A correspondent trying to get his head around old designs was challenged by the Tuned Plate Tuned Grid (TPTG) oscillator in common cathode configuration.

A superficial analysis is that the feedback to the grid from the anode via the anode to grid capacitance (Cag) is in phase with the anode voltage, which because of inversion in the valve means it is negative feed back. How can it cause self oscillation?

Well it does, so the superficial analysis is probably inadequate.

Text books tend to gloss over the detail of how it works. Understanding is not helped by some folk lore, try to differentiate between what you know that is truly fact and other ‘knowledge’.

Where does the additional 180° phase shift necessary for self oscillation come from?

Lets take a fairly high level approximation of what happens in a simple circuit, one with identical parallel tuned circuits in the anode and grid circuits, and a very very small equivalent Cag.

Some important concepts:

- We all think of a common cathode stage as inverting (we could say AC anode voltage leads current by 180°), but whilst that is approximately true for a resistive load, it is not true with a reactive load and the TPTG oscillator depends on one or more of its tuned circuits being reactive. A more detailed analysis is necessary to explain the TPTG oscillator.
- If Cag is very small and close to ideal (ie has very low equivalent series resistance), the grid AC voltage will be much smaller than anode AC voltage and the current Iag will lead Va by almost 90°.

Here are the steps around the loop considering only phase:

- At a frequency of half the half power bandwidth (HBW=f0/Q) below resonance (f0), so f=f0-f0/Q/2, the anode tank is inductive and the AC voltage developed at the anode leads the anode current by 180+45°.
- As mentioned, because Va>>Vg, and Cag is ideal, the current in Cag (Iag) leads Va by almost 90°
- Current Iag flows into the (identical in this case) grid tank circuit which is also inductive, and the voltage across the tank circuit leads the current by 45°.

The total phase lead from anode current to grid voltage is 180+45+90+45=0° which satisfies one of the criteria for oscillation. If the magnitude of the total loop gain is greater than unity, then the circuit will self oscillate at this frequency.

So how does it start?

If the stage is biased so that some anode current flows, it contains noise components, and if the feedback circuit has loop gain greater than unity and phase=0°, the noise currents will be amplified greatly and quickly lead to self oscillation.

Note that in this example, the necessary phase shifts mean the circuit oscillates just a little below the self resonant frequency of the two identical tuned circuits.

The circuit does not depend on identical tuned circuits, and just one could be variable to adjust frequency of oscillation over a small range. Though the explanation used a very small ideal Cag, again it does not depend on that though the frequency of oscillation is sensitive to the phase relationship between Iag and Va. Obviously phase shift may differ at each stage with these small variations, it is the frequency where the loop phase shift is 0° that oscillation will occur if loop gain is sufficient.

I have used an example with LC parallel tuned circuits, but one or both could be tuned cavities (for UHF and above).

Was it a good oscillator in its day?

Probably not for a host of reasons, and probably why its use died out pretty quickly. The last transmitter that I had that used a TPTG oscillator was a modified aircraft transponder on the 23cm band (it was originally a TPTG oscillator and needed more feedback to run at lower power in continuous mode)… that was in the late 1960s.