Question about the oscillators of the EM theremin

"Nah! I don’t give up! ;-)"  - FredM

The opening salvo in the great oscillator wars of '12!  ;-)

I pretty much agree with most of your points, any disagreement can likely be chalked up to intuition and personal biases.  Perhaps I should do an analog design to grapple with more of these issues directly.

Off for a week's vacation in the sweltering heat of NC!

Why would you want hundreds of volts of swing? I suppose the signal must not be too weak but why would such high voltages be useful?"  - edanielf

"High voltages will help swamp outside interfering noise sources (increase SNR).  Not sure if anyone has studied this, but it makes sense.  Using a linearizing inductor (like the EWS design) can jack this up even higher at the antenna" - Dewster

"Not sure if anyone has studied this".. I have not "studied" this, but I have expierienced the difference that the tank / antenna amplitudes make - In my first ventures, I did not see any relevance in the amplitude - I was used to short distance capacitive sensor design with plate amplitudes of 5V or even less, and they worked fine.. 

Based on this I developed an extremely complex multi-antenna instrument (Epsilon) and on the bench, with modules spread out, things seemed to be working.. As soon as the pieces were put together in an enclosure, all hell broke loose..

The impedence at the antenna needs to be high for the sensor to work over any reasonable distance - High impedence and low signal levels means that other signals can couple easily into the antenna, and even if this coupling only has a tiny effect on the waveform at the antenna, this can / will affect the cycle-by-cycle period of the oscillator - leading to frequency jitter and all sorts of other nasties.. My digital circuitry was coupling back to the antennas and screwing everything up.

There are some really strong radiated signals from common appliances - strong 30kHz (ish) signals radiated from "eco" light bulbs, some of which have strong harmonics up into the MHz, and other signals from other appliances which, although usually greatly attenuated by the time (distance) they get to the antenna will still cause oscillator jitter if the amplitude of the oscillator signal at the antenna is "low"..

I think it is a proportional thing - I think that all theremin oscillators will be affected to some extent by interference from external signals - But I think that the jitter may be proportional to the reletive amplitudes of the oscillator signal and the interfering signal - if this is the case, then the higher the oscillator amplitude on the antenna, the lower the jitter.

I dont think one needs to go to hundreds of volts to get better-than-acceptable immunity from interference - IME 100V P-P is better-than-acceptable , and 50V P-P is acceptable for most moderately noisy environments - Even 25V P-P is usually fine.. But go lower than this and I think you start to increase you chances of having problems. One shouldnt get silly and go too high - its a theremin you want, not a Tesla coil ;-) having lightening bolts jumping from your antenna probably aint a comfortable way to play!

Also, if you place a capacitor between the linearizing coil and the antenna, remember that this needs a voltage rating higher than the maximum voltage the antenna can attain.

There is one other aspect which I have observed - One gets a slight increase in sensing distance when the amplitude on the antenna is increased - I think this may be due to an increase in the space-charge arround the antenna having more 'density' due to the conductivity of the air around it - This is pure hypothesis on my part, but I wonder if the depth of this field changes the effective area (diameter)  of the antenna - probably only by < 1mm, but enough to cause drift when the properties (humidity and temperature particularly) of the air local to the antenna changes.. I have found that low voltage 'antennas' seem to have less drift due to environmental changes than those with really high voltage amplitudes on them.

Fred.

Dewster,

If you havent yet gone, have a good holiday!

I have tried to find your AFE schematic, but been unable to - If you are still here, and able to paste it up here, that would be great.. I try to challenge my biases as much as possible, and looking at your design might help me to do that!

"Perhaps I should do an analog design to grapple with more of these issues directly." - Dewster

Who knows - we might even change sides in this "argument" ;-)

Fred.

"I have tried to find your AFE schematic, but been unable to..."  - FredM

Page 2 of the Let's Design and Build a (mostly) Digital Theremin! thread.

Thanks for the well-wishes Fred!

Hi,

I would like to revive this this thread with one other question that one of you might be able to answer. I have been playing with this EM-oscillators design and I came up with a slightly alternative version that circumvents the use of the voltage divider R3 and R4. I thought to be smart and replace this with a part of the tank where by dividing the capacitor C1 into two capacitors in series. There is a lower signal between the two (see picture below) which can then be used for the feedback. However, the oscillator only works in theory :( . In qucs it gave a very nice result but on the bench it didn't want to oscillate. Now I am one of those people that can't quit until he knows exactly WHY it is not working. (Understandably, the building of this theremin is going at an extremely slow pace...) Could one of you possibly shed some light on this?

For simplicity I left out the capacitor between 12V and ground although I understand that it is essential for the final design.

best regards and thanks for all the help so far,

Daniel

Not really looked at what you are doing  - but a quick thing to check when a simulation runs but "real" circuit doesnt, is tolerances..

Capacitors can easily be +/- 20% or more - if C1 and C2 errors combine, the signal at T2 base may not be enough to maintain oscillation.. Try reducing C1 and increasing C2 on the simulation to the worst-case values and see what happens. One of the reasons perhaps that resistive dividers are used more than capacitive dividers is that it is far easier to get high tolerance resistors than it is to get high tolerance capacitors.

There is also the start-up conditions to think about - C2 will be discharged at start-up in the real circuit, but unless you set this initial condition in the simulation, the "real" could be different to the simulation.

Fred

One other thing - I see no tank capacitance other than the C1 loading (220 pF) - This (with the 470uH) is going to severely upset things when you get to adding the Eq inductance... You need a smaller inductance and a larger capacitance for the effect of the antenna linearizing to work..

My advice? Go back to the EM (unmodified) design!

Fred.

Hi Fred,

I will take your advice and go back to the unmodified (or only slightly modified) EM-design. However, I do want to understand why it isn't working. grrr ;) ... So I tried most of what you suggested. I also tried lower the value for C2 on the breadboard to increase the amount of feedback... to no avail. I put the startup voltages to 0 V in the simulations, which didn't change the theoretical result either. I measured the DC resistance of L1 (8 Ohm) and added it to the simulation but theoretically it still worked, even if I cranked it up to 20 Ohm just to be sure (I understand that the AC resistance goes up for coils at higher frequencies but the effect should still be rather low at 500kHz). I also ran the simulation for at least 0.1 second of simulated time to make sure that there is no large scale effect that dampens the oscillations. It still works in theory :) . So it is still a mystery to me why it doesn't work...

Coming to your remark about the need for a lower inductance and a higher capacitance. I agree, the value of 220pF is too low. I have read somewhere that the antenna has a difference in capacitance of about 40pF depending on wether the hand is close or not. Based on that I calculated with a small python script what the effective frequency range of the theremin would be if I choose different components for the tank. Maybe I missed something but it seems that the EM-Theremin doesn't have a very wide range. I also calculated the Q-factor but maybe it is not necessary to worry about that and maybe I should worry more about the linearizing. I thought the range of the theremin should at least be that of a guitar...

Hi Fred,

I ran a few more simulations. It does seem like there is some kind of (startup) problem in the simulations as well if I use more more proper values for the tank's components. But I still don't completely understand why it isn't working. In any case I'll continue with the original EM-design or something very close to that.

best,

Daniel.

" But I still don't completely understand why it isn't working" - Daniel

Oscillator start-up can be one of those tricky "unseens" - Not saying what follows is a full explanation - it is just a quick "off the top of my head" hypothesis which I have not explored with simulation or the like.. It may even be wrong! ;-)

With the original EM circuit the coupling C into the resistive divider forms a high-pass filter, which "kicks" the transistor and this ensures start-up.. Adding a capacitor in the divider acts as a low-pass filter, attenuating the transient - Effectively it is "noise" which ensures start-up even if supply is interrupted at a non ideal moment - a little bit of HF noise at the transistors base is enough to start the circuit oscillating (even thermal noise is enough) - A capacitor from base to ground will greatly attenuate any HF noise or power-up transients.

Usually it is simulations which fail when real circuits operate - Simulations tend to be "cleaner" and thermal noise from components is not usually included in the component model - also, start-up states need to be explicitly specified (if you try to simulate a 555 oscillator, you will probably find that the simulation will output steady DC levels unless you specify initial voltages on Tr and Th pins.. But a real circuit will always and unfailingly oscillate regardless of the initial voltages on these pins).. But sometimes, as you have seen, simulations "work" when a "real" circuit doesnt. Usually I have found this is due to component tolerances being sloppier on a real circuit than the simulation.

Fred.