Pi Coil ringing

[tgm]

I have thought about a possible improvement in PI receiver
coil design but don't know if it is worth doing. The idea seems
fairly obvious. There are probably good reasons why
it is not being used. I'd appreciate some feedback.

Most PI coil designs use damping resistors to eliminate
receive coil ringing. This changes the received signal to
a shape that the PI detector can work with.

Damping coil response reduces its Q and reduces the
target signal. Can't the unwanted ringing be eliminated by
using another coil in a bridge arrangement (see attached image)?

The small coil would have the same inductance as the search coil.
It could be shielded and put on the main circuit board. It's
magnetic field would not interact with that of the search coil.

If the ringing of the small coil was matched that of the search
coil it should be able to cancel out the search coil ringing
at the input of the differential op amp. Any remaining ringing mismatch
could be adjusted out by large damping resistors across one or both coils.
Larger damping resistors would have much less of a detrimental
effect than the original damping resistors.

I would like some feedback on this idea. Has it been tried before?

[tgm]

Coil Drivers

The previous diagram was only intended to explain the concept.
A real circuit would use synchronously driven but separate coil drivers.
When the high current driver pulse was turned off, each coil would be in
a separate high impedance circuit. Each coil would then start ringing
independently of the other.

If all components are well matched the ringings should start to cancel each other out.
It might be too difficult to cancel out the ringing if there are many ring cycles in the waveform.
Damping resistors across each coil would used to reduce the ringing to
a few cycles (say 3 or less). This should make possible a finely tuned ringing cancellation
that will leave only a stronger than normal search coil target signal at the differential op amp.

If the ringing is well canceled out it might also be possible to start sampling at an earlier
point in the received signal.

My questions for the forum experts:

1) will a circuit of this type have the advantages I described?
2) If so is a circuit of this type be stable enough to be practical?

[Davor]

IMHO PI with IB Rx coil should work even better. Monocoil rocks just because it is just one coil with minimum quirks that go with it, ringing being one of the quirks.

Your solution with two FETs may be worth further experimentation.

[Tinkerer]

IMHO PI with IB Rx coil should work even better. Monocoil rocks just because it is just one coil with minimum quirks that go with it, ringing being one of the quirks.

Your solution with two FETs may be worth further experimentation.

We could also look at the coil ringing from a different angle:

2 ways for eliminating the ringing:

Take away power, by burning off the energy in the damping resistor. This is the traditional way.

Or, prevent the ringing by adding power at the end of the Flyback. This is the TEM way.

Tinkerer

[tgm]

Your feedback was valuable. I now can see why my idea won't work.
Induction balancing the RX coil with the TX coil should eliminate RX coil ringing and allow only the target signal to be received. TX coil ringing is still a problem.

You have to wait for the TX coil ringing to dampen out to a reasonable level before looking at the target signal. Ideally, if the TX pulse was a step function, target sampling could begin immediately after the step.

With that in mind, I can see why my proposal would NOT work.
Ideally you want to expose the target to a single strong step pulse.
My proposal exposes the target to multiple ringing cycles. This make things worse.
You would have to wait for the effects of ringing on the target to die down before sampling the received signal. The effects of ringing on the signal would make it too
weak to be usable. Looks like it's time to go back to the drawing board!

[Tinkerer]

Your feedback was valuable. I now can see why my idea won't work.
Induction balancing the RX coil with the TX coil should eliminate RX coil ringing and allow only the target signal to be received. TX coil ringing is still a problem.

You have to wait for the TX coil ringing to dampen out to a reasonable level before looking at the target signal. Ideally, if the TX pulse was a step function, target sampling could begin immediately after the step.

With that in mind, I can see why my proposal would NOT work.
Ideally you want to expose the target to a single strong step pulse.
My proposal exposes the target to multiple ringing cycles. This make things worse.
You would have to wait for the effects of ringing on the target to die down before sampling the received signal. The effects of ringing on the signal would make it too
weak to be usable. Looks like it's time to go back to the drawing board!

I think it may be one of the ZEN ideas: "If your opponent is stronger than you, use his own strength against him." Or something of the kind.

You can also use the ringing for your purpose. The ringing after the step function, can itself be a powerful target exciter. You just have to extract the target information in a different way.

Going back to a non-ringing system, with separate TX and RX coils, both coils have to be damped separately, but it is possible to use different damping methods.

In the TEM_IB-PI, I use the classical resistor damping for the RX coil and the TEM specific adding of a DC current to avoid the ringing in the TX.

I think the TEM TX, by the way, can be looked at as a typical step function.

Tinkerer

[tgm]

Looking at the oscilloscope pictures in "Tinkerer's IB-PI" post #2, it is evident that the received target signal is present despite the RX coil ringing.

1) If the coils of IB-PI are really balanced why is there ringing in the RX coil? The RX coil should be seeing little or no TX signal. Can I assume that there is no way to eliminate the cause of the RX coil ringing by better coil balancing and that there will always be some
residual RX coil ringing?

2) It looks like the residual RX ringing simply superimposes itself on the target signal and has no other effect to the target signal. If this is true then the idea of canceling RX coil ringing at the op amp (using another coil as previously described) may still have some merit.

[Tinkerer]

Looking at the oscilloscope pictures in "Tinkerer's IB-PI" post #2, it is evident that the received target signal is present despite the RX coil ringing.

1) If the coils of IB-PI are really balanced why is there ringing in the RX coil? The RX coil should be seeing little or no TX signal. Can I assume that there is no way to eliminate the cause of the RX coil ringing by better coil balancing and that there will always be some
residual RX coil ringing?

2) It looks like the residual RX ringing simply superimposes itself on the target signal and has no other effect to the target signal. If this is true then the idea of canceling RX coil ringing at the op amp (using another coil as previously described) may still have some merit.

Could you give me a link to the post #2? There are so many threads and posts #2.

The balancing of the coils does not change or eliminate any ringing.

Even with the best induction balance there is always some residual signal. This does not affect the target signal, but it affects the possibility of amplifying the RX signal.

The RX signal is always the sum of many different signals like:

The ground response.
The residual from the coil self response.
Various types of noise.
The residual of the TX pulse.
The switching noise.
The EF signal
And probably a few more of lesser importance. Out of this sum of signals, we have to isolate the target signal, which is often in the nano volt range.
This is the art of metal detectors.

The good thing is that there are many ways to do it. Out of the box thinking to explore new ways or to improve on old ways can lead to a breakthrough.

Tinkerer

[tgm]

The link to the tread is here: http://www.geotech1.com/forums/showthread.php?t=18910
The oscilloscope pictures are in the post 2 and post 4.

You said that "balancing of the coils does not change or eliminate any ringing."
Careful balancing of the coils should minimize any inductively coupled TX signal in the RX coil. Ideally, there should be little or no RX coil ringing. From your list of RX signal components I am guessing the "residual of the TX pulse" is causing the RX coil ringing. Can't this ringing be
eliminated by directly adding a very small out of phase TX signal at the RX coil?

This could minimize the need for RX coil damping. The pictures in post 2 and post 4 show that the target signal an undamped RX coil sees is more than twice as strong that seen by the damped RX coil.

[Tinkerer]

The link to the tread is here: http://www.geotech1.com/forums/showthread.php?t=18910
The oscilloscope pictures are in the post 2 and post 4.

You said that "balancing of the coils does not change or eliminate any ringing."
Careful balancing of the coils should minimize any inductively coupled TX signal in the RX coil. Ideally, there should be little or no RX coil ringing. From your list of RX signal components I am guessing the "residual of the TX pulse" is causing the RX coil ringing. Can't this ringing be
eliminated by directly adding a very small out of phase TX signal at the RX coil?

This could minimize the need for RX coil damping. The pictures in post 2 and post 4 show that the target signal an undamped RX coil sees is more than twice as strong that seen by the damped RX coil.

Excellent.
I really appreciate your help.
Once we have generated a target signal, it is crucial to keep the signal integrity. I posted these step by step posts and pictures in the hope somebody would offer help in solving the problems.

Thanks for the feedback.

So let's look at it in more detail.

The TX step function, generates a Flyback of several hundred volts. The RX coil picks this up because the TX and RX coils are transformer coupled.
To reduce this Flyback in the RX coil, we add a Bucking coil made of a few turns, about 5% of the inductance of the TX coil. The Bucking coil is in series with the TX coil, but 180 degrees apart, so it generates a counter Flyback that nulls the TX Flyback.

Attached is a simulation that shows the above. I have also attached the LTSpice file, so you can experiment with different settings.

The yellow trace shows the TX coil current.
The red trace shows the Bucking coil current.
The blue trace shows the sum of the TX coil and the Bu coil currents.

R5 is a bypass. Coils in series have normally the same current. However, if we have a slight unbalance, the bypass allows to compensate for that by making the current different in the 2 coils.

You see no oscillations on the simulation. This is because the simulation is not perfect. In fact, there are always some HF oscillations at the Mosfet switch ON and OFF. These are the oscillations you see on the RX signal after damping.

Reducing these oscillations would help increasing the sensitivity. Mind, the sensitivity is quite high already, but we want to improve it still more.

Tinkerer

[Tinkerer]

I just managed to reproduce the switching oscillations in the simulation.
So now we can investigate what produces them.
It looks like it is stray inductance, but I will investigate further.

Tinkerer

[Davor]

There is a patent of a detector that uses these oscillations as a cause for calling their solution a hybrid PI VLF

My strong belief is that PI detection (Rx) will benefit from introducing some kind of time dependent weighting function. A sinus is a possible choice. With such functions sampling delay is not that much of a limiting factor.

[Tinkerer]

There is a patent of a detector that uses these oscillations as a cause for calling their solution a hybrid PI VLF

My strong belief is that PI detection (Rx) will benefit from introducing some kind of time dependent weighting function. A sinus is a possible choice. With such functions sampling delay is not that much of a limiting factor.

I have no problems sampling at any time, but would be interested in what you mean with the above statement. Would you be so kind and explain in detail?

Tinkerer

[Davor]

It is fairly easy to understand. Remember a or mu law in telephony? There you have it. A weighting function.

Now, with PI you have a time dependent function which is very sensitive on timing, and a few quirks that prevent you from early sampling. In case you transform your nasty exponential decay into something more manageable, you are able to extract more information from it. E.g. if you consider end of PI pulse as a t=0 and your weighting function is y=a*t with some arbitrary factor a (linear ramp function), multiply this function with Rx signal, and your nasty PI response becomes less nasty.
Other option is an alternating chopping step function that hacks PI response in a series of short alternating pulses that have favourable properties regarding (electric) ground reference, say you may supply alternating pulses to a differential amplifier and integrate the signal with no worries about zero reference.
Or you may use some natural mechanism that provides weighting function by itself, say a LC tank, and detect a RF component obtained by exciting that tank with PI Rx. That solution is already patented as VLF PI hybrid, and the commercial solution was not that much successful, but there it is.
You may also play with LC resonant frequency from the above example and obtain something as a ramp function, thus combining some of the above mentioned weighting function candidates. Remember sin(x)=x for small x-es, and this approach will surely conserve the pulse energy.

To be completely honest, I have no idea what approach will yield a practical solution with favourable results. So far I played with some weighting functions and PI responses in LTspice for various targets, and obtained some interesting results. My goal was finding a directly observable discrimination solution and some results looked promising, but not that much practical. I tend to dismiss solutions that are complicated.

[ODM]

I believe I saw something similar in a popular mechanics archived issue from 1941 - it was used as a mine detector. The balancing coil of the measurement bridge was housed in the electronics box containing an oscillator and headset amplifier.

The oscillator powered the bridge, and adjustments were provided for that reference inductor and a rheostat for balancing the resistive parts of the measurement bridge. Curious contraption, but apparently it worked enough to save at least some lives.