Automatic level control in oscillator circuit, how to?

Hello, I have build this oscillator circuit, followed by two buffers (only one shown) and I would like to apply automatic level control to the circuit, so that the output level is constant if different crystals are switched.
I have tested two ideas. The first is using a BJT as a variable resistor, to form a voltage divider, to limit the voltage to the oscillator, which limits the output power.
The second is, using an LDR to alter the resistor value at the gate of the fet.
A third way, which I have not tested yet (although it should work) is to limit the output power of the oscillator, by including a voltage divider at the output of the oscillator, set with an LDR as previously.

The first two ways I have tested worked up to a point but the output power still had some variations and the circuits behaved more like limiters rather than ALC. The output power was more constant but much less attenuated at all bands.

Can a better ALC circuit be applied to my circuit, that will allow for almost full and constant output level at all bands?

Looks unecessarily complicated. Varying the gain or bias point of the oscillator circuit itself seems the natural way to achieve constant level.

FvM said:

Looks unecessarily complicated. Varying the gain or bias point of the oscillator circuit itself seems the natural way to achieve constant level.

Click to expand...

You mean varying the emitter resistor, or the collector one?
semiconductors used as resistors may induce distortions (this was the case with the first diagram), that is why I have used the LED-LDR as a pure resistance element.

You may consider implementing the PLL logic to this circuit. Use an appropriate loop filter to set down bias of the FET other than that I don't see any solution except digitally controlled biasing.

Alper özel said:

You may consider implementing the PLL logic to this circuit. Use an appropriate loop filter to set down bias of the FET other than that I don't see any solution except digitally controlled biasing.

Click to expand...

Now I am confused. A PLL to set the bias of the oscillator transistor?

Stabilizing the output power from an oscillator with buffers is best done in the buffer-amplifier chain. Varying the bias in oscillator stage may cause problems like poor starting.
If the output signal is filtered and frequency change is not large, I would try using a limiter stage after the first buffer. Otherwise use the detector-DC bias circuit as n your schematic but do not touch the oscillator. Buffers are used to separate the sensitive oscillator from load changes.

jiripolivka said:

Stabilizing the output power from an oscillator with buffers is best done in the buffer-amplifier chain. Varying the bias in oscillator stage may cause problems like poor starting.
If the output signal is filtered and frequency change is not large, I would try using a limiter stage after the first buffer. Otherwise use the detector-DC bias circuit as n your schematic but do not touch the oscillator. Buffers are used to separate the sensitive oscillator from load changes.

Click to expand...

I totally agree with you. Also, the rectified voltage may have some ripple, which will be ripple the oscillator amplitude, effectively modulating it.
However, this oscillator is a special case, because it is broadband in a sense. A change of the crystal is the only thing needed to oscillate to another band. BUT different crystals have different characteristics, so the output level of the oscillator is not constant throughout bands. Adjusting the coupling capacitor brings the output level back to an normalized value, but I am trying to auto set this level so no adjustment is made.
Feeding back the rectified output signal from the output of the final amplifier to the input of the first buffer (2 stages are used after the oscillator) is a better idea but not with this oscillator that changes it's output level, so obviously the "correction" has to do in the oscillator I think.

I have tried the two schematics shown on post #1 and apart from the performance mentioned in post #1, the rectifiers were giving additional signals at the output of the transmitter (due to non-linearity). I will try buffering these with a simple bjt amplifier so that they are isolated, I have seen this technique in another oscillator.

I also do not know it altering the source resistor of the oscillator will alter the gain as well?

Try a dual gate mosfet as a gain control cell, either as part of the oscillator itself or as one of the buffers, feed a control voltage to the other gate.

Regards, Dan.

Dan Mills said:

Try a dual gate mosfet as a gain control cell, either as part of the oscillator itself or as one of the buffers, feed a control voltage to the other gate.
Regards, Dan.

Click to expand...

Yeas I have seen this technique.
Also a dual gate mosfet can be made discrete by connecting two fets together I think.
But isn't that the same result as connecting a fet in series with the oscillator fet?

I see that there are different opinions if controlling the oscillator level itself is appropriate or not. I don't see a need to question jiripolivka's viewpoint in this regard.

I would ask the other way around. Without level control, the oscillator must have excess gain so the oscillation magnitude rises until it's limited by circuit nonlinearity, e.g. clipped at the transistor saturation voltage. Typically involving more distortion than a level stabilized oscillator.

I have done some more experimenting. It seems that the source resistor of the oscillator is the best place to apply the feedback.
I have used an LDR and the level varies from very low to full. Harmonics at full level are below the set limit so I am very satisfied with the result.
Next I would like to try a bjt or fet as a variable resistor to see how distortion levels go, but first I need to reverse the ALC control, i.e at higher input voltage increase the resistor and not decreasing it (as it is now).
How should I do that?

As was mentioned above, touching the bias of the oscillator to get ALC (auto level control) is a mistake. If it works, is just a matter of luck, but is not recommend because it will affect the frequency stability and the phase noise.
Ideally the active stage of the oscillator should work in Class-A with stable DC bias, to minimize the limitations in the stage that drives the resonator.

Don't want to harp about on this point too much, but it seems obvious to me that a well designed ALC circuit inside the oscillator can improve start-up behaviour, phase noise and frequency stability. See e.g. **broken link removed**