New Concentric Co-planar Coil with Strong MF.

[Davor]

... which is not very useful
BTW, Vin is usually measured at 100 ohm, not 0 ohm.
There is a nice application note explaining the subject of input impedance and choice of opamp: http://cds.linear.com/docs/Design%20Note/DN355f.pdf

[Carl-NC]

It is the input impedance above which the sole contributor of noise is resistor noise. Below that the dominant source of noise is opamp.
In case your design calls for input impedance in neighborhood of 9k, NE5534 gives best bang per buck. Spending money on LT1115 will benefit you nothing, it will only lighten your purse.

Opamp noise consists of voltage noise and current noise. If the input resistor is zero, then all you get is the opamp voltage noise. As the input resistor increases, at some point the 4kT*R noise overtakes the Vn noise. That exact point is when R = (Vn^2)/4kT.

If the input resistor = Vn/In, then the resistor has 4kTR thermal noise, but the opamp current noise through the resistor also contributes. Which one is larger depends on Vn and In, there is nothing magical about Vn/In.

As the input resistor increases, at some point the In*R noise overtakes the 4kTR noise. That exact point is when R = 4kT/(In^2). Of course, the Vn contribution remains constant.

So we have:

Vn dominates for R=0 up to (Vn^2)/4kT
4kTR dominates* for R=(Vn^2)/4kT up to 4kT/(In^2)
In dominates for R=4kT/(In^2) or larger

So the bottom line is, the optimum value of R is zero.
-----------------

*Depending on the opamp, 4kTR might never dominate.

[Davor]

Never heard this before... "optimum" for what?

It is the input impedance above which the sole contributor of noise is resistor noise. Below that the dominant source of noise is opamp.
In case your design calls for input impedance in neighborhood of 9k, NE5534 gives best bang per buck. Spending money on LT1115 will benefit you nothing, it will only lighten your purse.

[ivconic]

Never heard this before... "optimum" for what?

That's because you are one ... trial&error "ee" and nothing else!!!


P.S.
Don't give me evil eyes and don't ban me! Ban Aziz and Davor! They are guilty for everything!

[Carl-NC]

Optimum input impedance of an opamp is calculated from voltage and current noise (Vn and In -> Ropt=Vn/In).

Never heard this before... "optimum" for what?

[Davor]

Could you give some more detail there? Not sure what your point is.

Both parallel or series resonance are the same phenomenon, but the way you take a signal off it makes it series or parallel. In series you see a very low impedance, ideally only parasitic resistances of the tank components, and you are basically measuring current that passes through. Equivalent circuit of an ampere meter is a short circuit. Simultaneously you may measure voltage across a coil and from that point of view it is a parallel resonance tank. In both instances a coil has some voltage across and some current going through. As the voltage is a function of magnetic flux, you are better off if you allow current to flow - thus maximizing power transfer. To ensure current flow you may either connect a capacitor to a coil and achieve resonance, or you may connect it to a low impedance preamp.

Because of the phase problems in resonance, I think the way around it is using a low impedance preamp. If you go too low you'll have some phase shift, but uniformly so, and not a step function as in resonance.

High side of low impedance is also a low noise. Optimum input impedance of an opamp is calculated from voltage and current noise (Vn and In -> Ropt=Vn/In). With NE5534 you get Ropt at ~9kohm, OPA134pa would have Ropt ~ 1800ohm, while with LT1115 you get Ropt at ~750ohm. Hence, LT1115 is better for low Z operation, and NE5534 is best for quasi resonant input. Option B would be using a common base or gate preamp, or some impedance transformer -just like pre-pre for MC phono.

Check out Hiraga pre-pre for ideas on low input impedance
http://www.bonavolta.ch/hobby/en/audio/prepre.htm

You may play with LTspice simulation above for all cases mentioned here. Try terminating a non-resonant branch with, say, 100 ohm instead of 4k7 and see what happens. Your criteria are voltage and phase. As the coil voltage is a function of flux, unless you go with input impedance below the coil reactance your coil voltage will be the same.

[simonbaker]

As a parallel resonance seen from a coil is in fact a series resonance with low impedance, the way around the troubles with resonance would be connecting a Rx coil to a low impedance pre-amp. You may play with the LTspice design here by lowering the 4k7 resistor (preamp input impedance stand-in), and see what happens to phase. Instant conclusions are that low impedance affects phase response, but in a mild way, and the voltage response remains mostly untouched.
To make a long story short: you may benefit from resonance pickup - without resonance - by connecting a Rx coil to a low impedance preamp.

Could you give some more detail there? Not sure what your point is.

Part of the appeal of an "on-resonant" design is the high gain of a high-Q RX resonant tank (using low resistance wire), which would seem to offer hope of improved signal-to-noise. However, I haven't analyzed far enough to know if such gains would evaporate due to some other factors.

Point with Tx coil being a part of an oscillator is not that much of a problem because whatever effect is dragging Tx frequency up or down, the same effect moves Rx resonance in the same fashion.

Good observation -- in fact, that might be a compelling reason to not use a frequency-stable oscillator, since the RX resonant frequency would then drift around relative to the TX freq!

-SB

[Davor]

As a parallel resonance seen from a coil is in fact a series resonance with low impedance, the way around the troubles with resonance would be connecting a Rx coil to a low impedance pre-amp. You may play with the LTspice design here by lowering the 4k7 resistor (preamp input impedance stand-in), and see what happens to phase. Instant conclusions are that low impedance affects phase response, but in a mild way, and the voltage response remains mostly untouched.
To make a long story short: you may benefit from resonance pickup - without resonance - by connecting a Rx coil to a low impedance preamp.

Point with Tx coil being a part of an oscillator is not that much of a problem because whatever effect is dragging Tx frequency up or down, the same effect moves Rx resonance in the same fashion.

[simonbaker]

My thoughts went to that direction as well. There are too much "wet grass" etc. effects called upon with a common denominator: E-field coupling. I'm almost certain my Rx coils will be bifilar with center tap to the signal ground, and Tx will be fed via separate pair.

Now regarding the IB coil systems. I made a LTspice simulation of a DD coil system, with values according to my gut feeling. I tinkered a bit with coupling to achieve some 120dB Tx suppression, and copied everything to show situation with resonant Rx tank, and a non-resonant Rx coil. This arrangement is as such for future experiments purposes - I did not put any target there yet.
Even here It is obvious that the tank resonance screws up phase so badly that any discrimination at resonance is doubtful. Also it is obvious that below resonance there is not much going on that would be any different from a non-resonant case.

So, have fun and play with this. This arrangement will be easy to adapt for concentric coils as well, and of course with more life-like values.

Yes, a number of people have gone through that exercise, but always good to have another sim to play with, thanks.

It does illustrate why Tesoro (and others) probably go with off-resonance designs -- any little component parameter deviation (manufacturing tolerances) would create a hefty phase bias if the TX frequency was at RX tank resonance.

It would be interesting to know how much heat effects those component parameters during typical MD operation. The reason is that I am interested in trying an "on-resonant" design some day, which would require dealing with the phase bias. Once you know the phase bias, discrimination could be performed easily, unless thermal effects cause significant movement of that bias during typical operation.

You would probably need to use a different kind of oscillator also, since ground composition probably significantly affects the TX oscillator frequency in the current TGSL design where the search coil is part of the oscillator. I would favor perhaps a relaxation oscillator (or crystal type) driving a resonant TX tank, although I'm sure that has its own challenges such as amplitude fluctuations.

I would probably tackle an on-resonant design with the help of an on-board microprocessor to help manage the calibration issues.

Just one of those pipe-dreams on my list...

-SB

[Tinkerer]

Hi Tinkerer:

My experiments (TGSL Experiments thread) seem to indicate that residual voltage is due to some non-magnetic coupling (capacitive?) in the wires of the coil and/or cable system. I got a much deeper null with a special Belden cable with shielded pairs than with a USB cable, for example. I do not know if that translated into detection superiority (couldn't test that). Also, both nulls happened to be well within the headroom of the pre-amp.

Magnetic coupling, in theory, should be able to be canceled quite well by shifting our coils I think. So maybe a little circuitry to cancel some of the "capacitive" coupling could be employed if needed to accomodate a coil whose residual null can't be knocked down enough by shifting the coils.

-SB

I agree, there are many factors that add up to generate the residual. Cable, shielding, inter-wire capacitance, are some of the factors. Add 2" of cable on the coil and be amazed how much residual that can produce.

Nulling the coils, means to obtain the minimum residual, but there will always be a residual.

Like everything else in metal detector design, it is a matter of compromises. We add here, only to loose there.
In case of doubt, we should fall back to the KISS principle.

Tinkerer

[Davor]

My thoughts went to that direction as well. There are too much "wet grass" etc. effects called upon with a common denominator: E-field coupling. I'm almost certain my Rx coils will be bifilar with center tap to the signal ground, and Tx will be fed via separate pair.

Now regarding the IB coil systems. I made a LTspice simulation of a DD coil system, with values according to my gut feeling. I tinkered a bit with coupling to achieve some 120dB Tx suppression, and copied everything to show situation with resonant Rx tank, and a non-resonant Rx coil. This arrangement is as such for future experiments purposes - I did not put any target there yet.
Even here It is obvious that the tank resonance screws up phase so badly that any discrimination at resonance is doubtful. Also it is obvious that below resonance there is not much going on that would be any different from a non-resonant case.

So, have fun and play with this. This arrangement will be easy to adapt for concentric coils as well, and of course with more life-like values.

[simonbaker]

Relative larger RX coils give more residual voltage. More residual means less head room for the pre-amp. Less headroom means less possible gain.

The best gain is at the pre-amp.

Tinkerer

Hi Tinkerer:

My experiments (TGSL Experiments thread) seem to indicate that residual voltage is due to some non-magnetic coupling (capacitive?) in the wires of the coil and/or cable system. I got a much deeper null with a special Belden cable with shielded pairs than with a USB cable, for example. I do not know if that translated into detection superiority (couldn't test that). Also, both nulls happened to be well within the headroom of the pre-amp.

Magnetic coupling, in theory, should be able to be canceled quite well by shifting our coils I think. So maybe a little circuitry to cancel some of the "capacitive" coupling could be employed if needed to accomodate a coil whose residual null can't be knocked down enough by shifting the coils.

-SB

[Davor]

There is no reason why it couldn't be well balanced as any other coil system. I dunno. I'll surely concentrate on possible causes of residual voltages. I have some ideas, but not too certain at the moment.

[Tinkerer]

Yes, it is a simplified view of the matter. In reality the flux thing is a bit more complicated, especially near the outer loop wires. Calculations tend to be quite complicated.
Considering the mismatch due to the ground proximity you are much better off with inner coil being closer to the outer one. Aziz already confirmed such coils to be better for deeper scans.

Relative larger RX coils give more residual voltage. More residual means less head room for the pre-amp. Less headroom means less possible gain.

The best gain is at the pre-amp.

Tinkerer

[ApBerg]

Thanks Apbergi
Ive saved the pdf file on your forms
Did you do a detailed construction pdf for making and setting up coplaner for the TGSL or any others would be helpful.

Many thanks

Regards

Hello Dave,

No.. I have not more detailed pdf's ... sorry...

Best regards.

Ap