FM oscillator circuit: questions

[John_BS]

The Armstrong Method was commonly used to get broadcast FM transmitters down to hifi levels of distortion.

The BBC employed a very linear modulator which ran at (originally) 3MHz (subsequently 10.7MHz in later designs) using a circuit topology which was the subject of a BBC patent (Packman / Johnstone). A crystal LO was used to mix up to final frequency. My ex-boss (the aforesaid Johnstone) was keen to point out that a synthesised VCO could wreak havoc with the transmitter combiner if it lost lock.

[regenfreak]

Of course, I never implied you would use the whole portion of the log graph because FM tuning ratio is very small anyway. As I wrote, you often use it with a series padder cap or two varactors back-to-back.
With two back-to-back varactors, you half the capacitance with reduced tuning range but the response is more symmetric.
For oscillator frequency 98.7-118.7MHz, the frequency ratio is 1:1.2026 so the capacitance would be 1.4462:1.

If you read the article in my post 8

Next, it is seen that the Clapp oscillators are comparatively compact as they employ a relatively small capacitor to tune the oscillator over a wide frequency band. This is because, here, even a slight change in the value of the capacitance varies the frequency of the circuit up to a great extent.

If you pick the less "non-linear" portion for reversed voltage from 5V to 10V, you get 15pF to 7.5P with a capacitance ratio of 2.0 which is way more you would need for 1.4462. I am neither an expert nor an electronic engineer.

[regenfreak]

But you never use the graph to calculate the tuning range regardless because you have to include the series and parallel capacitance of the LC in practice.

[Radio Wrangler]

Originally Posted by John_BS

My ex-boss (the aforesaid Johnstone) was keen to point out that a synthesised VCO could wreak havoc with the transmitter combiner if it lost lock.

Indeed it could. Most synthesiser loops (not all) use phase-frequency detactors and have integrator loop amplifiers with huge gain at DC. If these go out of lock, there has to be a definite failure somewhere in the loop. They inevitably drift off to one end of their range and sit there sulking giving an output at an extreme frequency.

The less common sort of PLLs use pure phase detectors or samplers and have controlled gain at low frequencies in their loop amplifiers. These tend to either modulate the output with a beat frequency or they twitch erratically, somewhere around the middle of their tuning range.

Neither of these behaviours is desirable to drive a transmitter. Ideally, you want a 100% trustworthy lock detector operating some means of killing the output if things go wrong.

Sometime around about 1990 The BBC bought a bunch of HP 8656 synthesised signal generators for use as drivers for their World Service transmitters. I never heard of any problems.

David

[Skywave]

Re: post 11, q.v.

I built a 'bread-board' style of this cct. on a chunk of copper-clad fibre-glass board as a ground plane. That build produced a very spiky non-sinusoidal waveform at approx 72 MHz and no FM whatsoever. I tried different varicaps - all of which had tested good on my basic 'varicap test rig' - but all to no avail as far as producing any trace of FM in this cct.
When I met this cct., I was intrigued by the basic oscillator design. I had no need for this cct's. basic function - and don't have now: I was driven by simple curiosity.
However, right now, I have many bigger fish to fry, so I am not going to investigate this cct. further.

Thanks to all members who made useful contributions to this thread, which from only my viewpoint, can now be closed.

Al. / Aug. 12, 2021