NuFlector
From BEAM Robotics Wiki
Contents |
[edit] aka BEAM Capaciflector
An active field sensor that produces a variable frequency output signal in response to changes in it's sensing field caused by the presents of a detected object.
Click > HERE < to view a PowerPoint animation (created by Richard Caudle) of the sensing field produced by the Nu-Flector, and how the dielectric lens affects of field.
[edit] Development
Richard Caudle described the developement process of this BEAMish capaciflector circuit as follows:
(From a post (message #57426) to the BEAM email list dated January 25, 2008)
I started out looking into the Capaciflector at the behest of my Boss so we could make a machine to grade and remove knots and voids in floor boards. I talked to the inventor of the circuit several times and he told me that I knew more about the circuit than he did! I didn't like the response frequencies that I was getting so I went to another circuit.
The second generation circuit I used a 556 timer. That got me from 2kHz to 3kHz to 100kHz. This is when I started using a Master/evalS { master with inverted slave :) } setup. The output of the master circuit is 180 degrees out of phase from the slave output. I was using a 12" aluminum dish inside a 24" aluminum dish as an array. The biggest problem with this setup was that it was broadcasting noise on the radios [AM and FM] and the telephone system. After a flyby from a military helicoper, my boss decided to retire that version.
The third version came about from a converstion with the Capaciflector's inventor, John Vranish. I asked him if he tried anything other than OpAmp circuits and he told me that they worked and he was on a deadline, so he didn't. I went on to use the 74HC14 simply because it was "a triangle like the OpAmp". I drew up the circuit like I thought it should work and it did! Amazing! At that point I tuned the output to about 20MHz and could get anywhere from 1MHz to 5MHz variation in output depending on the size of the room and the object being sensed. The responce seemed to change when the size of the room changed. I could use it to sense when there were people in the office below mine! It will operate over a broad range of input voltages, but it seems the most stable at around 5-6 VDC. Once I hooked all six gates in the 74x14 together and got over 100MHz output, but the circuit burned up after a few minutes. Pop! and it was done...
I sent a schematic of the "Original" Richard Caudle's Capaciflector (although Wilf will tell you that a capaciflector it ain't!) as I built it back in 1998. The semi-circle is the driven parabolic "Shield" and the small circle is the "Sensor Element".
I've had some good luck with using double-sided PCBs as sensor/shield pairs. The substrate of the board helps increase sensitivity of the circuit. Also if you place something made from HDPE or UHMW PE in front of the sensor/shield array it "magnifies" the field. I called it a dielectric lens. I thought that I was coming up with a new term, but it's already been done! Rats! I tried using wire as a sensor (antenna) but had little success.
[edit] More Comments from Richard Caudle
- The 1 MHz that I mentioned was the change in frequency from an unloaded state to a loaded one not the total output. I don't have a 'scope or a freq. counter anymore so I can't be positive about the actual frequency output. (From BEAM message #7113)
- … use the actual driver circuit as the sensor plate. That way you cut down on spurious detections. (From Alt-Beam Message # 1119, Paragraph 3)
- The first thing that I used was a regular copper-clad board (two sided) to make the sensor and shield. Actually, I take that back...the first one that
I built was made of chewing gum wrappers with leads soldered onto them (great soldering practice!), and held together with scotch tape. The shield has to be a factor of seven larger than then sensor.
There are lots of applications that this can be used. A Stamp2 based robot that did some work on a few weeks ago used a capaciflector circuit as antennae. instead of a shield and sensor, I figured the wavelength and divided it harmonically down to 2.1 feet and wound bell wire around a LEGO (blasphemy?) flexible hose thingy and it worked! I could tell if an obsticle was to the right or to the left of the vehicle.
The geometry of the sensor/shield combo is critical to the performance of the circuit. Also you need to keep the leads to the sensor/shield as short as possible as they too will generate a field. (From Alt-Beam Message # 1135)
[edit] Wilf's Comments
Begin Quote (From Message #7083)
I'll go over some old ground but here are some comments:
This circuit looks like a capaciflector but is different in that this circuit uses gain in the shield driver. That is a big departure from the fundamental principle of the capaciflector driving the shield with a unity gain buffer with exactly the same signal as the triangular waveform on the sensor lead. Matching the signals nulls the normally dominant near stray capacitance to the shield and ground plane and makes the sensor much more sensitive to changes in far stray capacitances in front of the "lens". These can be anything from a detecting an approaching body to changes in dielectric constant of wood (ie moisture content). The circuit responds to the change in stray capacitance by changing frequency. At 1MHz, a counter read and reset at fixed intervals should give the desired frequency counter function. Using a LM2907 frequency to voltage converter (F/V) chip, or alternately the more sensitive error signal from the voltage controlled oscillator (VCO) of a phase lock loop (74HC4046) is a way to detect the frequency changes.
End Quote
After confiming the capacitor value (.22uf) Wilf posted the following:
Begin Quote (From Message #7086)
So the mystery deepens: the RC (1M x .22uf) time constant is way off for a standard Schmitt trigger oscillator. This means that the 74*14|HC14]] are operating in a spurious mode, probably dependent on output resistance and the "shield to sensor" capacitance. I assume that changing the 0.22uf capacitor to 0.1uf does not affect the frequency very much?
End Quote
Begin New Quote (From Message #7110)
I realize that my reply was a bit obscure. What I meant was: if Richard's circuit frequency was generated by the conventional Schmitt trigger oscillator (as the circuit suggests) then the RC time constant is way off compared to the 1MHz frequency he reported for this circuit. So it can't be operating in the conventional Schmitt trigger oscillator mode.
My best guess that it is the inverting shield driver is that causes the oscillation and the capacitance between the shield and sensor is the dominant C and the output resistance of the inverter sets the R. Each of these assumptions is easy to check ie by adding 50 ohms in series with the shield driver output and adding some capacitance between the shield and the sensor to see what effect this has on the frequency.
End Quote
