Phase or frequency modulation?

Is your "crystal oscillator" a packaged VCXO? They generally lowpass the frequency control input, the varactor thing, pretty hard.

If your rig is all tubes, probably not. But I suspect the rolloff is in the transmitter, not the receiver.

Good question. It's a shame the All Identified Signals database

Danke,

Good question. It's a shame the All Identified Signals database

Danke,

Probably nowhere. Analog PM stinks—it’s no better than AM at low modulation index, and atrociously wasteful of bandwidth at high index.

Cheers

Phil Hobbs

No

The circuits are at:

As the gain of the reactance valve is varied by the audio signal on its grid, a variable amount of 90-degree phase-shifted signal is fed into the crystal oscillator frequency trimming inductor in the anode circuit of the reactance valve.

The RC time constant in the grid circuit is 3dB down at 3.4 Kc/s. I have tried removing most of the top-cut capacitors between the audio clipper and the input to the modulator but this made little difference as all those time constants took effect above 3 Kc/s.

The initial tests were done with the experimental transmitter but the audio response figures of the Icom were taken with a Marconi TF 2016A signal generator. This has an internal meter which allows the modulation level to be accurately set and monitored. The audio source was a Solartron CO 546 Wein-bridge oscillator which is stable to + or -

The sig-gen tests confirmed what my ears were already telling me.

Why not measure the FM and see who the bad guy is?

It's the former, and "it's a feature, not a bug".

As I understand it: the normal convention on the ham bands is to apply a 6 dB/octave high-pass equalization to the transmitted voice signal prior to frequency modulation. The time constant puts the "knee" of the curve above the voice band.

During reception, the signal from the discriminator is fed through a corresponding low-pass filter ("knee" below the voice band) before being fed to the audio amplifier and outputs.

It's a process similar to what's done in commercial FM broadcasting, but with different time constants in the filters.

As I understand it, this was done for two reasons: to reduce the incursion of high-frequency noise into the audio signal, and to allow compatibility with PM transmitters (which don't require or use the high-pass filter).

So, what you are observing is probably this: the signal you're transmitting (from your own modulator or from your test oscillator) is frequency-modulated, but the necessary pre-emphasis of the upper frequencies is not being applied. When the signal is demodulated by the Icom, it's passed through the de-emphasis low- pass filter, and what you're observing in your frequency chart is the response of that filter.

Many of the higher-end VHF ham radios have a feature which is designed for use with high-speed TNCs using the G3RUH modulation (direct FM of the carrier) at 9600 baud. In "9600 baud" mode, the TNC-input jack is disconnected from the microphone-input path (which applies the high-pass pre-emphasis) and feeds the frequency modulator directly. The TNC-output is fed a signal coming directly from the FM discriminator, bypassing the low-pass de-emphasis.

So, you could try putting your Icom into "9600-baud" mode, and look at the discriminator-output signal at the "to TNC" output jack. I'm going to guess that you find that you see a pretty flat frequency response from the receiver when you feed it a signal from your experimental modulator.

The FM was coming from a high quality signal generator, I have no reason to suppose it was causing such a bad frequency response.

Yes, that would make slightly less than 3dB down at 3 Kc/s - not the 18 dB down which I measured.

That makes complete sense and explains what I measured. (As an aside, it probably makes the manufacturer's specification look better too.)

The question now is where to put the pre-emphasis in the feed to the modulator? If I put it in front of the clipper it will be rendered ineffective by clipping - but if I put it after the clipper it is liable to over-deviate the signal.

There is already some pre-emphasis generated by the 100pf capacitor coupling between the anode of the first valve and Grid 1 of the second valve. I can further increase the pre-clipping effect by substituting a

If the pre-emphasis is needed after the clipper, I can take the feed from the anode of the clipper directly to the top of the 100k pot through a small-value capacitor and omit the top-cut network. My worry is that this could cause 'sideband-splash' into adjacent channels and lead to complaints.

Thanks, I'll look into that.

I meant measure the actual FM from your crystal oscillator.

I wonder if the crystal's Q limits modulation bandwidth somehow.

I have no way of making those measurement other than with a VHF receiver.

I had wondered about that but the measuements I made with the signal generator indicate quite clearly that the large drop in audio HF response is mainly (if not entirely) caused by the receiver. The reply by Dave Platt ofers a completely plausible explanation for this and confirms something I had suspected but couldn't find stated in any of my usual sources.

When I have added the appropriate gross pre-emphasis to the transmitter I shall be able to hear what I am doing and can then set about correcting any remaining nuances of the modulator's performance.

Use a receiver in CW mode. That will heterodyne the FM down and zoom up the deviation. Then look at the result on a scope to see the period variation.

A decent digital scope should display FM deviation directly.

I only have analogue equipment and the test with a signal generator has already told me what is going on. When I have the modulator working, I can easily check for unwanted harmonics by listening on the two adjacent channels.

Crystals modulate much faster than f0/Q, interestingly. The mechanical movement and the piezoelectric coupling are a single degree of freedom, governed by an ordinary differential equation.

ODEs have no internal state, so the oscillation responds instantly to changes in the capacitive load.

Something like a SAW resonator has lots of internal state, and so takes much longer to respond.

Cheers

Phil Hobbs

I presume you are only talking about frequency 'pulling'; any change in amplitude would be governed by the ratio of internal oscillatory energy to energy extracted or added.

I have never thought about amplitude-modulating a crystal oscillator before.

Interesting!

What are ODEs?

Ordinary differential equations. The crystal mode is a mass-spring oscillation, so an instantaneous change in the spring constant causes an instantaneous change of frequency.

There’s no need for the resonant energy to die out, unlike the case of external forcing, where a change in the forcing frequency takes on the order of Q cycles to change the response.

Cheers

Phil Hobbs