6
\$\begingroup\$

TL;DR I need to do is successfully transmit a wave across a single conductor between two battery powered circuits with no common ground. I already know how to create the wave and detect the wave when there is a common ground, but I don't know how to design a circuit that can detect the wave without the ground. My scope grounded to the receiver can detect the wave.

My friends and I are are building an open source fencing box to score matches (The kind of fencing with swords). Competitive fencing is scored electronically by using a sword rigged with a button and jackets made of conductive material. We want to improve on the existing designs available today which require expensive mechanical reels to maintain tension on the cables.

To make a long story short, we need to be able to communicate over one "wire" (the sword) between two Arduinos. Our plan was to use the Arduino PWM generators to generate tones that can then be detect by FFT of the analog inputs on the other side. We're using 7 total tones between the frequencies of 9khz and 18khz. We got all that working great. The problem is what happens when we go to batteries and no common ground. The PWM signal goes from 5V peak to peak to 1.2V peak to peak. I don't understand why this is happening. Any enlightenment here would be appreciated.

Can anybody recommend a circuit design for transmitting a wave over 1 conductor between two systems with no common ground? (And amplifying it on the other end if required) I'm thinking it's going to have to be similar to radio technology. Alternatively I can try to take what I've got and feed it into some sort of amplifier and see if I can run with that, but I hate doing that without understanding the forces at play, and so far none of the circuits I've tried have reacted to the signal.

Some additional Information:

We have a button on the sword. We can always detect that button press and use it to turn off/on the transmitter, so that if we cannot tell the difference between touch and not touch via attenuation of the wave we can just make it not transmit when there is no touch.

Improve this question
\$\endgroup\$
  • 5
    \$\begingroup\$ To make a long story short, sending a signal between two devices through a single conductor, with no other connections or comonalities at all is impossible. Voltage is fundamentally referenced to ground, without a ground, you cannot have a voltage measurement. The fact that your system works at all is due to the two systems coupling in some manner outside of the single wire connection. \$\endgroup\$ – Connor Wolf Feb 11 '12 at 19:55
  • 1
    \$\begingroup\$ @Fake Name So, radios work by magic? \$\endgroup\$ – Philip Feb 14 '12 at 21:38
  • 3
    \$\begingroup\$ @David HInkle - Oh god, this is in no way experts exchange. That site is an abomination. Also, you can turn on e--mail alerts, but they're off by default. Stack-exchange does not believe in spamming people! \$\endgroup\$ – Connor Wolf Feb 15 '12 at 22:36
  • 2
    \$\begingroup\$ @David: It is incorrect to assume that just because people are willing to help answer your question that they are therefore "interested in your open source project". You need to step back and realize the insignificant impact of your little project on the universe and the people here. Being willing to help and giving a crap to two very different things. \$\endgroup\$ – Olin Lathrop Feb 16 '12 at 14:22
  • 2
    \$\begingroup\$ Oooh Olin I like you! \$\endgroup\$ – AngryEE Feb 17 '12 at 17:36
4
+50
\$\begingroup\$

There will be a vague capacitive connection between any two separated bodies. The capacitance of this connection will vary with a lot of factors, including surface area, distance, humidity, phase of the Moon, the price of tea in China, etc. It's really not possible to predict even within an order of magnitude what the capacitance will be on any given day.

On the other hand, if two people P and Q are holding swords X and Y which are not touching, it's very likely that--because people's bodies have a much larger surface area than their swords, capacitor PQ will be large relative to PY and QX, which will be large relative to XY. For an AC voltage on PX to appear on QY, current would have to pass through small capacitor XY, which is very small, and some of that current would uselessly return through capacitor QX. On the other hand, if the swords touch, then the current would only have to go through the much larger capacitor PQ.

While one would have to do some experiments to determine the optimal coupling for signal generators and signal detection circuitry (my recommendation would probably be to use a single-frequency sine-wave generator in each suit, along with some LC-resonant tank circuits, so that each input circuit would be blind to its own generator, and maximally sensitive to the other). I would expect that with swords are touching, the strength of the coupled signal would be enormously higher than when they are not.

Note, btw, that higher frequencies will almost certainly work better than lower frequencies, though if you go too high (over about 100KHz I think) you'll start having to worry about being regarded as an "intentional radiator".

Improve this answer
\$\endgroup\$
  • \$\begingroup\$ Thanks for discussing some of the theory behind this situation. Do you have any recommendations for how to detect the sin wave? \$\endgroup\$ – David HInkle Feb 15 '12 at 20:42
  • \$\begingroup\$ You'll have to experiment. I would suggest that you design a circuit with a fairly narrow bandpass filter driving amplifiers of different gains, which in turn feed ADC channels (e.g. if you have a 12-bit ADC, have one channel feed the signal to the ADC with 5:1 gain, a second with 500:1, and a third with 50,000:1). Have the CPU blink lights or make noises to indicate which channels are seeing signals and which ones are saturated. A 12-bit ADC on a single channel would have a 4000:1 dynamic range, but with three channels you'd have a 40,000,000:1 range. \$\endgroup\$ – supercat Feb 15 '12 at 21:33
  • \$\begingroup\$ I really have no idea how strongly the signals would be picked up in various circumstances, and connecting any sort of lab apparatus to the device under test would totally corrupt the results. Therefore, having a large selectable dynamic range would maximize the likelihood of feeding something useful to the CPU. \$\endgroup\$ – supercat Feb 15 '12 at 21:36
  • \$\begingroup\$ @Philip: I'd like to hear what you come up with. Rereading your original question, one thing I'm not quite clear on--were you looking to detect touches between the two swords, between a sword and the opponent, or between a sword and "anything" that isn't the other sword? In the latter scenario, would you need to distinguish between touching "only" the other sword, versus touching something in addition to the other sword? \$\endgroup\$ – supercat Feb 21 '12 at 15:47
  • \$\begingroup\$ @supercat Once we get something functioning we'll put a writeup somewhere. I'll be sure to link it here. As for what we need to detect; yes, all of that. Touches are scored when your tip hits the opponent's metal vest. Off-targets on anything else are also announced. But if your tip hits their bell-guard (part of their sword), it has to be suppressed with no buzzer. We figured we could use different frequencies to determine what was hitting what when. This no common ground thing is a hurdle, but we'll get something together, even if we have to increase the cost and go zigbee or something. \$\endgroup\$ – Philip Feb 21 '12 at 16:29
3
\$\begingroup\$

At this frequency and distance, you're not going to get a self-propagating wave that takes no reference to receive. Like it or not, you will only be detecting a differential signal, and the signal strength and impedance of that signal will be dependent on the impedance of the whole loop, including the "ground" return connection.

However, this does not mean you have to have a specifically wired return connection. Each signal can use the body as a reference. The two bodies will have some connection thru the floor or other parts of the environment. The other pole of each signal would be on the sword at the sending side and on the jacket on the receiving side. You will get some coupling, although it will be quite high impedance. This means the receiver will pick up noise. You have to design the protocol carefully enough so that noise has a very low chance of being mistaken for the real signal.

I am actually surprised you are getting as much as 1.2 Vpp at the receiver without a deliberate return connection. This will vary as a function of the floor material. The more conductive, the higher the signal level. In any case, 1.2 Vpp must be well above the noise, so what exactly is the problem? Such a strong signal should be easy to detect, although I would make the receiver respond to much lower than that because it most likely will happen in different situations.

Improve this answer
\$\endgroup\$
  • \$\begingroup\$ Thank you for your comments. We know how to detect the wave when the grounds are tied. When the grounds aren't tied my oscilloscope (grounded to the receiver) can detect the wave but my circuitry can't. My electrical engineering understanding is not very strong. (I'm a programmer). My assumption is that although I have a voltage fluctuation that there is very little capacity to carry current so that whenever I put on any sort of load the voltage just goes to nothing. I don't know how to build a circuit that can detect and amplify the incoming wave form. \$\endgroup\$ – David HInkle Feb 15 '12 at 19:12
  • \$\begingroup\$ 1.2Vpp is what we're getting setting next to each other on the desk. We'll try it through the whole rig and test it tonight to see if it goes down. We're also not married to these frequencies, if you feel other frequencies are better we can move to them. \$\endgroup\$ – David HInkle Feb 15 '12 at 19:13
  • \$\begingroup\$ It's no big deal if we can detect the wave when the devices aren't actually touching. Because the swords have a button on them we can rig the sword to only transmit when the button is depressed if it's not possible to detect touching vs not touching based on attenuation. We'd still like the system to be as short range as possible, we don't want to be able to pick it up from several feet away on another box. \$\endgroup\$ – David HInkle Feb 15 '12 at 22:18
1
\$\begingroup\$

You are thinking about this wrong. You are thinking about this as a form of communication by contact. But since you don't have a closed loop, that's not what you want. There are at least two good ways you can do this -- the same two ways touch-activated lights work.

One way is to use what is essentially a capacitance meter. When you make contact, the capacitance will increase. Generally, you do this by making an oscillator and including the probe as part of the capacitative part of the oscillator. When the probe makes contact with something, the capacitance will increase, slowing down the oscillator. You detect this sudden drop in frequency. You can detect a specific thing by giving it a large amount of capacitance.

The other way is to use the other person basically as an antenna. Through the civilized world, we are all bathed in the 60Hz or 50Hz hum of our electrical wiring. A human body makes a decent antenna at these low frequencies. When your probe makes contact with something conducting to a human body, the level of 50Hz or 60Hz hum detected in the probe will increase. You detect this sudden increase in hum. (Note that this won't work in a forest or in places without electricity. And it can't distinguish one human-sized object from another.)

You could actually specifically induce a "hum" in the person and detect that hum. You would essentially be using the person as an ultra-short-range antenna. Stay under 9Khz and 1mW.

Improve this answer
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.