RCA THEREMIN

Over on the "scratchpad" thread, two seperate lines of investigation are sort of happening.. One being the design of a RCA "Clone" using semiconductors, the other being exploration of the theory on how the original RCA functions.

Both activities are needed - without understanding, one cannot replicate - or one may simply try to 'replicate' but miss vital aspects and have a "clone" which is clearly not a "clone" (my first ignorant design made assumptions which were sensible, but completely wrong)

I want to move the mainly theoretical discussion to this thread, and leave the other thread for the 'practical' aspects - there will be overlap -

Data on RCA components :

To fully verify the understanding, anyone having unpublished data or measurements of actual RCA theremin component values - Please, lets have these! - In particular, the inductance values are needed for confirmation.

I believe I have a copy of everything published to date (RCA Manuals, Art Harrisons analysis, Construction notes etc).. but am puzzled by some inconsistancies on these.

In particular, I think there is a tendency for those analysing the RCA to take "shortcuts" in the same way that I originally did - For example, you know the tank resonant frequency (172 kHz or therabouts) you know the tank capacitance (say 1000pF) so it is easy to calculate the tank inductance as 598uH.. This is close to the value I calculated, and is the value given for each T8 winding by Art Harrison in his documentation..

But it is wrong! - This value worked with my simulation when I built the simulation circuit the way I thought it should be.. An Armstrong oscillator with normal tank and tickler coil.

But the Lev oscillator isnt a standard Armstrong oscillator - When I "built" an oscillator exactly like the RCA, the simulation returned completely "wrong" results - the frequency dropped way below 100kHz for starters.

At this point I was puzzled (to put it mildly) and asked for data on what the real inductance values for T8 were.. Rob (w0ttm) calculated the inductance based on the physical (winding and diameter) data for the coils, and gave a value of 165uH - Plugging these values into the simulator, the frequency became same as that of the RCA.. But completely wrong for a standard Armstrong oscillator.

Understanding the oscillator:

Thr reason that the inductance calculated from circuit theory (known C, known F, unknown L - simple to calculate L..) is that the circuit is a series resonant, not parrallel resonant one - The two inductors are also coupled, and this coupling increases the combined inductance - it is a circuit like this:

----T8 Winding A ------ 100pf fixed capacitor --- 100pf Variable capacitor --- T8 Winding B --

All in series, with winding A and B coupled, quite tightly (using coupling of .95 in the simulation, with value of 165uH gave close to the right frequency - we know that the theoretical inductance required is is about 600uH, and the additive and coupling (effectively increasing the number of turns) result of coils wired as above gives this kind of value.

Then there is the real simulation challenge - the antenna circuit 'connected' across the lower (B) winding.. Having an antenna circuit across a parrallel tank is usual territory, having it across one half of a series resonant tank, well, thatsv new (to me, anyway!)

Assumptions:

I think wrong (but understandable) assumptions have been made when the RCA circuit has been looked at in the past.

The first "mistake" is probably calling the oscillator/s an "Armstrong" oscillator - It is certainly derived from an armstrong type design, but I believe it is probably a different type of oscillator - probably a Lev Termen invention.

Fred.

Here is the RCA pitch oscillator schematic, a re-drawn schematic which makes the series connections clearer, and a diagram showing just the tank and antenna resonant circuits.

Original:

Redrawn:

Resonant Circuits:

FULL UPDATED ANALYSIS CAN BE FOUND HERE: http://www.thereminworld.com/Forums/T/28796/rca-theremin?Page=5 (Page 5 of this topic)

from levnet, today:

"Rob, and all:
 What I was able to discern is that the RCA/Termen designed oscillator is a modified Armstrong oscillator, as you say. But Termen took it a step further. He biased the oscillator such as to not only oscillate at about 170 Khz but also pulsed like a blocking oscillator. In effect he had two oscillators in one.
 It is the supersonic frequency of the blocking oscillator, about 20 Khz, and its harmonics, superimposed on the r-f frequency that makes the sound unique at the mixer output.
 I believe that Termen was influenced by the operation of the super-regenerative detector which was an amplifier that  also oscillated at a "quench" frequency of 20 Khz to achieve a higher gain. Termen was not interested in the gain factor but the harmonics generated by pulsing the r-f frequency for the harmonics that were generated.
 Brilliant? I would say so, indeed!"
Uncle Howie

Also, for further reference;  here is a clear and complete schematic of clara's theremin,  by Adrian Bontenbal - he also built a "Clara" replica

-An "evolved" RCA;  the addition and location of variables (pots and a  cap) are   good indicators of where to tweak

That explains the multiple resonance points Fred found.

Super-regenerative design is not in my comfort zone. Everything I've played with has been superhet or direct conversion. I did build a standard regenerative receiver once, but it was a kit and I didn't take time to analyze it.

Time to hit the books.....

I've also studied the RCA and Rockmore theremin schematics for a long time, thus it could be that "Uncle Howie" is right with his regenerative pulse theory but I am not at all sure about that.   And even if he is right, I'm mot sure about the post-mixer effect of that. 

I'd really like to spend an afternoon with an oscilloscope and an RCA...

And I'm still not convinced that it is possible to emulate the RCA sound with FETs and without audio transformers... FETs act with a Uds>Up as voltage controlled current sources as pentodes do. But there are mostly triodes in the RCA which have a much greater output resistance (which depends also on the plate voltage) than any FET and thus behave more like voltage controlled resistors. The mixer tetrode will behave in the same way since the screen grid is not biased at +90V to reduce the output resistance but it is used as a second control grid...

I think that there are still lots of things to discover and to explore in Lev's designs and I'm happy that this discussion has been started here.

come on over ;-)

Here still a small "gift" for all those who want to refresh their knowledge: The RCA Radiotron Handbook from 1932 http://db.tt/acQ50G5G

"also the Volume Circuit?   Performance-wise it is pretty sluggish" Chobbs, from the scratch pad thread.

Clara's volume circuit is different from a stock RCA, but has the sought after tone.

I'm with Fred on this point. An opto-FET ought to serve nicely here and have much faster response compared to the original RCA.

Also notice that Clara's can switch different, un-bypassed resistances into the cathodes of both oscillators and the mixer that would change the bias point and gain.

"And I'm still not convinced that it is possible to emulate the RCA sound with FETs and without audio transformers..." Thierry.

I don't think we can exactly duplicate "the sound" without duplicating the whole thing, but I bet we can get darn close.

The output transformer in my tube guitar amps plays an important role in generating the desired sound. I've been able to get very close with silicon, but never equal. The difference is subtle, and to be honest, if played side by side there would be few people able to tell the two apart.

"So you think mpf102's will work?"  Chobbs, from the scratch pad thread.

Probably. I've used them as amplifiers at 50 mhz. The only reason I was going to use them is that I can get them at shack.

"And I'm still not convinced that it is possible to emulate the RCA sound with FETs and without audio transformers..." Thierry.

Nor am I, Thierry.

This was a first avenue of exploration - unknown territory.

I am convinced that, when we fully understand the RCA, we will be able to get extremely close to emulating it - We cannot exactly emulate it because, for one thing, no two RCA's are exactly the same! - No two tubes are exactly the same.. no two transformers are exactly the same..

Sorry if the above seems trite - its not meant to be! - Its just that I view this whole matter of "emulation" as far more than just 'copying' the electronics - To me, the critical aspect is what comes out of the "clone" at line level, versus what people have heard the RCA sound like from recordings taken via a microphone from the RCA loudspeaker..

Even if we produced a perfect electronic clone, it wont sound like an RCA unless played through an RCA loudspeaker - So we DONT want to produce an electrical output which is like the output one would get electrically from an RCA, we want an output which is close to what one would get from a microphone pre-amp connected to a microphone pointing at a RCA speaker being driven from an RCA theremin.

And the above, IMO, makes discussion about the effect of the audio transformers etc somewhat redundant.. As I see it, we need to get the oscillators, mixer and volume/VCA right - after that, we need to craft the following stages so that what comes out is like I described above - a composite of all the audio 'coloration' including the speaker..

I believe the above can be achieved using audio filters - probably quite complex filters.

Fred.

ps - thanks for the tube manual.. I have it somewhere on one of my HDD's, but this link was real timely .. ye gods, it takes me back - I was studying that manual when doing my ICS course back in 1970 ;-) .. Funny to think that, if I was to take the exam I took back then, I would fail now!