RCA THEREMIN

"I suspect these will come in handy" - w0ttm

LOL ! - You dug into your junk box, and found a pair of 27's, and some 24's ? .. Hell - Im starting to think there may be metaphysical events occurring! ;-)

I was thinking that if I really wanted to see what was going on, I would need to get these tubes and get someone to build a couple of oscillators and mixer for me to test - but the idea was not appealing - all the other stuff like PSU etc, and my limited space..

So I will spend my time in the virtual world - working my way through the theory.. first challenge is to understand the tubes well enough to construct some reliable models - then I can "test" the circuits on my laptop - if a real set of data comes along, the simulations can be compared against reality.

I really dont think there should be a huge problem generating the waveforms and mixer behaviour using modern electronics once I know whats really going on.. I believe I probably have a method of achieving the required end result in my hands at the moment - I have a working additive engine which allows harmonics to be mixed to produce any waveform - but its complex and expensive (each harmonic adds about £10 of actual component cost) and not disclosable at this time.. So I would be looking for simpler techniques.

Thanks Rob and Thierry - Your thoughts about whats going on at the tube are really helpful - And yes.. What you say makes sense.. my idea of the tube as a some sort of glass encased fet was a bit simplistic ;-)

Fred.

Found this article in Popular Science (Aug 1929) which discusses some of the "new" tubes - in particular the UY-224.

Damn diagrams!

Can I confirm something - I am reasonably sure I am right, but..

On the "Art Harrison, Rev.11 3-4-04 " diagram, the VPO is shown wired (via R16) to a grid marked SG  (the Screen Grid, which, in the tube symbol, is drawn close to the cathode) - The Reference Osc is shown connected to a grid marked CG - ( the Control Grid, which, in the tube symbol, is drawn close to the plate).

As I see it, this is misleading - What is wrong here (im sure something must be!)? - As I understand things, the SG is closest to the plate, the CG is closest to the cathode.

Is the symbol wrong? - Looking at the awful RCA schematic it looks like the grid closest to the plate is connected to the VFO..

Fred.

You are right, the screen grid is closest to the plate (or anode as we say here) and the control grid is closer to the cathode.

In all three theremin schematics which I looked at (original RCA, Rockmore, Rosen) the variable pitch oscillator is connected to the screen grid.

This is IMHO not a problem since the screen grid is not biased as a screen grid but rather at a second control grid (but with higher leak current and capacitance). That's why you find a smaller potentiometer (0-25k) in series with it in the Rockmore theremin, while the control grid has a 0-50k series pot.

In the Rosen Theremin you'll find fixed resistors (1k between VPO and SG, 15k between FPO and CG).

Thanks for the confirmation, Thierry -

I was really wanting to check why, on the RCA, there was a resistance from the VFO to the 224 (mixer) grid, but none from the Ref Osc to the grid - My confusion came about because I believed this resistor would make more sence on the grid closest to the anode, not the one closest to the cathode! .. I just read the symbol as if it was right without paying notice to its labels..

I am slowly getting there I think - trying to understand the operation.. but getting the position of the grids muddled in my head is a bit like getting a symbol of a PNP transistor in a schematic when it should be an NPN.. Easy to see the error when you understand transistors and their circuits - but not helpful to someone trying to learn about transistor circuits!

What I really want now is some charactaristic curves for the 224 and particularly 227 tubes.. I have obtained software which allows me to trace over a .gif image, enter parameters, and (with some fiddling) hopefully generate data I can feed into my spice model parameters, so I can model these tubes..

its ok - found these in the RC-10 manual (not shown in the selected tubes from this manual)

Fred.

"(1k between VPO and SG, 15k between FPO and CG)."

Errr.. You sure? (my head goes back into "I am now confused again" mode) - Shouldnt the higer resistance go to the SG, as this is closest to the Anode?

Yes I'm sure and it seems to be logical to me. The screen grid has a larger leak current and a larger capacitance, thus a smaller resistor is needed for the same bias and the same time constant.

BTW: Lucie's theremin had a much more nasal and a less stringy timbre than Clara's.

Fred, don't waste too much time with the 227 and 224 tubes as I did. The secret of the waveforms is elsewhere in my opinion. I found out that both interstage transformers were not 3:1 or 5:1 as one would normally expect, but 1:5. The first audio transformer is directly connected to the grid of the (high z) control grid of the 1st audio stage, there is no impedance matching which makes that the 224 sees an inductive load at the plate and the 1st audio 227 will most probably be (over-)driven by a kind of pulse response with over- and under-swings and will, since there is only a weak negative bias (control grid leak current x 5000 ohms DC resistance of the secondary winding of the first transformer) rather act as a simple switch for lower audio frequencies, creating another pulse in the second audio transformer. 

That's only my theory, but it could explain the unexpected waveforms, which made me rather think of deflection stages in old crt tv sets (or the simple switched ignition systems in old cars) than any RF mixer.

Ok - Just checked the available diagrams I can find.. Lucy Rosen's theremin has 15k to the grid shown closest to the anode on the symbol, and 1k  to the lower grid (closest to the cathode) -

This (if the schematic symbols are right) would mean 15k to SG, 1k to CG.

But is the symbol right?

Clara's theremin schematic (drawn by Bob Moog) shows 0-25k variable resistance to the symbols "upper" grid (closest to anode) to the VFO.. - and this grid is numbered as "3"  ..

The  Ref osc is wired through a 0-50k variable resistor to what is shown on the symbol as the grid closest to the anode, and no number is given to this connection. The anode (plate) is numbered as "2" and the cathode is numbered as "4".

Looking at the numbering on the 27 tubes, these show Anode = 2, Grid (control grid) = 3, and Cathode = 4.

Looking at the socket connections, both the 24 and 27 share the same footpring, but the "grid" connection on the 27 is replaced by a "screen" connection on the 24.. The control grid on the 24 being connected to the cap.

So everything in Bob's diagram looks right - the upper grid in the symbol is the screen grid, and the lower grid is the control grid, the VFO is connected to the SG, the Ref oscillator to the CG, as the RCA.

Lucy Rosen's theremin schematic also follows the RCA theremin topology if the schematic and symbol are correctly drawn - VFO to SG via 15k, and an additional (as in, not fitted in the RCA) 1k between the Ref Osc and CG.

Clara Rockmore's theremin's schematic is probably the most reliable we have, having been drawn by Bob Moog.. This again follows the RCA topology, but the interesting thing IMO, is the inclusion of variable resistors on both SG and CG.. Why is the CG value adjustable to 50k, but the SG value only adjustable to 25k ?

Is it perhaps because the impedance of the CG is much higher than that of the SG, so in order to effect any significant change on the level / waveform of the REFO signal, a higher value resistance would be needed?

Yeah - its all slowly starting to make sense..

Fred

>> It appears that the convention in diagrams is for the screen grid (SG) to be drawn below the plate (anode) and the control grid (CG) to be drawn directly above the cathode.. This is what RCA drew, its what Bob Moog drew, there is a question mark still remaining over the diagram of Lucy Rosen's theremin diagram http://www.caramoor.org/images/Theremin_diagram.jpg 

The only obvious case where a screen grid (SG) was drawn as the closest to the cathode was Art Harrison's drawing of the RCA.

"The first audio transformer is directly connected to the grid of the (high z) control grid of the 1st audio stage, there is no impedance matching which makes that the 224 sees an inductive load at the plate and the 1st audio 227 will most probably be (over-)driven by a kind of pulse response with over- and under-swings and will, since there is only a weak negative bias (control grid leak current x 5000 ohms DC resistance of the secondary winding of the first transformer) rather act as a simple switch for lower audio frequencies, creating another pulse in the second audio transformer." - Thierry

Ouch! - That sounds like something from a Tab book! ;-)

Heres my approach - you probably went a similar route..

My first objective was to understand the oscillators - the next is to try to understand the mixer.. All I have to work on is the "hands on" data from people like Hobbs - So when I hear about "balancing" the inputs to the mixer, and how this affects the quality of the sound, I must explore the possible reasons..

Add to this the tonal differences between the RCA and the other two Lev theremins we have data on, and the spotlight seems to settle on the mixer - at least, we can say that changes here have a direct effect on the sound..

Clara's theremin sound is, IMO, more desirable than the sound from any of the others, it is (circuit wise) also unique in that there is a variable resistor coupling to the AF pre-amplifiers grid.. and the other variable resistors.

What I am aiming for is to get an output from the mixer which is close enough to what is seen on Clara's theremin.. If with solid state we can get that, I think (hope wish ;-) that  with filters (perhaps even employing gyrators) we can emulate the resonant behaviour of any audio components (transformers, speakers) which follow the mixer, and add any other kind of filtering required to achieve this.

It may be futile - The best approach may be to forget about whats inside the "originals" and just design new circuitry which is crafted based on the output signal - I have been studying every sample I can get my hands on, and believe it should be possible..

I certainly have no interest in using tubes or audio transformers - I would, quite honestly, sooner go to a DSP which could be programmed to simulate the sound.. I know this is blasphemy! ;-)

I also have a strong possibility that I will not be able to complete this "project" - So my main focus right now is to cover as much ground in terms of understanding and documenting (here) as much as I can to enable others to pick this data (and the contributions from yourself and others) and take the project to completion.

Fred.

"I would, quite honestly, sooner go to a DSP which could be programmed to simulate the sound.. I know this is blasphemy! ;-)"

The analog engineer half of my brain agrees, but the musician half doesn't care, as long as the tone is there. Yes, I do have DSP effects in my guitar audio chain.

I was unable to get any FET's, so I'm just going to order some, as well as IFT's, inductors, etc. I'll just keep playing with the sim and PC board CAD until they arrive.

In my oscillator simulations, I've increased the source resistor a bit, and sent a signal from that point to the inverting input of an op amp and LED to act as a tuning indicator. The non inverting input goes to a pot to set bias. Minimum brightness is at antenna resonance. The schematic is so basic that I'm not going to bother with posting it, unless asked.

In simulations at least, this oscillator loads down at antenna resonance a lot less than any other I've simed, including uncle Bob's EW. My Kustom, in sims and real life almost stalls at resonance.

I plan to build this project first, then finish my Kep, but after further excavations of my junk box, I believe I now have enough vintage parts to build a Clara clone. I even have several AF coupling transformers. All I lack are the big coils. I'm a bit torn on that subject. Modern inductors should work a bit better, but those big ones look so cool. 

Why not just attempt to build a clone and see what does or doesn't work?

Surely a clone of an RCA will have to use valves (tubes)?