Armstrong's Audion principle

Comparing the efficiency of the regeneration principle according to Armstrongs US pat. Nr. 1,113,149 with the more widely used inductive feedback (von Bronk) questions were raised about its features. Because this functionality is not so obvious an analytical approach should help. In the enclosed article a simple formula is derived which describes the impedance at the grid of a triode dependent on the plate impedance. The resulting expression can easily be implemented on a graphical calculator for optimizing regenerative audion circuits.

Thanks go to Joe Sousa for encouraging me to post the article and to Prof. D. Rudolph for giving valuable improvement comments.

More comments are welcome!

Felix A. Schaffhauser

Attachments:

• input impedance of a triode with complex load (153 KB)
• Input impedance of a triode with complex load (112 KB)

To thank the Author because you find the post helpful or well done.

 Nicely done and well presented.  I've added this to my archive.  Thanks

To thank the Author because you find the post helpful or well done.

Congratulations, Felix, on the fine work! This will be now a standard reference for me too.

Your work also helps clarify that most 1920's radios were under the effect of Armstrong's Audion principle, unless they were explicitly neutralized, or had other means to eliminate the inductive plate load. However, Armstrong understood it and applied it in a controlled fashion.

Your analysis is also applicable in cases for non-neutralized RF amplifiers with inductive loads. One example is my 1924 Silvertone-Travler with a single tuned loop and two broadband transformer-coupled RF stages. The audio transformer load at the Audion detector also allows for some Armstrong-style regeneration.

The "Volume" control of this radio places a variable positive bias at front stage that also serves to control the regeneration between the high Q tuned loop and the inductive plate load. This stage would be the most regenerated stage because the low loss loop does not need a lot of reflected negative conductance from the inductive plate load for oscillations to occur.

Regards,

-Joe

---

p.s.: Audion was the name that Lee DeForest gave his triode detector.

In Germany, Audion has come to mean a grid leak detector. Audion in the US usually just means an early triode made by DeForest.

DeForest came up with this word with the contraction of Audio and Ion. Audio, because it produced sound, and Ion because he found that the gas was essential for sensitive detection. This also derived from the work that DeForest had been doing with flame based detection.

The presence of gas did make the detector more sensitive with internal plasma regeneration, albeit more unpredictable and more easily saturated. RCA made a soft vacuum detector based on this principle, it was the 00A.

To thank the Author because you find the post helpful or well done.

Thank You James and Joe, for the nice comments.

It also occurred to me during my analyses that most of the receiver sets of the 1920's and maybe after, somehow were affected by the feedback explained in my paper. They mostly used either transformer coupling between the stages (as Joe's "Silvertone travler") or/and inductors as plate loads. Earlier I wondered why this relative complicated approach (winding coils etc.) was chosen so often. Of course the most obvious reason is that with the relative low plate battery voltage used less volts are "wasted" compared to resistive plate load. Next: transformers allow for an optimal coupling from stage to stage so the signal level is improved. And last but not least, the regeneration effect caused by the inductive load aided in sensitivity and signal level as well. Keeping in mind the relative poor performance of the tubes (at a considerable price!) at these times and the weak transmitter power, sometimes far away, it was important to get as much out of a set as possible.

I think it is important to understand the cause and parameters of the feedback caused by an inductive load in order to control the stable operation of such sets and being able to neutralize the sets after tube changes or due to aging effects of the components. For this, I hope, my little program can help.

Felix

To thank the Author because you find the post helpful or well done.

> I wondered why this relative complicated approach (winding coils etc.) was chosen...
> ...less volts are "wasted" compared to resistive plate load.
> transformers allow for an optimal coupling
> regeneration effect caused by the inductive load aided in sensitivity

Well, and:

In the 1920s, resistors were NOT such a standard part. Coils could readily be hand-wound from standard wire. Reliable resistors came later. Especially in the higher values used for plate-load. (30-ohm resistors for filament heat adjustment were common heating (cooker) wire, but you can't easily make a 10K resistor with heating wire.)

Yes, the AC/DC ratio is far higher with coils.

"Coupling" is a partial answer. Plate to grid transformers are often near unity turns ratio, although tuning and separation affect gain.

The "big" problem with resistance coupling in RF is the stray capacitance. A simple R-C network will roll-off. For 10K resistance and say 15pFd stray capacitance (and maybe a little Miller), gain is down 3db at 500KHz. When a coil is used instead of a  resistor, and tuned to match the signal, gain can be 10db or more higher. 10dB added gain was a LOT in triodes.

The "big" advantage of coils, aside from gain, is selectivity. As soon as there were two transmitters we needed a way to pick one or the other. (Or to discount God's transmitter: random thunderstorm static.) By the 1920s there were many dozen broadcast transmitters. AM has no "lock-on", it will (with simple detector) pick up ALL stations at once. We must filter-down to the one station desired. This needs a fairly steep-side filter. In fact radio design starts from both the desired gain AND the desired selectivity. We need several tubes for gain AND we need several L-C tanks for selectivity. This leads to the 2-tube (and 2 tanks) for local reception with few interfering signals, the 3-tube (and 3 tanks) later when more stations were on the air and it might be desired to get a distant station instead of a local station.

Here's a hasty/sloppy analysis. The simple R-C stage with 15pFd stray is falling before the broadcast band, will pick up "all" stations at once, but the lower the better. Adding a coil to tune the 15pFd stray can give a lot more gain at 1MHz, a typical broadcast station. But "stray" is very variable with tube, with voltage, with hands or bodies near the radio. But if we add 185pFd of "good capacitor" (not random stray) we can use a smaller coil to get slightly higher gain and much better rejection of other frequencies.

To thank the Author because you find the post helpful or well done.