Pulse-Counting FM Broadcast Receiver

The goal of this project was to build a companion vacuum-tube FM receiver for the OTL Audio Amplifier I had built for my office and to use as many parts on hand as possible.

I found this article that described a tube-based FM receiver that sounded simpler to make than the ones that I saw when I was a kid. The advantages of this design were:

A ~150kHz IF which does not require interstage coupling transformers (which I didn't have).
A detector that does not require tricky alignment like a ratio detector or discriminator
A novel circuit that would be fun to explore.

The author, John Hunter, wrote out a detailed description of how it worked as well as how well his various designed worked. It is a most excellent and useful site and John provided lots of helpful advice to me along the way (THANKS!).

Theory of Operation

Fundamentally, demodulation of an FM signal requires converting changes in frequency of the IF (in this case 150kHz ± 75kHz) to changes in voltage corresponding to the audio signal.

Traditional FM detectors rely on modified versions of slope detection. Slope detection sets the center of the IF on one of the 'side skirts' of a tuned circuit's frequency response. As a result, changes in frequency move the signal up and down along the slope of the response, converting changes in frequency to changes in amplitude.

Pulse-counting FM demodulation uses a different principle. This is described in two articles in Wireless World from 1956:

Low-Distortion FM Discriminator M.G. Scroggie April 1956 (page 158)
Unconventional FM Receiver M.G. Scroggie June 1956 (page 258)

This is shown below:

The VHF FM signal is down-converted to a low frequency IF (~150kHz).
The IF signal is amplified from a few mV to several volts RMS.
The sine wave signal is converted to constant-duration pulses.
These pulses are integrated by a low-pass filter with an audio-range time constant.
The resulting AF modulating signal is then amplified and enjoyed.

RF input

(88-108MHz)

Frequency Converter

IF

(150kHz ±75kHz)

IF

amp

sine wave

Limiter &

Pulse shaper

pulses

Integrator

(LP-Filter)

Audio

Signal

Out

Why is it important in (3) that the input to the integrator be "constant duration pulses"?

Because the integrator/LPF is averaging the input signal, the average value of the input signal must change with changing frequency.

The figures below show this:

75kHz 150kHz

The graphs compare "squaring up" the sine wave (through a very high-gain amplifier) with "pulses" triggered by the rising edge of the sine wave.

Notice that, as the frequency increases, the width of the "square" wave gets shorter, so the average value remains constant. However, since the "pulses" are constant duration, the average value increases, reflecting the increase in frequency.

In this case, the output of the integrator/LPF is, at least theoretically, a completely linear function of the input frequency over a wide range of frequencies. This is different than the slope-detection-based methods which depend on linearizing the non-linear response of a tuned circuit.

Both Scroggie and Hunter use differentiation of the very distorted (almost square) output of a limiter to generate very short pulses of fixed duration. This has the advantage of simplicity but it has two drawbacks:

unless the limiter is fully saturated all the time, the pulse amplitude will depend, at least a little on the amplitude of the input signal
the pulses are relatively short, making the output of the integrator/LPF low amplitude

Another way to generate pulses of constant duration is to use a monostable multivibrator. This has the advantages of:

generating longer pulses, thus yielding higher AF signal amplitude
greater noise immunity as, while the monostable is going through it's triggered cycle, it is immune to re-triggering by noise pulses

Using a monostable multivibrator in a pulse-counting FM receiver is not exactly an original idea. There are solid state examples (including a patent) on the web as well as some discussion of this idea.

To be continued: