Keyed power oscillator frequency drift, why and how?

[neazoi]

Hello, this is an HF power oscillator that is keyed for CW. The output is about 4W @ 50R and it uses a power transistor.
The second picture shows that there is some frequency drift (~200Hz) of the oscillator and this is normal because it is unbuffered. You can also see the times on the right, which show that the greatest drift occurs in the first 5 seconds of keying or less, then the oscillator stays pretty much stable.

I was wondering why this is happening, Is it a thermal issue?

If so, what are the components that I must suspect for this issue, in order to find out what can I do (if anything) to compensate this instability?
Maybe some negative ppm caps coupled to the thermal source?
Or keeping the oscillator pre-heated at all times somehow?

A hint: By touching the components of the circuit, none is heated in the first 5 seconds. The only thing I cannot touch is the crystal, as the crystal is inside a metal can and I cannot find out if it is heated or not.

 

[jiripolivka]

Hello, this is an HF power oscillator that is keyed for CW. The output is about 4W @ 50R and it uses a power transistor.
The second picture shows that there is some frequency drift (~200Hz) of the oscillator and this is normal because it is unbuffered. You can also see the times on the right, which show that the greatest drift occurs in the first 5 seconds of keying or less, then the oscillator stays pretty much stable.

I was wondering why this is happening, Is it a thermal issue?

If so, what are the components that I must suspect for this issue, in order to find out what can I do (if anything) to compensate this instability?
Maybe some negative ppm caps coupled to the thermal source?
Or keeping the oscillator pre-heated at all times somehow?

A hint: By touching the components of the circuit, none is heated in the first 5 seconds. The only thing I cannot touch is the crystal, as the crystal is inside a metal can and I cannot find out if it is heated or not.

Such effects were observed already in 1920s. The solution was found in using a buffer amplifier and keyed PA. For a good stbility NEVER try keying the oscillator.

 

[neazoi]

Such effects were observed already in 1920s. The solution was found in using a buffer amplifier and keyed PA. For a good stbility NEVER try keying the oscillator.

I know that, as I said already. The point of the post is to find out WHAT causes this behaviour so as to see if this can be corrected or minimized.

 

[FvM]

I guess, the transistor junction temperature and related variations of transistor capacitances and current gain are making the major effect.

In any case, I don't believe that it's possible to compensate dynamical effects like this with respective cap TC. Maybe ham practitioners know a better single transistor power oscillator circuit? Otherwise, need to go for multistage.

 

[jiripolivka]

The transistor is always heated when DC current starts. Crystals are often sensitive when starting, and every crystal tends to behave differently. You may try to write a Ph.D. thesis on this problem, but the practical solution is the above.

 

[volker@muehlhaus]

It is expected than an oscillator needs some startup time before the frequency is stable. It starts from amplified noise and then feedback filters the voltage that starts building up. I would expect that this is a fast process, limited mostly by the time constants in your circuit. (e.g. R and C in supply, L and C in tank circuit, change in transistor parasitics with operating point). Drift with seconds of time constant sounds like thermal effect to me, caused by internal drift of operating point.

 

[jiripolivka]

Also I would add that it is WRONG to use a crystal as indicated if the oscillator generates more than 10 mW output.
PLAINLY WRONG design, you will see soon your crystal broken!

 

[neazoi]

Also I would add that it is WRONG to use a crystal as indicated if the oscillator generates more than 10 mW output.
PLAINLY WRONG design, you will see soon your crystal broken!

Yes I have been warned about it. However I am using vintage crystals in larger packages (maybe they are more robust?) I do not think the tiny tuning fork or even some hc-49 full size crystals would work without a problem, eventhough the author of the circuit tries to minimize the RF through the crystal by the capacitor divider.
Saying so, I keep these vintage crystals operating for 1 minute (as seen in the picture) with no signnificant noticed effects.

- - - Updated - - -

Drift with seconds of time constant sounds like thermal effect to me, caused by internal drift of operating point.

Hm... it is as I suspected it. Any guess which component can cause this? I might be lucky to compensate it a bit, so this 5 seconds period does not drift too much.
An idea I had was to keep the oscillator running at low power (using an emitter resistor) and then switch to full power by shorting this resistor. But I must ensure that reliable oscillator will start at the lower oscillator gain.
Any other ideas?

 

[FvM]

To see if it's the transistor, you could make a test circuit that is keeping the transistor DC biased in an operation point with similar power dissipation, keying the feedback path.

The conclusions about crystal power rating are somehow unsure. Reminds me to a student who claimed in a seminar presentation to have found by experiment that TTL ICs can be safely operated with 12V supply voltage...

 

[neazoi]

To see if it's the transistor, you could make a test circuit that is keeping the transistor DC biased in an operation point with similar power dissipation, keying the feedback path.

What if I just connect a switch to the crystal end that connects to the transistor base?
This will keep the transistor biased as you suggest, but it will oscillate only when the feedback is connected to it.

 

[FvM]

Yes, something like this.

Consider that the transistor power disspation may be different without oscillations, the bias circuit may need to be switched, too.

 

[neazoi]

Yes, something like this.

Consider that the transistor power disspation may be different without oscillations, the bias circuit may need to be switched, too.

Thank you, I will try this and let you know.

The other way is to keep the oscillator running at all times and switch it's output between the antenna and a dummy load. However I do not like the idea of having the oscillator running at all times, basically for interference to nearby circuits (eg an RX at the same frequency), but also for crystal stress issues. Remember, this is a power oscillator (~4W)

A third way would be an emitter resistor to vary the gain of the oscillation, but the way you propose, if it works, is preffered to me.

In either case the advantage of zero power consumption when the oscillator is not keyed, is lost. But we can't have everything!

 

[chuckey]

Keep R1 feeding the transistor when the key is UP. Its not the same but it might help, in fact you could look up the transistor and see what its maximum base current is, and fiddle the circuit so in the key up, the transistor is running at its maximum base current to help keep it at the same temperature.
Frank

 

[BradtheRad]

From what I understand, it is all right if the oscillator takes a few cycles to start up.

Hams do not recommend that you start transmitting a carrier wave too abruptly. It comes through like a loud click.

 

[neazoi]

From what I understand, it is all right if the oscillator takes a few cycles to start up.

Hams do not recommend that you start transmitting a carrier wave too abruptly. It comes through like a loud click.

No this is not a start up issue, please read the previous ports, it is all about frequency drifting at start up (5 seconds or so)

- - - Updated - - -

From what I understand, it is all right if the oscillator takes a few cycles to start up.

Hams do not recommend that you start transmitting a carrier wave too abruptly. It comes through like a loud click.

That might help but the transistor bias setting affects harmonics and oscillator start up and efficiency. It should not set up in an extreme point. Also, as I will describe below, this does not seem to be a transistor thermal issue finally.

- - - Updated - - -

Yes, something like this.

Consider that the transistor power disspation may be different without oscillations, the bias circuit may need to be switched, too.

Hi again,
This does not seem to be a transistor thermal issue.
I have tried to put the key switch between the base and the crystal. Again exactly the same drift.
I have also tried to put the key switch at the transistor emitter to ground. The same drift again.
I have also tried to disconnect the diode completely, to see if it has to do with it, again the same behaviour.

If it was a transformer thermal issue I would expect it to continue drifting much more than 5sec, as the transformer would gradually heat up.

So the only component left is the crystal?

If it helps, when I run the oscillator from lower voltages (les power output) I have a bit less drift (I think).

Any other ideas?

 

[BradtheRad]

No this is not a start up issue, please read the previous ports, it is all about frequency drifting at start up (5 seconds or so)

If the drifting really is caused by the crystal, then a crystal oven may be needed. It provides a constant high temperature climate, so the crystal operates in range where the frequency stays more constant, despite changes in temperature.

 

[neazoi]

If the drifting really is caused by the crystal, then a crystal oven may be needed. It provides a constant high temperature climate, so the crystal operates in range where the frequency stays more constant, despite changes in temperature.

It might help. However I am not sure if the crystal frequency will stay the same even in an oven if it is thermally shocked from inside because of large loading (the case in this circuit). I do not think the oven can easily compensate for the internal shock within 5sec.

The above is IF this drift is a crystal thermal shock issue.
So I will stay on post #15 to find out what is the source of this drift.

Any ideas to try would be helpful.
It would be nice to see what causes this in this keyed power oscillator and other similar designs.

 

[FvM]

O.K. sounds like a crystal porblem, as already foreseen by jiripolivka. In this case, I see no other help than running the crystal at lower RF power and preferably continuosly.

It's been said.

You may try to write a Ph.D. thesis on this problem, but the practical solution is the above.

 

[neazoi]

O.K. sounds like a crystal porblem, as already foreseen by jiripolivka. In this case, I see no other help than running the crystal at lower RF power and preferably continuosly.

It's been said.

Ok, since you say it is a crystal thermal problem, then I see no other way to compensate on it (unless you have an idea), other than having the oscillator running coninuously and switch the output between a dummy load and an antenna.

 

[chuckey]

One way would be to compensate the drop in frequency, with a varactor diode that increases the frequency over 5 seconds. One way would be to feed the keyed Vcc via a 1M and a 10K to the cathode of a varactor who's anode is earthed. Feed the cathode to the base via a 50 PF capacitor. Decouple the junction of the two resistors with a 4 MF capacitor. The theory is that when keyed the varactor has no volts on it so it has maximum capacitance, after a couple of seconds the volts across it rises and its capacitance falls, so hopefully raising the crystal frequency to offset the fall in frequency caused by your present circuit.
Frank