High voltage RF oscillator

[LukeW]

This is the RF oscillator stage in an ion-trapping time-of-flight spectrometer.

The RF output to the ion trap is about 1 MHz, 4 kV p-p.

There’s no crystal or input from an external oscillator, it just sits at whatever frequency determined by L/C, and changes subject to things like the geometry of the electrodes, cable length into the vacuum feedthrough etc.

I want to tap something in to the circuit to inject stable, crystal-locked frequency at some known frequency eg. 1.000 MHz.

Is this realistic? What would be the best way to do this into this circuit?

[awallin]

one way is a helical-resonator where the big coil is placed in a e.g. copper box/tube, the big coil is connected to the capacitive trap, and is driven with a smaller antenna-coil, like so:
https://arxiv.org/abs/1106.5013

the Q value might be 500-1000, so you need to drive this at (almost) the LC resonance-frequency in any case (your circuit above self-oscillates at the LC resonance, when driving with an external signal generator you'll get max voltage in the trap only at the LC resonance).

you probably will need about 1-5W (maybe) of RF input to the driving antenna-coil, and with a Q of 500-1000 you should get in the range of kVpp in the trap. With a signal-generator giving maybe 10Vpp out (into 50R) that means you'll probably need a 1-5W RF amp.

[graybeard]

You could easily change it from an oscillator to a driven circuit by removing the feedback from the base drive transformer input winding and injecting a signal there.  The feedback connected to that winding would might need to have a resistor substituted for the winding, or shorted, but analysis beyond simple inspection would be required.

[T3sl4co1l]

Wonder what the "loop" (to J7-2) is doing.  There isn't a controller, is there?  And what's the tube doing?  Seems to be shorting it out when keyed, weird.  Or pinging it, but there's no plate supply voltage for that.

Anyway, injecting some reference to T201 or R3 or R4 should do something, assuming the tank has wide enough bandwidth to be pulled as much as needed.  A better solution might be power varactors on T202, or even a servo to run a small air variable cap to Q301,,, wait what the hell that's an inductor not a transistor.   In any case, you'd have a PLL against the reference, which is likely to have a wider tuning range and lower noise than injection locking?  Not that pulling T202 will do all that much because its output voltage won't be too high a fraction of Q301's, I think.

Note that modulation will pull by a similar amount [as varactors on T202], due to Q203/4 collector capacitance varying with supply voltage.

I would much rather leave the tank to its own devices and time everything else against it..?

Also, I don't get, or know, if this is just an idle question, screwing with an existing machine that likely isn't broke (which as the saying says..), or new design for a mass spec you're making, or...

Tim

[awallin]

Wonder what the "loop" (to J7-2) is doing. 

with a helical-resonator we use a short stub of wire in the resonator-box (close to the main coil) to get a voltage monitor that is proportional to the voltage across the trap electrodes. The "loop" in the image might be a similar probe-point, to see what is going on in the main coil. To get good Q for the resonator it is best to load the "loop" only with e.g. a 10x oscilloscope probe or similar. The monitor probe reads e.g. 1 Vpp when the trap voltage is 500 Vpp - that ratio needs to be calibrated.

[jonpaul]

Hello: Much simpler:

Classic class C Armstrong power oscillator, (tank has feedback "tickler" and bias is RC self bias)

JUST one tube, no semis, triode like 811A, 813.

See induction heaters and Tesla Coils, of tube design, circa 1930s...1960s.

Tank will determine frequency.

Built these many times.

Bon chance


Jean-Paul

[Quarlo Klobrigney]

What signal is coming into the op-amp aka "tube trigger"?
Is it a sine wave or other?
The grid of the 807 has a fixed negative bias with DC coming out of the LM311.
Just wondering if there is also a frequency component.
Without a signal, the 807 should clamp off.