"Joop" wrote in message
...
On 16 Mar 2007 06:12:27 -0700, wrote:

On Mar 16, 5:49 am, "W3JDR" wrote:

The series resonance is, for practical purposes, invariant. The motional
parameters (L and C) of the series resonance are such high reactances
(small
capacitance; high inductance) that external components have only a tiny
influence on the series resonance.

Yes....this is the point of a VCXO...to allow an almost
infinitessimally small, but still useful, variation about the crystal
frequency while maintaining most of the crystal's stability.

The series resonant frequency is the lower of the two crystal
'resonances'.
The parallel resonance is above it. When you make a VCXO with any
substantial tuneability, you're probably operating the crystal at its
parallel resonance. This leads to the common observation that you can
'pull'
a crystal up in frequency more than you you can pull it down.

Nearly all VCXO's I've run across work the other way. You can pull
the frequency down substantially while maintaining good stability
(typically on the order of 0.1%), but not up. This certainly applies
to the circuit for which the original poster provided a link.

Do you have any examples of practical circuit schematics which use
parallel resonance and which can be pulled substantially up in
frequency ? I assume it should be possible to do with a parallel
inductor, for example in a Franklin oscillator circuit, but as was
pointed out the inductor values can be inconveniently large.

Steve
I have to agree with Joe. Basically there is no such thing as an
crystal oscillator in "parallel resonance". However there are
oscillators that use the crystal in the feedback path to add
substantial phase shift. Such as in the Pierce oscillator where it
behaves inductive. The phase shift changes so rapidly that it can
still make a low-drift oscillator.

In the book by Matthys where he compares various oscillators there is
one in chapter 10.6 where the deviation from the (series) resonant
point is the highest. It is a circuit where the crystal sees a very
high impedance as opposed to regular circuits where highest Q is
obtained with very low impedance. This in effect makes the crystal
load to be around C0 of the crystal with some output and input
capacitance. And therefore it is probably the smallest effective
physical series capacitance obtainable and thus the highest frequency.

Looking up the Franklin oscillator you mentioned, I notice this also
is providing a high impedance to the resonant elements. So yes, it
seems a valid way of implementing an alternative to Matthys' example.

Now also cancelling the effect of C0 of the crystal by adding parallel
inductance might push it a bit further. Right now I would not be able
to predict the effect on loaded Q of the crystal. Lowering Q is
normally not done, but in this case we are primarily in quest for wide
pulling range right?

In a low impedance Butler (overtone) oscillator I have seen C0
cancellation by using parallel inductance as well. There sometimes is
an L plus series R used to lower the Q of the L/C0 combination. This
seems not appropriate for a high impedance oscillator circuit. I would
expect best effect if the Q of the inductor is high (low Rs).

Sorry this is still theory. I have no examples of VCXO in this
context.

The reason for the parallel inductance in the overtone mode is the
low impedance of the crystal self capacitance at high frequencies,
this on its own can be quite low, and the series resistance
at overtone can be quite high, so this can allow the tank circuit to
dominate rather than the crystal unless it is cancelled.

Not sure about the resistor not seeing the circuit.

the statement all crystal circuits operate in series mode stems from the
fact that the internal eq circuit of a crystal is a series LC, any
capacitance in parallel with the crystal can only be in series with the
internal motional capacitance.

If your lucky you may be able to find a spurious node, but it would be hard
to make it oscillate at this point.

Colin =^.^=

On Fri, 16 Mar 2007 22:51:34 GMT, "colin"
wrote:
The reason for the parallel inductance in the overtone mode is the
low impedance of the crystal self capacitance at high frequencies,
this on its own can be quite low, and the series resistance
at overtone can be quite high, so this can allow the tank circuit to
dominate rather than the crystal unless it is cancelled.

I know why the inductor is usually necessary. The thing is that for
the purpose of the Butler the effect of C0 can be neutralized.
The question is whether it can be made to have a similar effect in the
"pulling arena" as well. Of course it should not have to much side
effects in normal operation of the oscillator.

In the overtone butler there is also another LC resonance circuit
present that determines the possible operating frequency (read desired
overtone). Such a thing might be necessary in most circuits where an
inductor is placed in parallel with a crystal.

Joop

In message , Joop
writes
On Fri, 16 Mar 2007 22:51:34 GMT, "colin"
wrote:
The reason for the parallel inductance in the overtone mode is the
low impedance of the crystal self capacitance at high frequencies,
this on its own can be quite low, and the series resistance
at overtone can be quite high, so this can allow the tank circuit to
dominate rather than the crystal unless it is cancelled.

I know why the inductor is usually necessary. The thing is that for
the purpose of the Butler the effect of C0 can be neutralized.
The question is whether it can be made to have a similar effect in the
"pulling arena" as well. Of course it should not have to much side
effects in normal operation of the oscillator.

In the overtone butler there is also another LC resonance circuit
present that determines the possible operating frequency (read desired
overtone). Such a thing might be necessary in most circuits where an
inductor is placed in parallel with a crystal.

Joop


I've eventually found some good plots of the frequency responses of
crystals (see below).

http://g4oep.atspace.com/crystal_fil...rystal_filters
..htm#4)%20Single-Crystal%20Ladder.

This info from G4OEP is about filters, rather than VXOs, but it
graphically illustrates how a crystal can be neutralised in a bridge
circuit, using a small capacitor rather than large inductor. The
capacitor will be the same value as the parallel capacity of the
crystal.

The plots in Figs 9, 10 and 12 also show how the throughput peak moves
around to some extent as the neutralization is adjusted. The circuit is
essentially the same as I used when I was testing a load of crystals to
see if they met spec wrt frequency accuracy and ESR.

You should be able to use the circuit of the single crystal filter
(in-to-out, without the resistive padding etc) as the resonant part in a
VXO, and adjust the neutralization to pull the frequency of oscillation.
It's worth a try.

Ian.
--

On Tue, 20 Mar 2007 12:08:57 +0000, Ian Jackson
wrote:

You should be able to use the circuit of the single crystal filter
(in-to-out, without the resistive padding etc) as the resonant part in a
VXO, and adjust the neutralization to pull the frequency of oscillation.
It's worth a try.

Ian.
The images primarily showed how the stopband notch moved around.
So I decided to put it into spice.

Compared to L compensation it does seem to have a benefit as higher
frequency peak. It also is not troubled by the side-effect of
L-compensation as passing lower and higher frequencies than those
around the crystal frequency.

Trying to make the most of the balancing compensation I placed a small
capacitor in series with the crystal. This moves the pass band
frequency up a bit more. But the smaller the series cap, the less
pronounced the peak seems to be. Also the circuits starts to attenuate
more and more. This might cause difficulty in an oscillator setup
where the loop gain should stay more than one.

Also with the balanced compensation circuit, the phase is changed
around the peak frequency. Without series cap around -45 degrees,
climbing to -83 with 12pF. This should be accounted for in the
feedback loop of an oscillator. (A properly dimensioned L-compensated
crystal does not change phase.)

But all in all it might be a (complex) method of shifting the working
frequency up.

Joop