Yet another "extending the range of a theremin" thread

Posted: 6/1/2017 10:18:52 PM
carebare47

Joined: 6/1/2017

Hello all,

First of all, apologies if this goes against forum rules because it technically isn't a theremin. Looking through this forums I can see that some of you guys have years of experience with building theremin systems and so if any help is available it would be greatly appreciated. 

I can see that there are already a few posts regarding getting theremins to work at ranges of a few meters, and that for most applications this isn't the best approach for the job at hand. As far as I can tell capacitive sensing could be a good solution for this project as linear audio response isn't needed but omni-directionality is. 

I'm currently working on a project in which the hope is to employ capacitive sensing to make a tree covered in LEDs responsive to congregating persons. The system is for a music festival in a couple of weeks. The idea is to strap an antenna to a tree covered in RGB LED strip lights whose colour or modulation frequency change depending on how many people are by the tree (or under a particular branch or something).

I built the system described here (http://www.strangeapparatus.com/Theremin.html) a couple of years ago for audio purposes. In a perfect world I would like to use the pitch variable and reference oscillators along with the mixer circuit in the strange apparatus theremin to attempt to detect people near the tree. The output frequency from the hetrodyne would be converted to a DC voltage and sampled by a micro-controller which will then generate the appropriate light patterns.

From what I've read on the forum something like this may be possible, although I don't know whether it's figure-out-able in a week. 

From parallel plate capacitor equations I can see that, as distance increases, larger plates are needed to maintain the same capacitance. I know that higher frequency hetrodyne circuits would provide a greater frequency change for the same capacitance and so this is a parameter that could be tuned for distance.

I have read in different places on these forums both that higher frequencies will be needed for greater distances, and that larger antennas don't respond as well at higher frequencies. Having very little experience with systems like this, I am unsure as to whether I should opt for a smaller antenna at higher frequencies or a larger antenna at lower frequencies (or neither or both :P).

I found the following from user dewster on these forums:

 

Make a cylindrical antenna, roughly the same height and diameter as a person, and put it in the center of the room.  Or use a plate to make it more directional.  Use higher frequencies than normal, say 1 MHz or higher, for the oscillators so as to increase the heterodyned sensitivity.  Don't pad or load the antenna with any extra oscillator capacitance, use the intrinsic self-capacitance of the antenna as the C in the LC tank.

 

[EDIT] The antenna should be raised a foot or two to minimize capacitance with the floor.

 

So my plan is to up the frequency to closer to 1MHz and try some larger antennas. 

I can see that dewster recommends a cylindrical antenna. Does anyone know how essential this is? Does anyone think the side panel from a computer case hanging under a branch could work? 

If anyone has any other ideas for omni-directional low resolution sensing of groups of people over a few meters then please feel free to suggest.

I'll try and get the system battery powered ASAP and see how it responds hanging from a tree in a local park, but any ball-park frequencies or suggestions for materials /shapes of antenna would be fantastic.

Thanks, 

Tom

Posted: 6/1/2017 11:05:54 PM
dewster

From: Northern NJ, USA

Joined: 2/17/2012

Hi Tom,

Having done more theoretical work since I wrote the above, I can currently say that the shape of the antenna won't influence directionality much.  You will need a large antenna to detect people sized objects over several meters.  The antenna mount should be insulated.  The antenna should be kept some distance from the tree and ground (perhaps a foot or more) and this distance should be constant, otherwise you will be detecting the movement between the antenna and the tree - I don't think hanging it from a branch will be anywhere near stable enough.

I believe RF transmission will limit the maximum Q of the LC resonance, and Q directly sets the "step-up" of the drive voltage swing vs the swing at the antenna.  So you want to stay away from extreme antenna aspect ratios and keep the operating frequency low - don't let the wavelength of the drive approach the physical dimension of the antenna.

For stability I'd recommend you use a single layer solenoid for the coil.

You might be able to drive the LC with a constant frequency (at resonance) and measure the phase shift between the drive and the antenna (sensed via a 1pF/100pF capacitive divider).

You can experiment with all of this pretty easily with a plate antenna, a coil, a function generator, and a two channel scope.  Get a feel for what you're up against in the analog domain before you get anywhere near constructing stuff and coding things up.

Also, hum filtering can dramatically lower the detected noise floor, giving you much better SNR.  If you don't want to play music on this you can go for really low detected bandwidth (<1Hz) and crush hum that way.

Posted: 6/1/2017 11:06:09 PM
oldtemecula

From: 60 Miles North of San Diego, CA

Joined: 10/1/2014

Tom you are very welcomed, TW could use some new conversations. That is a lot of writing you did but it is misguided. What you need is best done with ultra-sonic sensing. Use two or more, these boards have a spot on the board where a DC voltage rises and falls based on distance. The range is out to 8 feet.

Two weeks to figure it out? If you lived down the street I could help you out.

Christopher

Edit: He wants results in a week with 3 usable meters of sensing, that is a 27' diameter under a tree, I must be sniffing a different type of glue than everyone else?  If the tree is in a bucket above the ground then make the tree the pitch antenna, if not capacitively swamped it will still have about a two foot sensing range. Keeping it tuned in the open outside environment will be a nightmare. There is a big difference between theory and practice. Some people here giving advice have never demonstrated anything original that works.

Posted: 6/1/2017 11:43:31 PM
carebare47

Joined: 6/1/2017

Thank you both for the quick response.

 

Dewster, thanks for the great information, exactly what I was after. Hadn't considered movement of the antenna relative to the tree, I'll try and come up with something for that. 

How big an antenna are you thinking, and what kind of frequencies? The advice to stay away from extreme aspect ratios is appreciated, I was considering a scaffolding pole. A computer case panel at 1MHz shouldn't couple the wavelength too much, right?

If the Q of the LC is important, do you think it would be worth using heavier gauge wire to connect the antenna to the circuit to try and reduce the resistance?

Regarding the coil, are you referring to the L in the LC? 

I'll have a look into checking the phase shift between the antenna and the driver and see if it looks like that gives more usable signals. My first ever electronics project was a touch lamp that I believe did something similar (but with a latched binary output instead of an analogue one). I'll see if I can find the schematic. 

 

Oldtemecula, thanks for the response. The ultrasonic sensors I've used tend to have rather narrow cones in front of them so many would be required, and I think it would be harder to waterproof them against morning dew. If it was one person I was trying to detect and I needed to know exactly how far away they were then I think I'd probably end up using ultrasonics. I'm hoping that capacitive sensing will give me a vague approximation of the number of people under a part of the tree. Ultrasonics or PIR are the backup plan. 

 

EDIT: 

Dewster, I'm not quite sure what you mean regarding going for a low bandwidth to get rid of the hum. Are you saying that if I can design it so that the difference frequency from the hetrodyne has a range of only 1Hz then I'll be able to low-pass filter the input of whatever is measuring that range to get rid of mains hum?

Posted: 6/2/2017 12:31:26 AM
dewster

From: Northern NJ, USA

Joined: 2/17/2012

"How big an antenna are you thinking, and what kind of frequencies? The advice to stay away from extreme aspect ratios is appreciated, I was considering a scaffolding pole. A computer case panel at 1MHz shouldn't couple the wavelength too much, right?"

I'd aim for an antenna with roughly the surface area (or maybe the surface area of the 2D projection, or shadow) of the thing you want to detect.  1MHz with a computer case panel should be OK, just keep the wavelength in mind when scaling up the antenna dimensions.  For a human sized antenna you might want to drop to hundreds of kHz.  The increased intrinsic C of the panel will help lower the resonant frequency somewhat.  Just don't pad the antenna with any extra capacitance.  Adding 1pF to sense the phase should be entirely OK though.

"If the Q of the LC is important, do you think it would be worth using heavier gauge wire to connect the antenna to the circuit to try and reduce the resistance?"

The resistance of the connecting wire won't have much, if any, influence on Q.  The resistance of the coil wire will be higher and have more influence, and even that probably won't influence Q much.

"Regarding the coil, are you referring to the L in the LC?"

Yes.

"I'll have a look into checking the phase shift between the antenna and the driver and see if it looks like that gives more usable signals. My first ever electronics project was a touch lamp that I believe did something similar (but with a latched binary output instead of an analogue one). I'll see if I can find the schematic. "

Look here too.

"Dewster, I'm not quite sure what you mean regarding going for a low bandwidth to get rid of the hum. Are you saying that if I can design it so that the difference frequency from the hetrodyne has a range of only 1Hz then I'll be able to low-pass filter the input of whatever is measuring that range to get rid of mains hum?"

No, not filtering the range but the response time.  And if you go the phase difference route there won't be any heterodyning going on.  I'm talking about low pass filtering the detected signal to trade bandwidth for increased precision (and reduced hum).  You'll have a signal that goes all the way to DC coming out of it, and you can filter this down almost all the way to DC if you don't care about playing music on it.

Do play with this before you get down to brass tacks to see if it is feasible / what you can expect.  Hook a coil to a function generator, and the other side of the coil to a suspended plate.  Look at the generator drive signal with one channel of the scope (trigger to this), look at the antenna voltage through a 1pF caps with the other scope channel.  At resonance you should very clearly see a ton of phase change as you move your body around.

Posted: 6/2/2017 6:14:09 AM
Buggins

From: Theremin Motherland

Joined: 3/16/2017

Hello,

 

You can try phase shift sensor.

If audio is not needed, it could be optimal solution. May be simplified - two stage passive RC filter and Arduino 12 bit DAC should be enough IMHO. Single layer air core coil may be used.

This schematic requires reference frequency generator which can be tuned in steps ~1% or finer (setPwmFrequency, digitalWrite). Not sure if Arduino can do it. If it can't, I would recommend Teensy 3.X MCU instead.

 

You can try this schematic easy, on a breadboard. Only one 74HC86 IC is needed.

 

Best regards,

    Vadim

Posted: 6/2/2017 2:00:16 PM
rupertchappelle

From: earth

Joined: 5/8/2017

Chief problem is that at distance the changes to the pitch are minimal.

If people are under the tree, individual theremin units could be hung from the branches like fruit. I would suggest a hemisphere for the antenna and paint it to attract the players and connect it to your lights. Metal bowls would suffice or if you want large, a metal snow saucer. The rounded surfaces would invite people to touch them. 

You can also fashion them as alien plants on the ground.

But the range of play will be limited and it takes a lot of time to make a series of hanging theremin bowls.

So, you may want to go digital for a future attempt:

https://musictech.mit.edu/ims

http://music.arts.uci.edu/dobrian/motioncapture/NIME03DobrianBevilacqua.pdf

https://www.vicon.com/

https://www.google.com/#q=kinect+for+music

https://www.google.com/#q=body+controllers++for+music

 

 
For the Art Farm project I was involved with I thought a field of mushroom theremins would be cool for people to walk through and interact with, but I am not rich so a field of magical steel mushrooms remains just an idea. Probably not a very good one in practice.

Good luck.

Posted: 6/2/2017 2:46:45 PM
dewster

From: Northern NJ, USA

Joined: 2/17/2012

"Chief problem is that at distance the changes to the pitch are minimal."  - rupertchappelle

I think you're overgeneralizing your experience with conventional Theremins, which have smallish surface area.

If you read research papers on calculating capacitance, you'll see that most, if not all, of the equations are ratiometric (which can make the results rather difficult to apply in the real world).  So if you scale up the plate area, then the distance between them scales the same, as does the mutual capacitance.

The ways to extend the physical range of a Theremin are:

1. Increase the plate / hand area.

2. Increase the SNR of the detection.

3. Decrease the intrinsic C of the antenna.

(1) is easy, though non-traditional, and can impact linearity; (2) requires stability and filtering; (3) requires active shielding.

Posted: 6/4/2017 3:16:41 PM
carebare47

Joined: 6/1/2017

 

Oldtemeculaare you suggesting making the entire tree the antenna? I'd considered this but I'm not sure how to proceed. What kind of variables will alter the turnings outside over time? Humidity? Temperature? 

Dewster, I attempted the setup described by you, but didn't really get anything useable. Connected a function generator (at between 100k to 2MHz) to a 10mH inductor, and the other end of the inductor to the computer case panel.

I connected one scope channel to the fgen side of the inductor, and another probe to the panel via a 4.7pF capacitor (smallest I could find).

It was hard to measure due to massive amounts of 50Hz on everything, but as far as I could tell the phase shift varied by about 3 degrees depending on my proximity to the computer panel (from a few meters to a centimeter or so).I then connected the old theremin circuit (well, only the pitch variable, reference oscillators and mixer) built from here: http://www.strangeapparatus.com/Theremin.html to the computer case, and observed a very small change in output frequency. I then disconnected the panel and reconnected the original antenna (literally a normal sized cutlery fork) and observed a significantly larger change in output frequency for nearby movements (although still nothing for movements >50cm away). 

RupertChappelle, so far that's looking like the best option. I get a much bigger change in output frequency for my smaller antenna. Increasing the frequency of the oscillators should give me more range right? 

Larger antenna seem to produce much less change in output frequency. Does anyone know why this would be?  

Posted: 6/4/2017 7:47:17 PM
dewster

From: Northern NJ, USA

Joined: 2/17/2012

"It was hard to measure due to massive amounts of 50Hz on everything..."

The negative output of the function generator should be earth grounded, as should the negative input to the scope probes (scope probes are usually grounded to the AC outlet ground, which is an earth ground, but function generator outputs are often floating).  You should also be using a good quality RF choke here or an air-core hand wound single layer solenoid.  If you do these things, believe me, you'll know when you hit resonance!  Resonance can be quite narrow for a high Q coil, and it's probably easiest to locate using sine wave drive from the function generator.  A small resistor (~1meg or so) from the antenna scope probe to ground can help kill hum.

You must be logged in to post a reply. Please log in or register for a new account.