Buried wire detector

[Kiwi Bruv]

After 3 mnths of experimenting with my wire following robot project I've cracked the sensor. This is no small achievement for me. Now I need some more help.

I have an astable 555 driving an approx 100hz signal into a wire with about 80 mA @ 12V.

My inductor is picking this up, I've used this freq so I can actually hear it using an earphone.

I need some help with two things:

1. Audibly, the signal is only really discernable 50 mm from the wire, and maximum vol when beside the wire instead of on top of it. What is the best way of doubling or tripling this detection at least to 100mm ? More current? I don't want to do this because the wire loop will probably run off batteries eventually.

2. I know I need to clear up the signal and measure the strength somehow, the response below talks about bandpass filtering. Can someone explain what this means perhaps with an example?

Also the response below talks about wiring the wire as a multi turn coil - what does this mean and how does it help?

Very grateful, previously frustrated robot builder




"I'd use a buried loop, driven with a 1KHz to 5KHz signal as an invisible border.

[audioguru]

Hi Kiwi,
1) Your 100Hz is far too low for the inductive loop. Use 5KHz for much better efficiency of your "air-core transformer" and it should be very audible in your earphone unless you are very old and deaf.
2) Using multiple turns on the primary of your air-core transformer steps-up its voltage ratio by the same amount. Of course the greater length of the coil requires a thicker wire to maintain the same current.
3) Inductive loop systems eat batteries. Increase the efficiency with a multi-turn primary, multi-turn pickup and much more gain in the pickup's amp. Then you can reduce the loop's current. Maybe you can feed the coil's drive in bursts to reduce average current.
4) A bandpass or highpass filter in the pickup amp reduces mains hum and its harmonics. Then the amp can have more gain.

What circuit drives the very low impedance of the loop? Maybe it has a high loss with such a low impedance load. A car audio amp is probably suitable.  ;D

[Guest Alun]

Calculate the inductance, or better still measure it (I'm sorry I've not got the formula handy) then add a capacitor in paralell with the loop to form a tuned circuit and thus increase the power factor. A very high-Q tuned circuit could have a curent as high as >10A in the inductor while only consuming <1A from the supply. I agree with audioghuru 5KHz would be a lot better than 100Hz and yes, use lots or turns of thick wire for a high-Q.

[audioguru]

Hi Alun,
You have a good idea to tune the big coil. But I think its inductance would be very low since it is huge and has few turns. Also its Q would probaby be swamped by its driver.
Of course the pickup coil should be and can be tuned, then amplified with an amplifier with a high input impedance.  ;D

[Kiwi Bruv]

Many thanks guys, more questions below and

[Guest Alun]

Hi Alun,
You have a good idea to tune the big coil. But I think its inductance would be very low since it is huge and has few turns.

I can assure you the larger the radius the higher the inductance, but I'd agree the inductance will be very low and this would make it very hard to tune to 5KHz and still have a decent Q. I would recommend going up to 50KHz, or you could include an inductor wound on a ferrite ring or core (you could even buy these pre-wound)  in series to boost the inductance even more.

Here's a good web site that deals with AM antennas but it could also be adapted to this application, it even has a link to a JavaScript calculator.

Also its Q would probaby be swamped by its driver.

What do you mean? ???

I've made a simple CW experimental transmitter working at about 500KHz with an AM radio ferrite antenna with capacitor inseries and it worked very well. When the MOSFET was driven from a square wave the output was a pure sinewave. With a matched tuned ferite antenna I could power a neon lamp from when next to the transmitter (when it was powered from just 24V) and about 10mV or so when I was a few hundered meters away. The current consumption was just 2 amps and it ran cool so reasoned most of the energy was being converted to RF. I can't remember which made the most difference the transmiter being well tuned or the receiver or both.

Now I know this circuit is different as the transmitter is a lower frequency and it's Q is lower but I'm convinced that tuning it would increase the efficiency a lot.

Of course the pickup coil should be and can be tuned, then amplified with an amplifier with a high input impedance.  ;D

I agree but if the transmitter is tuned too the power output would be higher and the power consumption would also be lower.

[audioguru]

Hi Kiwi,
1) Change the frequency of the 555 by reducing the value of its timing cap by dividing its existing 100Hz value by 50. Then "tune" your pickup coil to 5KHz with a capacitor in parallel. Then tune your amp to 5KHz with a bandpass filter or use a highpass filter.
2) I am glad that you understand about using multiple turns on your primary coil.
3) Leave the duty-cycle at 50% for the primary coil's drive. I was thinking about saving battery power by transmitting the 5KHz in bursts with a duration of maybe only 100ms then a pause for maybe 900ms. Then it will use only 1/10th the battery power.
4) A bandpass filter in the amp circuit will tune it to 5KHz. Frequencies lower and higher than 5KHz will be attenuated. A highpass filter will pass 5KHz and above and attenuate frequencies lower than 5KHz. Search Google for typical circuits.  ;D

[Kiwi Bruv]

Many thanks. :)

The audio tone is much clearer now and even my ears can detect it at about 100mm from the wire.

Just about to research the band pass thing.

I don't know how to "tune" the pickup coil, how do I determine the capacitor I need? The coil is a 820 microH choke.

I can't describe the Eureka feeling I have about this. Having built something and knowing this was the most difficult part of the project its very satisfying. I wasn't sure I could do this at all. You're forum is brilliant and very tolerant of us novices.

[Guest Alun]

On the right is the formula for F you re-arrange the formula to make C the subject see the Left.

So for your circuit:
F = 5000
L = 820*10^-6

So:
C = 1/(820*10^-6*(2pi*5000)²) = 1.2356*10^-6 or 1.2356uF

The problem is that such a value would be electrolytic (altough cermic capacitors of this value do exist) the dialetric loss would be high and the tollerence would only be +/-20% so the coil would be very poorly tuned.


I still think it would be better to go for a much higher frequency if you want a it to be more sensitive. You can't hear 50KHz but it would make it very easy to tune both the transmitting and recieving loop, and there are ways to convert the 50KHz signal to an audiable frequency once it's recieved.

Now lets repeat the above calculation for C with 50KHz:

F = 50000
L = 820*10^-6

C = 1/(820*10^-6*(2pi*50000)²) = 12.35*10^-9 or 12.356nF

You could then use a 10nF and 2.2nF capacitor in paralell and with a 300pF variable AM radio tuning capacitor to tune the coil.

If you want optimise the ransmitter too (and I recommend you do) I would recommend using the same techniques for the reciever.

[Kiwi Bruv]

Many thanks for being so explicit and taking the time. I understand perfectly.

[Guest Alun]

I'm glad we've helped you I'm interested to see how well your circuit works could you please keep us updated.

[Kiwi Bruv]

Thanks. :)

I put the multi core cable on tonight and wired it to give 8 turns, it makes another big difference. Audibly we are out to about 15cm from the wire now and when on the wire it's screaming.

I want to complete this circuit functionality before optimising the signal and trying to reduce the primary loop power requirements some more.

I have one more question if you don't mind.

I now have the 5KHz signal nicely amplified. I believe the signal will be oscillating between the op amp power rails (nearly) and so shows about 6V average volts. It will show this when detecting or not.

For the input to my microcontroller I will put the signals from two sensors through a comparator but I know for this to work there must be a voltage difference for the comparator to work with.

How do I emulate my ears and brain and detect the presence or absence of the signal when my op amp will only show 6v avge irrespective of input signal?

As before mush obliged for pointing me in the right direction.

[audioguru]

Hi Kiwi,
That's great that your new coil and higher frequency are much more efficient!  ;D
The opamps for your two sensors should be amplifying the sensors' signals and not oscillating. The power in your transmitter coil must be reduced so that the receiver's opamps don't "clip" but have a linear output whose level can be measured.
Then each opamp output must be rectified and smoothed with a capacitor to provide a DC voltage that corresponds with level. A comparator works with the DC voltages produced.

[Kiwi Bruv]

Thanks guru.

I see how the rectifier would work and I can use that as an input directly to my microcontroller.

I have reduced the power supply voltage to 6v , but I can't get my multimeter to show a voltage change as the sensor approaches the wire - but the audio output says different?

I've attached my detector circuit. I have not built the meter part, my circuit finishes at the earphones connection.

[Guest Alun]

The reason your meter doesn't measure anything is because it's set for DC and it won't measure the 5kHz AC signal.

Try using your DMM on its AC setting.

A rectifier followed by a low pass filter sould give a DC signal that coresponds to the proximity of the sensor to with wire.

[audioguru]

Hi Kiwi,
Your LM351 opamp isn't made anymore, so full details about it are no longer available. Its datasheet shows it with a minimum of 10V for its supply voltage, but its text says that it will perform with reduced spec's with a supply voltage as low as 8V. I suggest using a more modern opamp that operates well at the exhausted voltage of 6V for a 9V battery.

The 150pF cap across the 2.2M feedback resistor reduces signals of 485Hz and above, and reduces 5KHz to about 1/10th. I suggest removing the cap. I also suggest changing the opamp circuit to be non-inverting, so you can accurately predict its gain and use its input resistor with the coupling cap as a highpass filter.

When you plugin your headphones, they will cutoff the rectifier transistor. The headphones jack needs its own coupling cap.  ;D

[Kiwi Bruv]

Thanks.

I'm now using an LF353 at 12V, just because I had it available.

I have 13mV to 1300 mV op amp output.

It also seems to work OK down to 9V. I doubled the transmitter power to 180mA and this made a big difference in stability. Output now starts at 9mV  at 200mm from the wire and goes up to 1700 mV on the wire very smoothly.

My microcontroler analogue input will resolve to 5mV so I think it's in a usable state now.

I think I have it in a position to use with my robot. I will build the rectifier circuit tonight and try your other suggestions.

Many thanks with all the help. This is my first analogue circuit and it's been a challenge! I've learned a lot and now it's working I can play with it to learn more.