Question about low noise oscillator amplitude or frequency instability

Hello, I design this low phase noise and distortion crystal oscillator for 10.7MHz
http://neazoi.com/refosc/lno.pdf
I am testing it now and getting a 0.1V output on 50R. The signal waveform seems undistorted on an analogue oscilloscope.
The problem I have is that the oscillator waveform amplitude seems to change amplitude at random intervals (very little amplitude variations), like a GND is not touching well. I checked all connections and they are ok. These fluctuations are heard on a shortwave receiver BFO as a slight changing in the tone.

I changes the PSU caps just to make sure that a faulty capacitor is causing this. The problem was not fixed.

Could anyone suggest what to look for?

I wonder why no voltage regulation for the oscillator supply is implemented? Did you check stable operation of the thermostate controller?

I have made two pcbs for it, one for the oven and another one for the oscillator. I am just testing the oscillator one now (not ovenized).
The oscillator has been designed for lab use, meaning that one could be able to supply different voltages to get different output characteristics, mainly on power. A voltage regulated PSU prior to the oscillator is required, which will feed the oven as well. The reason for including a separate internal regulator is to clean up the power supplied to the oscillator, not to stabilize the voltage.
FvM, I noticed, after about an hour, these fluctuations are much rarer. It may be coincidence, but I would like to know what is happening there. Is there any way the oscillator loop capacitors to cause this? I have used 1% silver mica types and 0.1% resistors on the oscillator part. For the variable capacitor I am using a miniature silver plated air variable.
Also, is there any way a faulty transistor to cause this? (have you ever encounter such a situation)
Before I start replacing components I would like to have a clue where to look for.

Can you bridge the regulator and feed the oscillator from another supply with exactly same output voltage as the regulator. You may slowly increase and reduce the supply voltage to make sure your oscillator is OK. Changing the supply voltage may change the frequency, but if you notice frequency steps, there is something wrong with the dimensioning of the oscillator.

I did not look in detail, but is it a proportional heater controller, or a bang bang controller. You want proportional. Also is there NO dc connection between the heater circuit and the crystal, like 10 M ohm between them?

Try running heater and oscillator off different supplies.

Try replacing the capacitors with different manufacturer caps, sometimes the caps have microcracks or are microphonic.

As I mentioned, I am not testing the oven now, I have left it disconnected. Just consider only the oscillator circuit.
I have done that, I have reduced the voltage in the whole circuit to see the effects. The frequency shifts a bit and the amplitide reduces, but no steps, it is all smooth.

- - - Updated - - -


I just test the oscillator, the oven is left unconnected and has no effect on the circuit. in fact the oven is on another PCB far apart.
The capacitors of the regulator was the first I thought and I replaced them, but still nothing... I still get these "microcracks" or better suddent variations in the amplitude (like bad gnd connections)

Does the cracking sound changes with supply voltage?

Cracking effects in components can be problem. Once I built a 900 MHz VCO with the former philips teflon trimmers. In production these were changed to surface mount trimmers (new from the reel). very low frequency phase noise became extremely high and finally the expensive philips trimmers were used. Now I know why they were expensive.

No it does not changing. the cracking is at random intervals. I do not know if this is coincidence, but the cracking starts after 1-2 mins, like the oscillator is overheated then. But no components are overheated.
I changes all components in the oscillator section, but it does not seem to make any difference.
Maybe it is something with the psu?

Ok I isolated the PSU, this is not a PSU problem.
I replaced R8 with 2K2 and the oscillator runs at about 6.5v now, whereas the amplifiers run at about 11v. The problem is much less now, in fact rarely occurs.
Do you think the loading resistor of the oscillator (isolation) was not enough or the voltage was too much for the oscillator before?

As far as it goes I cannot see any "cracking" for many minutes.

Did you change the supply voltage of the oscillator (only Q3)?

Changing R8 does not change the AC load to the oscillator (because of C8), but does change current and voltage through the oscillating transistor, so in fact you are changing the collector voltage and current. You changed 2.2k to 2.2k, so power supply change would be very small, or do we have some confusion now?

Oscillators can have several "stable" operating modes. When these stable modes/states are very close together, an oscillator may hop between these modes, resulting in cracking like effects.

Other thing could be parasitic oscillation of the JFET, that may be present depending on drive level. Changing R8 does likely change the drive level for the JFET, so may change the stability of the JFET.