Hello all,

I have been playing with robotics for a while. About a year ago I came across some DIY capacitive sensors. In particular I came across a digital Theremin circuit, which offered omnidirectional proximity sensing. Then a little while ago, someone else posted here about theremins, getting me thinking about this all over again.

I wanted to use the digital theremin circuits to make a cheap proximity sensor, but I want it to be highly directional.

My basic understanding of the circuit is that you create a frequency of pulses. On a parallel circuit you interface an antenna to a similar circuit running at the same frequency. The output is the difference between the frequencies. So if you used R/C to build your frequency networks, the capacitance of the antenna will have an affect on the frequency output from that leg, and an object that is close enough begins to affect the capacitance of the antenna. Supposedly this can measure the capacitive different down to the pF's.

All models of a theremin I have seen use an omni-directional antenna. Why don't they use an array, yaggi, or rhombic?

I guess my main question is: Does the directional electromagnetic sensitivity of an antenna array have the shape and/or pattern as it's directional capacitive sensitivity (or does it even HAVE a directional capacitive sensitivity)?


Thank you all very much, sorry for the weird question. This has been bugging me for a while.

 

Hello,

You could shield the directions you do not want to be detected.

You will need to tune the circuit, the capacity has changed due to the shield.

Greetings,
Bertus

 

Yeah, I guess this would dampen the inputs from the "Shielded" regions. But on a fundamental level I was hoping to find out if the directional affect of an antenna arrangement affected it's total capacitance in the same additive way. In particular, using an array, would you increase capacitive sensitivity along the same tx/rx sensitivity lobes?

Thank you for your reply.

 

Hello,

You can also take two of these theremin circuits.
If the tone from both circuits are the same the object is in front or back of the antennas.
(you will get some 8 shape sensetivity pattern).
The 8 pattern will be more narrow if the antennas are wider apart.

Greetings,
Bertus

 

That may not sense non-reactive things like walls. You could try a pair of LED's angled to have their beams overlap at a small distance, and avoid bright spots.

 

Thanks, I have been moving closer to the idea of using as many types of cheap sensors as possible. That way I could reap the benefits of them all.

I was just hoping to find something new that I could use to increase the sensitivity of the sensor and keep the costs down. The theremin circuit is super cheap and easy to build on my own. I was hoping to get more feedback on the directional capacitance of antennas though.

In the end, if this proves to be the case (directional capacitance), I could build antenna networks out of layered circuit boards. Either using rhombic or arrayed antennas, I'd be able to triangulate between neighboring arrays. This should yield 3D sensing along the plane of the networked sensor. I got my hands on an FPGA and I'm looking to throw it into something. I'm looking to make a "skin" of these printed out on stacked circuit boards and draw the base of the antennas to their simple theremin circuits. Then pass them all off to the FPGA.

 

A lot of theremins have two antennas -- pitch and volume. Typically the pitch antenna
is vertical a vertical line and the volume antenna is a horizontal loop.

The antenna is one plate of the capacitor your hand the other. The capacitance change is around 2pF. You get the maximum change when your hand is parallel to the antenna.

The vertical line is essentially omni-directional since it is not practical to approach from the top or bottom. The loop is directional since approaching from the top with your hand
parallel to the loop is essentially the only way to affect a large change in capacitance.

A picture of the the Moog Etherwave is at http://tinyurl.com/5wsoxr

(* jcl *)