RCA THEREMIN

I'm also coming up with pitch rod capacitance in the 12-14 pf range in sims, math, and experiments with my Kustom.

Re - Antenna capacitance..

I think that the problem with theoretical calculation of antenna capacitance may be due to several factors - as I understand it, the calculations are based on an infinite ground plane (horizontal) with the base of the antenna set at some height above this ground, and a vertical antenna from this point for given length, with the antenna having a given diameter.

One could hardly have anything further from reality than the above!

Theremins are (usually) operated inside rooms, there are walls having their own grounding properties, there are cables and metal pipes inside these walls, there are extraneous other grounded or conductive objects with their own capacitive or galvanic coupling to ground.. Even with walls / objects being quite distant from the antenna, the cumulative effects of these objects becomes extremely significant.

I have found the most practical way to accomodate the difference between theoretical "infinite free space" calculations and "real life" calculations is to add a constant "background" capacitance to the computed value.

I suppose that a multiplication of the computed value by some "fiddle factor" might work - but I have found that, apart from determining base-line value (which is always miles too low in the computation) the formula holds up quite well once a background "tweek" has been added.

Perhaps the best way is to forget about calculating the base-line value, simply select a value which works, and add the hand capacitance to this.

Fred.

"Perhaps the best way is to forget about calculating the base-line value, simply select a value which works, and add the hand capacitance to this."  - FredM

My spreadsheet does the calcs, which generally seem a bit low, so there is a "Cstray" adder fiddle factor. 

If the RCA linearizing inductor is around 70mH, then antenna capacitance is around 9pF for 200kHz operation.

One of the things that is crying out to be described mathematically is the capacitive behavior of a hand moving in the space between an antenna and the player's body - particularly when the hand is nearer to the body.  I've seen several videos where the Thereminist just about has their pitch hand on their chest.

W0ttm Said:

The volume antenna coil looks like a step down transformer with a high ratio. With some serious voltage across the primary, I can see the potential for substantial current in the secondary.

I agree with the above - but what puzzles me about it is the loading effect from the step-down secondary.. This current (which drives the heater) would, as I see it, greatly affect the sensitivity and general performance of the antenna...

The current through V5 is determined by the filament current - the hotter this is, the more current will flow - V5 is probably something equivalent to a current controlled resistor in series with a diode. 

The VCA itself looks to be a transformer coupled, single ended triode, with the supply voltage / current to the transformer controlled by the 120 triode heater in a stock RCA,

As I see it, the transformers role here is minimal (it merely couples the AC audio signal) - The gain is controlled by the current into V4's Plate, and V4 acts as a VCA (in fact, A CCA Current Controlled Amplifier)  because of this. 

or bias on a dual triode  -as in Clara's.

(I have used schematics from http://www.thereminworld.com/Article/14289/reader-re-creates-clara-rockmores-theremin)

I think Claras is an entirely different configuration - more like present day volume circuits.. The volume is not controlled by heater current - the oscillator is similar to the pitch oscillator, current to the antenna resonant circuit is coupled to a fixed frequency filter, and the amplitude of the ourput of this filter (which will be dependent on the Vol Osc Frequency) is taken to the grids..

These triodes look to me like they are configured in some sort of follower mode, with a high (5k?) value resistor, and I suspect that the voltage across this resistor is quite substantial - High enough to supply current to the anode (plate) of the VCA Triode through transformer T5 (which couples the audio in a similar way to the RCA circuit) and the triodes are also probably performing rectification of the HF volume signals applied to the grids. (The control voltage is taken from the junction of the two [3V and 18V] batteries)

I suspect that the triodes are biased so that the 210V supply on their anodes is 'split' - perhaps theres 150V across their Cathode <-> Anode when the theremin is silent / low volume, with 60V between Cathodes and ground, and this drops to perhaps 105V A<->C when volume is max, with C<->GND rising to 105V... These are only numbers I sucked out my thumb just to illistrate the idea ;-)

The significant differences in my view, are the replacement of the heavy current / slow response from the RCA's heater control mechanism, by a coupled tuned circuit into  high Z grids - a fraction of the current drain required by the RCA.

But I must re-state.. I really dont have a "feel" for tubes, have had no "hands on" expierience with them for more than 30 years, and could be spouting borelocks!

Fred.

Looking at Clara's theremin circuit, I really dont see the point in trying to "clone" a RCA - Clara's theremin is far more sophisticated, only slightly more complex by '30s standards, no more complex by todays standards - And has everything the RCA has and more.. In fact, if one was to try to clone the heater volume idea - well - youd need to be mad for one thing, but aside from that, it would be a real pain to implement with semiconductors!  ;-)

Thanks, Fred. You have confirmed what I was thinking.

I don't know of anyone that prefers the, as Clara put it, "molasses" response of the heater control scheme.

I'm sure the stock volume circuit does load quite a bit. Note that they used a 71 power triode as an oscillator. That bad boy should be good for at least a couple of watts in class B or C. I'm looking at one now, and the plate is massive.

I have a thought on Clara's.

A few years ago, I built a guitar for my bride out of vintage and reissue Fender parts. What should have taken a couple of days stretched out into weeks because this one was for her. I wanted every detail to be perfect.

I'm convinced that Lev poured his soul into Clara's for exactly the same reason. Everything about it is Lev's absolute best. Even Lucie's "September", the Marshall stack of theremins, is not as elegant.

Kitty's Stratocaster and Clara's Theremin are both labors of love.

The split tank inductor of the RCA oscillator is extremely clever, but even though there is an AC ground capacitor for the VCC side of the coil, it seems to me that, with one side connected to VCC and the other to ground, noise could easily couple into the tank via the power supply.

"noise could easily couple into the tank via the power supply" - Dewster

Yes - it does depend a bit on how one defines "easily".. But supply decoupling is one of my "hobby horses" - LOL.. There is no topology which is entirely immune from the problems of supply bourne noise or cross-talk - With a 10k from the supply to the decoupling capacitor, I think "cross talk" or coupling from active current sinks connected to this point is a more likely potential problem than noise from the supply.

Looking at the RCA schematic (Art Harrisons Rev 11) one sees that R14 (10k) is connected to the filtered 140V supply at one end, the other end feeds C15 (2uF) and the junction of R14+C15 feeds both the variable and reference oscillators.. The inductance of C15 will mean that some HF will pass from reference to variable oscillators - this could cause oscillator synchronisation - Even without the inductance, there would be SOME coupling.. Just how much, and how significant this coupling is - well - with all the unknown variables it would be extremely difficult to compute.

However - This coupling MAY be essential to the RCA's sound - We know that the oscillators do pull each other quite substantially, and that this changes the waveforms, particularly when the oscillator frequencies are close to each other at the bass end.. And I wonder if the values and perhaps even the type of capacitor used for C15 were deliberate - perhaps found by experimentation.

I do not actually see any deliberate oscillator coupling mechanism on the RCA (other than possibly their common connection via the R14/C15 junction), I have crudely analysed some waveforms extracted from recordings (I will sort these and hope to post them ASAP) and to me it looks like oscillator synchronisation plays a far greater role in the RCA sound than on any other theremin I have analysed - More particularly, the "pulling" seems to extend a lot higher into the audio spectrum than on other theremins - I am not sure about this - a lot of possible errors exist due to the nature of the samples I have - and what I am seeing may not be due to synchronisation.. But I dont think the coupling is likely to be soley the result of the RCA's Layout..

When "cloning" an RCA, one does not (I think) need to worry about replicating this coupling mechanism, if its important.. It should be possible to easily introduce a deliberate controllable (perhaps even a user control) level of oscillator coupling through one of the modern methods.. but if not, it should still be possible to apply the same coupling path and make this variable.

If isolation was required (and I think it probably is - certainly if one wants to make coupling adjustable), the simple way to do this would be by having a seperate R + C (R14 and C15) for each oscillator -

When designing using modern components, having low ESR / ESL capacitors fitted close to any active current sink or source, and ensuring these are large enough to provide or sink the required currents, is usually enough - my boards are covered with decoupling capacitors - every active device must have one. With HF circuits, there is usually enough inductance in the tracks to provide isolation of transient events - in cases where one can accomodate voltage drops, and needs more signal isolation, a series resistor can be fitted - in cases where voltage drops are not tollerable, an inductance can be fitted -

IMO, this is the area digital designers most often get wrong - they often forget that their board does not consist of ideal point-to-point connections - it is always a complex network of inductances and capacitances.. I have lost count of the number of times I have been called by a digital engineer who is baffled about why their 48MHz signal which is appearing on the output of some logic device, has vanished or dropped hugely in amplitude, or become a triangle wave, at the input to another logic device 200mm away ...

Expierienced "old school" digital designers dont have this problem - but most of those coming out of university do - even if theyve designed boards for years afterwards, they are often 'lucky' but get severely 'grounded' the first time they encounter a problem like this.

If supply decoupling is not designed into the system, the system may work - it may not work - but usually it doesnt work as well as it would, and is not as reliable or suited to manufacture, as it would be if decoupling (and power supply design and distribution generally) had been given a high priority.

Fred.

Clara's has just such a controlled coupling device in the form of a small variable cap between the oscillator plates. Because the shaft is hot with both RF and DC, player adjustment would be problematic, but there are many other ways to skin this cat.

"but there are many other ways to skin this cat." - w0ttm

Oh yes - there certainly are.. one method I have used is a transformer (common mode choke) with one coil capacitively coupled between oscillators, and the other connected to an H11F1 (Fet) - The fet shorts out the winding, which reduces the other windings inductance - control is remote by the current supplied to the LED.. change in the inductance changes the coupling..

I think with a simple oscillator configuration like we are playing with here, an H11F1 in series with, or shunting connections (via small C's and large R's) between the oscillators may be the easiest way - series connection probably being best.

One can also use varicap diodes in series or shunt mode, or other circuits like the tuning configuration on the EW.

The only little problem with variable sync is that this does impact a bit on linearity - not a problem if only minor adjustments are made, but it does limit the extent of the tonal changes one can introduce through this mechanism.

So I am not sure that a user control for a RCA / Clara clone is a good idea - a coiple of pieces of wire twisted together and used as a variable capacitor is usually fine (a few of my H1's had this fitted)

To get full control of synchronisation without any impact on linearity is possible, but a lot more complex.. and unrelated to the sort of circuits we are looking at here.

Fred.

I have just played with simulating the response of the Lev oscillator split tank inductance, in combination with antenna resonator (virtual inductance across one of the coupled tank inductances)

First, a disclaimer - LOL.. Simulating coupled inductances is complex and cannot be relied on - I think my model is working (certainly works in that the combined inductance is accurate when external inductances are not placed across any of the coupled inductors, and seems to give about the same results as I get when I physically connect an inductor across a winding of my actual transformer and measure the combined inductance with my test kit) - But I cannot be sure it is giving the correct results for these simulations..

What I am finding is that there is a huge difference in the effect of the antenna resonator across the split inductance compared to it being across the total inductance.. And if I change the ratio of top and bottom inductances (as in, wind fewer turns on the top, and more on the bottom) one extends the frequency range (which would translate to covering more octaves)..

I have not yet managed to compute the changes in linearity resulting from modifying the ratios.

I have also found that one can achieve the same results from a parallel LC circuit if one taps the tank inductor and connects the antenna resonant circuit to the tap .. Split the tank inductor (for example, if the tank inductor is 100uH, tap its winding in the middle - each half will be 25uH)

It takes too much time to prepare the images and documentation - I am on a "roll" and dont want to stop to do this right now - but I am saving each step, and will go back and gather these and post them as soon as I relax.. ;-)

Fred.

Update:

I have now verified the operation by substituting the simulated virtual inductance with the full series LC antenna circuit, and switched this across different points in the tank coil windings, and changed the ratios of upper to lower over almost the full inductance range (with the series configuration I am limited to about a 1:5 ratio in order to get ticker feedback).

It looks like the sensitivity (as in, frequency change for givan antenna capacitance change) is almost linearly related to the balance of the upper / lower inductances..

Not verified this yet - but it is looking like a 50/50 balance gives optimum linearity (or at least does for the values I have used, which are almost what the RCA has) and that as the lower windings inductance is increased with the upper reducing proportionally to maintain a constant total inductance, the linearity decreases as the range increases..... But I have NOT checked this in a spreadsheet yet, just going on patterns I think I recognise - And I often make mistakes doing this! - And this is all simulation - there may be demons in the detail - I will only be sure when I have built the circuit.. Main reason for doing the sims is to guide my design of the real test oscillator.

Update 2:

I have now verified my simulations with a quickly thrown together circuit on solderless breadboard - The simulations are getting quite deep - I didnt want to spend hours running them only to find they were useless.

Yes - Lev's split series tank resonator circuit was deliberate, it was not done that way just to cater for tube biasing or whatever.. it does improve linearity and increase stability, and I have no doubt in my mind anymore that this is the major "secret" of Lev's linearity.. IMO, it is one of the cleverest bits of circuit design I have ever seen - Possibly surpassing my former "top scorer" which was Bob Moog's VCF... It is just incredible to me that we have not noticed Lev's design before - that we have ALL been building theremin oscillators with the antenna resonator strapped across the whole tank resonator.

My simulations indicate that a parallel tank LC with tapped winding can behave in the same way - the position of the tank capacitance in the Lev oscillator is dictated by the split topology, but I think (unverified) one could have a  tapped inductance in (say) an EW oscillator and tailor its components a bit, and get improved stability and linearity..

As an aside - I really dont see any possibility anymore that there is a "blocking" action on the oscillator - On this, sadly, I think Uncle Howie is wrong. Also, the "twin peaks" I saw before proved to be simulation anomolies - I think that the sole secret of the Lev oscillator / EQ is its stability and linearity, and possibly its better ability to synchronise (as in, distort) over a larger frequency range before going into lock (I have not verified this) - nothing special about its output waveshapes except perhaps its extended sync range.