Low voltage tube mixer ?

Hello Dominique,

Because you bring up one of the most fascinating threads from the past I must respond. I am just a garage hacker with a little amateur radio background. WD6EDZ - Fred below (RIP) does a nice intellectual reasoning of how theremin sound color comes about, using a transformer. I also figured this out a few years earlier, this approach is on my PCB dated 2013, the only commercial board I ever had made. There are many reasons why a first theremin design will not work properly but she is a good teacher and will discipline you as you go along. She will humble most engineers, no.... all of them. 

I could give you some simple tips to aid in your journey but I do not want to spoil your fun... ok one, a 1N914 diode works fine for me mixing RF and detecting the audio signal, used with vacuum tube or transistor oscillators. I always found cooperative RF blending before the diode to be the real trick. Shaping the sound starts with the pitch oscillator right on to and thru the speakers.

In this older thread we are messaging I am avatar RS Theremin, today the theremin has seasoned and aged me, so now oldtemecula.

Good luck on your adventure.

Christopher

FredM said: Unlike you, I am not absolutely convinced that tubes add anything which cant be created more simply using modern components.. I have no interest in cumbersome electronics and HV supplies etc unless these are absolutely required for "good" sound.

My thinking is in the lines that IF tubes are "needed" then they will be needed probably for the mixer and (perhaps) output stages only - the mixer being by far the most likely important role. I cannot see any sonic benefit in replacing oscillators with tube ones, but there might be a thermal / stability benefit.

I also think (and Thierry is the person who got me to look more closely at this) it likely that to get the "classic" mixer sound, the mixer needs to drive a transformer, and the characteristics of this transformer (and related resonant components) will be critical in 'coloring' the tone.

I mostly agree with FredM. It is no sonic difference between a transistor class A amplifier and a tube class A amplifier because they will produce the same kind of distortion (the same harmonic content). You can also make a class b output amplifier with transistors and an output transformer; it will sound pretty much the same than a valve class b amplifier.

For the mixer, it is plenty of choices from passive to active ones. And yes, you can use an heptode to make a combined mixer-volume control, but I am sure you will get a better end result with one tube for the mixer and one for the volume control. For now, I begin with something very simple, a mixer with 2 capacitors in serie between the output of the oscillators. Their common point is coupled to the grid a the first audio preamplifier, its output go to the volume control. A limitation of that topology is than the 2 oscillators are synchronizing on each others when their frequency difference go lower than 20 to 30 Hz. I think than to add buffer stages between the oscillators and the mixing caps can eliminate that limitation and allow to get an audio range that start from 0 Hz. But I didn't experimented that in practice, maybe later, or more more later, I have other priorities for now. I begin to have something that work well, which imply I will soon begin to experiment with wave shaping. I will also do some measurement on how the antennas are reacting to the hand movements. But that's another subject and I must first finish to finalize what is already working (the hardware is still a little bit messy).

And don't be ashamed by your radio amateur background. Your community get the worst frequency bands regarding wave propagation and you succeeded to use them very well. Without you, the submarines in driving would still be unable to communicate. 15 years ago, I made my own TV antenna, an Yagi/Uda array of loops. It was not wideband but was outperforming all commercial TV antennas regarding signal quality and gain.

Hello Dominique,

Off topic:  So you're an RF person?  Do you know how to calculate the RF emission (power) of a 0.5m long x 10mm diameter Theremin antenna swinging 200Vp-p @ ~1MHz?  I tried to figure this out but hams speak a different EE dialect of impedance mismatch, SWR, etc. 

For the case of a 3D antenna I would think a sphere antenna would give the highest intrinsic capacitance for the lowest RF emission, and for the 2D case I believe this would be a circular plate antenna - with both driven from the center (I think the drive point is important?).  

If a person is using an air-core inductor vs. a ferrite core inductor, do you think there would be significant RF emitted from the inductor?  I was told this was true (long ago by member livio).

I ask the above because the LC energy lost to RF places an upper limit on LC tank resonance Q, and there is always radiated compliance to worry about (no Theremin builders / manufacturers seem to address this - though there is a compliance sticker on the bottom of the Theremini - go figure!). 

And it's just one of those things I'd like nailed down in my understanding of this stuff.  If you could just point me to an equation or two that would work too.  

Hi dewster:

From "Modern Antenna Handbook - Balanis, Constantine A. (2008).pdf":

At 1MHz, h/lambda=0.5/(300/1)=0.5/300=0.00166666. And a/lambda=.01/(300/1)=0.00003333.
After that the calculation is quite complicated. The Balanis explain it and show a curve where we can find Xa (reactance) et Ra (feed point resistance) at the feed point. For these values of h/lambda and a/lambda, we are out of that curve, but it show than Ra=0 and Xa can be extrapolated to something like 750 ohms, but it doesn't matter.

Now, the antenna radiation efficiency is Ra/(Ra+Rohmic). As Ra=0, it will always be 0. Which imply such an antenna will radiate no RF power. It is just too short.

If you want, I can send you privately a digital copy of that book. It is very complete and of a very high level, It also include both the theory and easy to read curves when the math becomes too complicated to be useful in practice. Definitely a must if you are making antennas.

Dominique, thank you very much for the pointer to that book!  I'm reading chapter 10 "Small and Fractal Antennas".  The discussion on small ka highly resonant antennas seem quite germane to Theremins (frequency detuning from nearby objects, etc.).   I believe even a tiny RF power loss will dominate other losses in a well-designed and constructed Theremin oscillator + antenna.

Hoping I can get some kind of number that makes sense from it.  In simulation I use a 470k to 1Meg resistor from antenna to ground in order to make the voltage swing come out closer to what I see with the actual circuit.

About the heater supply. For an audio device, it can really help to have a DC heater supply in order to leverage down the hum, that especially with high gain amplifier. For the current limitation, I would make it to work only at startup. An easy way to archive it with a LM317 is to make the voltage to go up slowly. But I would use a voltage regulation because in case of a heater breakage, the remaining heaters will see their nominal voltage when with a current limitation, the voltage will increase and the remaining heaters will start to over heat. The exception being a serie heater supply, in that case you must implement some kind of current limitation, otherwise the startup current will just be too high. In old TV sets, it was done with a PTC.

With an AC heater supply, the datasheet gives us the best topology in regard to hum. As exemple for the philips ECC82, it said that with Vf applied to pins 4+5 and 9, and the center tap of the heater transformer connected to the earth, the more favorable triode section with regard to hum is the section connected to pins 6, 7 and 8. 

You can also polarize the heater toward the cathode. The space charge is very near the cathode, and if it remain concentrated near the cathode, this result into a more concentrated flux of electrons and less noise. It is very easy to do. You must have a DC supply of a few volts, a 9 V battery will be fine. Instead of connecting the middle point of the heater to the ground, you connect it to that DC voltage when observing the output noise. It must be positive in order to attract the space charge. The exact DC voltage will be the point where you get the less noise. As the first amplifier stage is the main contributor to the output noise, this setting must be done with a low noise source like a bobined resistor connected at its entry. On some power tubes, you can see the influence of that voltage on the blue light inside the tube (when the light is turned off), it become more focused.

For minimum output noise, it can be interesting to polarize the input tube near the Lejay point. It is the point where the current coming into the command grid from the external circuit (normally 0 in class A) will be equal to the current coming out of the grid from the space charge, resulting into a sum of zero. This will give the minimum input noise. To find that point, the best is to measure it with a high Rg resistance, something between 10 to 20 Mohms. A voltmeter of similar input impedance can be used to replace Rg. You can replace Rk with a potentiometer and vary it. When the voltage on the voltmeter is equal to 0, this imply it is no Ig. The tension on Rk is the bias voltage for that point.

In a theremin, the first audio tube is the mixer because it "create" the audio. It is why it is so important, as was pointed out in that thread, and as my current work in progress just showed it to me. I just found that: In search of the ideal mixer. The experimentation of that guy is very interesting because he get better results with no anode supply than with an anode supply. Which imply my precedent comment about low anode voltages can be wrong in more cases than stated and must not be taken like a dogma. Does anyone experimented such a mixer topology?

Dominique,

That is a very odd mixer topology, using an accumulation of electrons on the grid (and the resulting grid current) to generate the output signal.

I have never wired such a beast, but from a theoretical standpoint it would certainly work without a B+ supply because as far as the electrons are concerned there still is one - it's just AC (IF+RF) without a DC offset. When the signal swings positive the plate pulls more electrons away from the grid. When it swings negative it pulls fewer electrons).

Removing the positive DC offset from the plates would make that effect more pronounced & give you a stronger signal at the grid (since the grid is the only positively-charged element left in the tube it will attract all the electrons, grid current will flow, and the grid voltage will drop proportional to R1/R2 & Ig).
Care must be taken not to exceed the grid current limit for the tube in use, lest the grid wires overheat (sag, melt, or otherwise self-destruct), but at low voltages with big resistors duty cycles <=50% that probably shouldn't be an issue: I've made mention of my 12V bench rig on another thread. That runs a 12AX7 with 12V on the plate and +3V to +5V on the grids. Grid current is constantly flowing and that tube has yet to suffer any ill effects even after being left on overnight because I forgot about it.

Re: Heater wiring, empirically from having built WAY TOO MANY guitar amps I wouldn't worry too much about AC heater hum in a theremin unless you're being really sloppy with your wiring: Ground the heater center tap, make sure the audio tubes aren't first in the chain (so the current flow near them is minimized), and make sure the heater lines are tightly twisted. (This appears to be more care than the RCA/Lev-Built instruments show, feeding their 2.5V heaters through the same cable bundle as the final audio signal going back to the power amplifier tubes).
You could go with a voltage-regulated DC heater supply but if you're a cheap-and-lazy sod like me you'll probably say screw it and run the heaters right off the filament transformer.

We're not talking about crazy high inter-stage gain in a theremin (whatever the gain of the mixer tube is, which you could tune to unity if you want, and whatever the gain of the volume amplifier tube is) so there isn't much chance for the hum to get amplified to audible levels, and 60Hz heater hum in the power stage usually isn't an issue and if you use a push-pull topology as in the Rockmore or Rosen theremin amplifiers it should be largely self-cancelling.

I'll find out in a few weeks if that theory all holds true with a theremin built from modern tubes - parts for my build should be arriving on Thursday

For me, the best analogy to understand that mixer is to see the triodes as 2 vacuum diodes, one of them controlling the other. With a vacuum diode, the heater will constantly provides electrons to the space charge. The space charge is negative toward the cathode, which imply you will get a current between cathode and anode even when Uac=0. That current is into the micro ampère range, but can be exploited with high impedance circuits like detection or AGC. Here we have an AC swing on both the anode and the grid. During the positive alternances Ia will flow but will also be controlled by Ug. Also, when Ug is positive, the grid will act mostly like a diode toward the cathode.

What surprize me in this mixer is that the local oscillator input is on the grid when typically its voltage amplitude is at least 10 dB over the RF input signal into such a radio mixer. This force the use of a negative bias, which will result into a loss of gain and sensibility. Maybe this is the price to pay to get a low distortion with that mixer.