Continued from Part 1 re-introducing GenRad.
= = = = =
GenRad’s chart recorder followed a common pattern, with the usual GR extras to satisfy peculiar needs. The common pattern was a pinfeed drive for the paper and a solenoid controlled by negative feedback pushing the pen back and forth.
The pen was driven by a differential solenoid fed by a differential amplifier. One side of the differential amplifier picked up the incoming signal from the task being monitored. The other side was fed by a rheostat that tracked the pen. This strictly regulated the movement of the pen, regardless of inertia or friction. If the pen tried to overshoot, the rheostat would sense it and pull it the other way. If friction slowed down the pen, the rheostat would pull it more strongly forward.
I used to play with chart recorders. You could push and pull the pen and feel the feedback fighting your foreign motion, just like the gamma system fighting friction or overshoot in a human muscle.
Here’s a chart recorder used in sync with a GenRad audio oscillator, to perform an automatic hearing test. Our Martian is the subject, and he is indicating when he hears the tones by pushing a lever.
The chart recorder has a steady-speed motor driving the chart forward, without slippage thanks to the pinwheels and perforations. The accessory chain drive moves the frequency control of the audio oscillator in sync with the paper motion, so the horizontal axis of the chart represents frequency and the vertical represents responses.
Here’s a closer view of the pen itself.
= = = = =
The oscillator is also interesting. It uses the same principle as a superhet receiver, but instead of demodulating an incoming signal it’s mixing two freqs in the same range, well above audio, to get a difference freq in the audio range. The purpose is similar, though the output and input are converse. The parts needed to firmly regulate an audio frequency are far too large for a practical piece of equipment. When you get above 100 Kc, the coils and capacitors are a more manageable size. By generating two waves around 200 Kc and using the beat or difference, all the parts are practical.
I’m showing the process here. In the schematic, 1 is the variable oscillator, 2 is the fixed oscillator. 3 is the mixer, 4 is the rectifier, 5 is filter and amplifier.
I’ve used a slow wave for (1) the variable, a faster wave for (2) the fixed. At 3 the two frequencies have been modulated or multiplied, giving a complex pattern (3) with two main frequencies. The two freqs in this wave are the fast-moving sum of the two originals and the slow-moving difference of the two. At 4 I’ve rectified the result so it’s only on one side; then at (5) I’ve filtered out the faster wave giving only the slow difference frequency.
= = = = =
In the digital world, though the methods are different, the same size factor applies. A precise digital audio oscillator starts with a crystal generating a high freq in the megacycle range, which can be done with a miniature crystal. Instead of modulating to form a difference, the digital system calculates a very long stepped sine wave, with very small steps for each pulse of the original crystal. This sine wave can be nearly pure, requiring only a filter to get rid of the tiny wiggles.
A crystal that could oscillate at 10 cycles would probably be several feet in diameter, and I suspect it wouldn’t even vibrate at all.
Direct generation of very low frequencies basically requires either a physical pendulum like the Riefler clock, or a fluid pendulum like the Vreeland oscillator. You could also do it with a literal analog sine creator like a geared-down sinusoidal cam controlling a rheostat, powered by a synchronous clock motor; but I’ve never seen a device like this.
= = = = =
Part 3 delves into an earlier and much stranger GenRad oscillator and recorder pair.
Polistra's Mill 