[time-nuts] Crystal Drive Level

Bruce Griffiths bruce.griffiths at xtra.co.nz
Tue Aug 12 18:37:46 EDT 2008

Peter Putnam wrote:
> Greetings...
> I have a well-specified quartz crystal:
> Bliley BG71A
> (vintage mid 70's ?)
> 3.000 000  MHz
> parallel resonant, 100 pF
> 80 deg C
> tempco of <1.5/10e7/deg C over range 79 to 81 deg C
> in a glass enclosure with 9-pin miniature tube base
> that I would like to utilize in an oscillator circuit such as the one 
> proposed by Bruce Griffith in his note to this list on 8/11/2008.
> How can I determine the crystal drive level and be certain that the 
> drive level specified as "10 microwatts into10 ohms" will not be exceeded?
> Regards,
> Peter

10 microwatts in 10 ohms corresponds to a crystal current of 1 mA.
The question is is this an absolute limit or a recommended drive level?
If it is a maximum then you should operate with a crystal current 
somewhat below this.

Thus all you have to do is ensure that the crystal current is 1mA or less.

It would be helpful to know the crystal ESR and if it is a fundamental 
or an overtone crystal.

If its an AT cut fundamental crystal the ESR may be as high as 40 ohms 
or more.
In which case the first stage of the oscillator buffer can be simplified 
to a common base stage.

An AGC loop sensing the crystal current and comparing it with a 
reference voltage can easily be added.
Another CB stage could be driven by the other end of T101's secondary  
and  a 1K impedance load  used at its collector output.
An emitter follower could then be used to  drive an  RF detector for the 
AGC loop.

Alternatively one could use a configuration with a high input impedance 
buffer as in Wenzel's circuit and drive the AGC (or monitor the level) 
from the buffer output.
Its not too difficult to arrange for such a buffer to drive a cascade of 
common base stages to achieve a very high reverse isolation with 
relatively low distortion.
However a few RF transformers are required.
Using a buffer amplifier transistor with a resistive collector load is a 
bad idea (unless the resistor is shunted by an inductor or the primary 
of a transformer) as it significantly increases the phase noise of the 
amplifier over that which is possible with an optimised buffer amplifier 

I'll post another version of the crystal oscillator circuit in a day or 
so that utilises the alternative high impedance buffer circuit driving a 
CB cascade.
It will also include AGC.

Another possibility is to use the Driscoll  two stage oscillator where 
the crystal current is limited by  setting the collector current of the 
second stage.
The second stage is driven into cutoff limiting the RF collector current 
of the first stage which is equal to the crystal current.
However this circuit will significantly degrade the crystal Q unless the 
crystal ESR is relatively high (~40 ohms or more).
Variants of this circuit which use AGC instead of cutoff to limit the 
crystal current are also possible.
Another technique is to use diodes to clamp the oscillator output stage 
collector RF voltage peaks thus limiting the crystal current.
For some variants of such Driscoll oscillators see:

NB using a LED in an oven may be somewhat problematic, so a different 
low noise method of biasing the various transistors may be required.


More information about the time-nuts mailing list