Colpitts VCXO

[Andrew Holme]

I'm trying to make an 8.192 MHz VCXO using a JFET Colpitts circuit. I had a
100pF trimmer and a varicap in series with the crystal between gate and
ground, and I was expecting the tuning range to centre on the crystal
frequency; but the output frequency was too high. I tried it with 6 MHz and
7.3728 MHz crystals from my junkbox and these worked fine: I could adjust a
few hundred Hertz either side of the nominal (marked) crystal frequency e.g.
5.999900 to 6.0000100. I tried changing the Colpitts capacitors and even
removed all other series capacitance, but the 8.192 MHz crystal refused to
oscillate below 8.193 MHz; so I tried series inductance and finally it
worked. Is this because I have a mixture of series and parallel cut
crystals here? The hobby supplier I bought the crystal from doesn't even
specify load capacitance, let alone cut, so it's just as well that I
prototyped the oscillator before doing my board layout. Any tips?

TIA

 

[Meindert Sprang]

I'm trying to make an 8.192 MHz VCXO using a JFET Colpitts circuit. I had a
100pF trimmer and a varicap in series with the crystal between gate and
ground, and I was expecting the tuning range to centre on the crystal
frequency; but the output frequency was too high. I tried it with 6 MHz and
7.3728 MHz crystals from my junkbox and these worked fine: I could adjust a
few hundred Hertz either side of the nominal (marked) crystal frequency e.g.
5.999900 to 6.0000100. I tried changing the Colpitts capacitors and even
removed all other series capacitance, but the 8.192 MHz crystal refused to
oscillate below 8.193 MHz; so I tried series inductance and finally it
worked. Is this because I have a mixture of series and parallel cut
crystals here? The hobby supplier I bought the crystal from doesn't even
specify load capacitance, let alone cut, so it's just as well that I
prototyped the oscillator before doing my board layout. Any tips?

This looks like a cheap "processor" grade crystal to me. In my experience,
most desings use too low caps on the crystals. Look in CPU datasheets and
you'll always see the "standard" 22p caps specified, while most crystals are
spec'd for 25-30 pF load. This means that you will need at least 33p caps to
get near this value.
Once on a DSP design that needed to be sampling at exactly 16384Hz and
synchronous with the signal source for several seconds, I went up to 68pF on
the driver side and 22p plus a 68p trimmer on the other side to get the
exact frequency.

Meindert

 

[colin]

I'm trying to make an 8.192 MHz VCXO using a JFET Colpitts circuit. I had a
100pF trimmer and a varicap in series with the crystal between gate and
ground, and I was expecting the tuning range to centre on the crystal
frequency; but the output frequency was too high. I tried it with 6 MHz and
7.3728 MHz crystals from my junkbox and these worked fine: I could adjust a
few hundred Hertz either side of the nominal (marked) crystal frequency e.g.
5.999900 to 6.0000100. I tried changing the Colpitts capacitors and even
removed all other series capacitance, but the 8.192 MHz crystal refused to
oscillate below 8.193 MHz; so I tried series inductance and finally it
worked. Is this because I have a mixture of series and parallel cut
crystals here? The hobby supplier I bought the crystal from doesn't even
specify load capacitance, let alone cut, so it's just as well that I
prototyped the oscillator before doing my board layout. Any tips?

Without seeing the circuit it does indeed sound like your crystal is
designed for series resonant mode, wich means it will oscillate a khz or so
when used in parallel mode, an inductor in series will compensate.

You could always use a series mode oscillator (eg butler type) wich are
often considered superior.

Colin =^.^=

 

[Mark]

Is this because I have a mixture of series and parallel cut

crystals here?

Yes if it is a series mode crystal you need an L in series with it to
pull it to freq below the stamped value...

Many designs will use a series crystal that is cut for a freq a little
lower than they need and then "pull" it up...

The crystal parmameter known as "motional capacitance" will control
how easy the crystal can be pulled...

For a VCXO crystal you NEED to specify the mode (series or parallel) ,
and the motional capcitance in order to get repeatable results.

Mark