the above is the link to the site where the above circuit is.
U1 could maybe use an output series current limiting resistor, say 22k or so at pin 4, to not overdrive the tank. An inductor in series with the antenna could improve linearity and reduce the possibility of static damage, as would a series resistor at the input of U1. I would experiment with where the antenna is connected (try the other side of the current limiting resistor). If you limit the drive, there are sine waves available at both oscillators, I think I'd try picking those off instead of the square wave scheme here.
There's no reason CMOS can't make a really nice Theremin (I don't understand why more aren't made that way) and this simple circuit doesn't need to be changed all that much to improve it quite a bit.
"U1 could maybe use an output series current limiting resistor" - Dewster.
I think that putting a small (say 100pF) capacitor between A1 and the circuit, would be better than a series resistor on U1's output.. A resistor will severely affect stability - The only condition I can see where excessive current could be drawn would be if A1 was shorted to ground - A capacitor on the antenna would prevent this DC current route.
This is one of the best 4069 based circuits I have seen.
"There's no reason CMOS can't make a really nice Theremin (I don't understand why more aren't made that way) "
Well, there are good reasons for not using a circuit like the above - One can get equal or better performance using single transistor oscillators, and these are far less prone to ESD destruction, and there is no cost advantage or disadvantage.
"I think that putting a small (say 100pF) capacitor between A1 and the circuit, would be better than a series resistor on U1's output.. A resistor will severely affect stability - The only condition I can see where excessive current could be drawn would be if A1 was shorted to ground - A capacitor on the antenna would prevent this DC current route." - FredM
When I was playing around with similar circuits, a current limiting resistor at the output of the inverter prevented tank overdrive, and gave a nice sine wave at the input of the inverter. Tank overdrive negatively impacts stability, no? I believe all LC Theremin oscillators limit tank current / voltage swing one way or another.
Even better might be a lower value resistor driving the tank to higher voltages, and a capacitive voltage divider at the inverter input. The use of a ceramic resonator in the reference oscillator is interesting.
"One can get equal or better performance using single transistor oscillators..."
Single transistor oscillators operate with offset thresholds, whereas these CMOS oscillators operate more uniformly about their input threshold point. So I would expect these CMOS oscillators to be less sensitive to supply voltage and temperature variation? = more stable? Also, Bob Moog uses differential pairs in the EWS.
I don't fully understand the function of T2 here.
Hmm.. Oscillators are a dark art! (LOL) - It is true that the EWS uses a transistor pair - but interesting that Bob used a single transistor oscillator on the E-Vox!
IME the stability of an LC oscillator is primarily down to stability of the L and C - thresholds do not have as substantial an effect as, for example, varying 'DC' offset currents through the L.
"So I would expect these CMOS oscillators to be less sensitive to supply voltage and temperature variation?"
The simplest CMOS gates are effectively just a couple of fets with quite loose tolerances - after all, they only need to work well for logic applications - and the thresholds do vary with both temperature and supply voltage variations - not so much as to be a bother in any logic application, and not enough to be a problem for anything but critical timing applications.. But they are no "wonder" component - they are just fets!
One advantage in using discreet transistors (BJT or FET) over using CMOS, is the packaging.. With discreets, you can plan the layout and insert whatever decoupling etc one wants, to limit interactions.. You cannot put a small RC filter on the supply to an individual inverter!
But hey - we all have our favorite ways of doing things, and I am realising that much creative electronics is more art than science .. Just one look at Bob's VCF design and one sees that - WTF! - Its genius, but so crude! Every time I see one I get goosebumps - and I wonder, was that "designed" or did it just "happen" ? - oh, I know it was designed - but I am not sure that its perfection was deliberately designed in.
"Even better might be a lower value resistor driving the tank to higher voltages, and a capacitive voltage divider at the inverter input. The use of a ceramic resonator in the reference oscillator is interesting."
Driving the tank to higher voltages is a good idea - one of the problems with CMOS oscillators (for theremins) is the low amplitude of the signal on the antenna - this, combined with the needed high Z, makes them extremely susceptable to noise interference. My biggest theremin disaster project was "epsilon" - I used CMOS oscillators on this (it had 4 antennas in a horizontal configuration, and computed hand movement above, giving X,Y and Z coordinates which drove VCO's VCF's and VCA's) - Things were working reasonably, but when I put the prototype stuff into the box, everything went crazy - The antennas were picking up signals radiated from the MCU board and LCD.. I had to redesign and rebuild the whole front-end, with higher voltages on the antennas, before it worked (it was unplayable, so abandoned..)
I have not played with CMOS LC oscillators much - I have used 455kHz resonators extensively - they make great reference oscillators!
"I don't fully understand the function of T2 here."
Pre-mixer waveshaping! (one of my passions) - The VFO and reference outputs are square waves, T2 removes upper harmonics from these, and I think it likely that the mixed waveform could be quite pleasant as a result.
ps - I am working on a spreadsheet which I hope will compute linearity (give distance - frequency graph, and tell octave spacing / number of octaves, etc) .. I am not doing any heavy maths - Im using a piecewise aproximation scheme.. But I have a bug in it that I cannot locate! I need help!
Anyone with Excel + Maths skills who is willing to look at what I am doing, and call me a fool .. ;-)
("It can't be done like this! You need to use eleventh dimentional quadratic hyperbolinated exponentiated fourtysecond order equations..")
Please email me - see my avatar. (particularly if you think you might be able to help - LOL)
Fred, as usual you are a fountain of knowledge! So sorry to hear about the problems you had with "epsilon" - though I'm always a bit shocked that anything as sensitive as even the garden variety Theremin works at all.
But I'm still confused about this design: T2 would shunt low frequencies to ground, correct? So how would that remove upper harmonics? I'd think it would be in series with D1 rather than going to ground.
I'm not an Excel expert but I've spent some time using it & so may be able to help - email away!
"But I'm still confused about this design: T2 would shunt low frequencies to ground, correct? So how would that remove upper harmonics?"- Dewster
This is a parallel LC not a series LC (and it is not just an inductor) , as such, in this configuration, it will attenuate frequencies on both sides of its resonance.. I think perhaps you never noticed the capacitor? - T2 is a 455kHz IFT with 680uH inductance betweens pins 1-3 and a 180pF capacitor inside the can between pins 1-3.
(remember that the 3rd harmonic is >1.3MHz, and so, by adjusting T2, one can apply some control over the tone.. A similar principle is used on the Moog 201 where two seperate tuned transformers give 'octave' and 'quint' enhancement )
(note - there is more filtering in the following cmos amplifier)
"I'm not an Excel expert but I've spent some time using it & so may be able to help - email away!"
I dont have your email address! - If you email me, I will then email you! ;-) ...
But I think I am probably trying to simplify something which cannot be simplified - I thought I was making an error in the arithmetic or referencing in excel - but I now think that I cannot solve the unknown without more elaborate maths 8-(.
"Fred, as usual you are a fountain of knowledge!"
LOL - Thanks Dewster - But I have one big advantage over many.. I have a pile of ancient data books from when Cmos and the like were new and engineers were exploring the potential uses.. These books gave schematics of what was inside the chips, and explored applications (critically) regarding their uses, often outside of their "intended" purpose.
Back in those days, when a 741 opamp was "state of the art" and was a fairly expensive component, there was a big attraction to having 6 multi-purpose (amplifiers / inverters / oscillators etc) on a cheap chip.. And I went for it!
But I did get my fingers burnt at times.. Fabrication methods changed, and with this, subtle changes to the devices which affected their "unintended" applications occurred - the devices still met all their published specifications, but their innards changed and the schematics for these innards sometimes no longer applied, with disasterous results for some designs which relied on features of the original parts. My worst expierience of this was with 4016 switches which I was using for VCF frequency control (PWM'ing these switches in the resistive path of the filters).. The specification improved - and my VCFs needed complete redesign (it was a crap design, actually - but I didn know that back then! ;-) .
These days I avoid using anything where the specification is not specifically suitable for a requirement - There are plenty of active analogue components at acceptable prices, and fets or BJTs where one needs something which is not on a chip or where the chip has disadvantages or is not cost effective..
But my perspective is always "tuned" to production - I dont want to design something "wonderful" for myself and not be able to put this into production if I want to. For someone without a "production perspective" almost any components or topologies can be used - One can fiddle till it works the way you want - and this way can be cheaper and quicker than my way.
going back to : "Even better might be a lower value resistor driving the tank to higher voltages, and a capacitive voltage divider at the inverter input. "
This has a disadvantage.. The capacitive divider would be in parrallell with the antenna capacitance, and therebye reduce sensitivity...
Oh - Perhaps disregard the above..
looking at the circuit again, I see a 33p extra capacitance added - So perhaps putting a small C between the T1 and input would increase amplitude on the antenna and increase sensitivity!
Fred, thank you very much for such a complete response! I understand now.
I saw that "switched resistor" filter trick back in a Don Lancaster cookbook IIRC. The old data books were quite a learning experience for me too.
And I definitely share your perspective when it comes to design. It makes me feel a bit queasy using CMOS 4000B in quasi-linear mode, but I looked around for voltage comparators and couldn't find anything that would function the same in a DIP package. It would be nice to have a guaranteed switching threshold.
Capacitive dividers are really neat, they chop down the voltage and unload at the same time, with no lag that a resistive divider tends to introduce. I can see why they are in so many one transistor oscillator topologies.
When I was still working IIRC we had lots of trouble with Philips HCT processor bus buffers (what they were actually designed for). When is a generic part not a generic part...
Sent you an email.