change by a factor of 10, down to 0.4 mA. With low levels like
this, you stand a better chance of operating the oscillator with
current limiting where you avoid transistor saturation.
2. It has been my experience that varactor diodes can be terrible
elements so far as temperature stability. I would recommend that
you pull the varactors out of the circuit, replace them with a fixed
capacitor or an air variable, and evaluate the stability. Get the
limiting in line and compensate the oscillator to get reasonable
stability without the varactor. Then put the varactors back into
the circuit and see what happens.
An example of the exercise presented here is Fig 4.33 of EMRFD. In
that case, I did temperature runs in a home "Thermal Chamber" without
the varactor diode. The standing current is low and this is a low
L, Hi C tank, so the impedance is fairly low. There is no
transistor saturation in this circuit. After finding that the
stability was reasonable, I added the varactor. Note that the
biasing of the varactor is between about 5 and 9 volts. The signal
voltage on the collector never gets high enough that the varactor is
driven into conduction; it is ALWAYS reverse biased. If varactor
conduction occurs, you alter the oscillator limiting mechanism in the
same way that transistor saturation does. Anyway, when the
varactor was first added, I was able to determine its temperature
coefficient. Diode D2 was then added to the circuit. This
provided some temperature compensation. I picked a value for R1
that established an operating point for the diode that set up the
required compensation. This scheme is one I lifted from the work
of Fred, WA7TZY, in a VCO that was part of the phase lock system in
the Tektronix 7L13 Spectrum Analyzer.
The oscillator described is part of a transceiver that has been used
on several backpacking treks. Stability, even during long QSOs in
the field, has never been an issue.
I hope this helps. (Remember the "E" in EMRFD.)
73, Wes, w7zoi247 2006-12-18 13:20:00 Rick Re: VFO Drift
that when a second harmonic (or other even harmonic) is added to a
fundamental, the spacing of the zero-crossings may become uneven, but
the average phase does not change. Oscillator frequency is ultimately
determined by the phase around the feedback loop.
If a little bit of the oscillator fundamental (or an odd harmonic)
leaks back into the oscillator loop (for example from a PA, the
antenna, a buffer stage or via power supply lines) the oscillator
frequency shifts to compensate. But if the output of the oscillator
is frequency doubled, those high level signals and power supply lines
will have 2nd harmonic energy, which will not change the frequency if
it leaks into the loop.
Doubling was standard practice in ancient times, and works for a
number of reasons. (I have also observed that drift goes up as
approximately the square of fundamental frequency in my home built
VFOs, which makes sense from the basic math if you think of LC as a
single quantity.)
Premixed VFOs are attractive for the same reason, because the VFO is
translated to a different frequency before amplifying and using it.
But thinking back....all of my best contacts have been with equipment
that drifted a little, starting with my novice days, some portable
work in nasty weather, and on up into microwave weak signal work....
There is a "good enough" and once you get there, continue the
experiments on the air.
Best Regards,
Rick kk7b249 2006-12-18 14:20:44 Nick Kennedy Re: VFO Drift
I guess I didn't realize that the transistor should not saturate until I started browsing the oscillator section of IRFD this morning. I had always sort of thought that excess gain was required to assure oscillation except in special cases (Wein bridge) and that this would result in non-linear device current. So I wasn't paying particular attention to that aspect. The circuit I used as a starting point in SSDRA is figure 2a in chapter 3, page 34.
Funny that I'm now getting some guidance on increasing standing current, or at least leaving it alone. (Well, actually I've had advice in both directions.) Looking at a few circuits made me decide to reduce it from 4mA to 1.5mA. How important is self heating with a BJT oscillator, and what level of current would be OK in a little plastic transistor such as the 2N3904?
Regarding the varactor, for the preliminary tests, I had it tied up to 8VDC via a 48k resistor. In the final version, I wanted to tune from 1V to 8VDC, hoping 1 volt would prevent any forward conduction, given that it is in series with 10pF.
I wonder if LTSpice is accurate enough to allow me to experiment with configurations and watch device current so I can avoid saturation?
OK on the subject line. The inappropriate one was kind of an "editorial cartooon" on the state of things in QRP-L and doesn't really have the context here.
73--Nick, WA5BDUOn 12/18/06, Wes Hayward <w7zoi@easystreet.com> wrote:Hello Nick, et al,
This is in response to posting #238 regarding drift in a VFO. The
title on the posting had nothing to do with the content, so I'll
start a new one.
You were not specific regarding the circuit in SSD that you elected
to duplicate, so it is hard to do any analysis. But there are two
things that could be a problem.
1. You mention a current of 4 mA with only 4 volts or so across the
resonator. It could be that the voltage swing on the resonator is
so high that the transistor goes into saturation during part of the
cycle. This produces an oscillator with a stable operating level, but
the limiting mechanism is the saturation resistance of the
transistor. This really trashes the Q and can do terrible things to
the thermal drift. There are two answers. First, you can drop
the standing current by increasing the emitter resistor value. Try
a change by a factor of 10, down to 0.4 mA. With low levels like
this, you stand a better chance of operating the oscillator with
current limiting where you avoid transistor saturation.
2. It has been my experience that varactor diodes can be terrible
elements so far as temperature stability. I would recommend that
you pull the varactors out of the circuit, replace them with a fixed
capacitor or an air variable, and evaluate the stability. Get the
limiting in line and compensate the oscillator to get reasonable
stability without the varactor. Then put the varactors back into
the circuit and see what happens.
An example of the exercise presented here is Fig 4.33 of EMRFD. In
that case, I did temperature runs in a home "Thermal Chamber" without
the varactor diode. The standing current is low and this is a low
L, Hi C tank, so the impedance is fairly low. There is no
transistor saturation in this circuit. After finding that the
stability was reasonable, I added the varactor. Note that the
biasing of the varactor is between about 5 and 9 volts. The signal
voltage on the collector never gets high enough that the varactor is
driven into conduction; it is ALWAYS reverse biased. If varactor
conduction occurs, you alter the oscillator limiting mechanism in the
same way that transistor saturation does. Anyway, when the
varactor was first added, I was able to determine its temperature
coefficient. Diode D2 was then added to the circuit. This
provided some temperature compensation. I picked a value for R1
that established an operating point for the diode that set up the
required compensation. This scheme is one I lifted from the work
of Fred, WA7TZY, in a VCO that was part of the phase lock system in
the Tektronix 7L13 Spectrum Analyzer.
The oscillator described is part of a transceiver that has been used
on several backpacking treks. Stability, even during long QSOs in
the field, has never been an issue.
I hope this helps. (Remember the "E" in EMRFD.)
73, Wes, w7zoi