Advanced PI coils

[Aziz]

Input Required on AI DD Coils

Hi all,

I need your input now. There are anti-interference DD coils available (co-planar).
But there are many options possible:

Option 1:
Transmit: One D TX=RX1
Receive: Other D RX2
Differential: RX1 - RX2 (anti-phase connected or electronically subtracted)
RX1, RX2: induction balanced geometry or not?

Option 2:
True figure-8 type
Both D's TX, = Both D's RX (anti-phase orientation)

Option 3:
Round TX (covering both DD's)
RX1, RX2 (separate, inside TX, anti-phase connected or electronically subtracted)
RX1, RX2: induction balanced geometry or not? Or just figure-8?

Other options?

Which option is most used?
Which option is most interesting to you?

Cheers,
Aziz

PS: I'm referring to the pulse induction (PI) DD coil of course (but AI configuration).

[IBGold]

Hi Aziz,

At present on our Tinkerer GG project which of course is TEM-TX I am testing a DD that consists of one side std type DD TX coil but the RX coil is a differential coil bifilar wound the other test coil is a concentric with differential RX but I am still having problems with the processor board so things have come to a standstill but hopefully am about to change to a UNO32 board when I get some code interesting times.

Regards, Ian.

[Aziz]

Hi Aziz,

At present on our Tinkerer GG project which of course is TEM-TX I am testing a DD that consists of one side std type DD TX coil but the RX coil is a differential coil bifilar wound the other test coil is a concentric with differential RX but I am still having problems with the processor board so things have come to a standstill but hopefully am about to change to a UNO32 board when I get some code interesting times.

Regards, Ian.

Sorry Ian, but your input doesn't help me further.

The differential RX coil you are meaning does not subtract induced signals in the RX coil (its adding). It is merely subtracting common mode signals on it's leads. Magnetic induction will be a differential mode signal and thus not being subtracted.

The "differential" I am meaning is, that the induced signals in RX+, RX- coil are subtracted (a typical anti-interference operation). How you do that is irrelevant: Either anti-phase serial connected or subtracted in a differential amplifier.

I'm am referring to the pulse induction DD coils, where you can change it's operation mode:
- Double D, Monoloop, Cancel mode

TX (left D), RX (right D):
windings overlapped like an induction balance (IB) state or not?
A PI machine does not really require the IB condition.

Cheers,
Aziz

[Mechanic]

Hi Aziz,

In post #142 you posted the data for the TX/RX (separate) concentric co-planar PI coils (not IB, not AI!) Which the best results were obtained with the receive at .33 tx with the result at 30" 1.034. Then in post #149 you posted the data for the same coil except with an AI coil mounted 10" above the coil assembly. The result is substantially less than the non AI variant. My question is, would there still be enough target response received by the AI coil to subtract that much response from the target? I would have thought that another 10" above the primary receive with the target at 30" that there would be little effect on the target response.

Cheers Mick

[Aziz]

[ATTACH]22447[/ATTACH]

Some words in a few sentences would help what you are meaning Sergey.
And don't make the sketch too small. I don't have eyes like an eagle.


Cheers,
Aziz

[Aziz]

Hi Aziz,

In post #142 you posted the data for the TX/RX (separate) concentric co-planar PI coils (not IB, not AI!) Which the best results were obtained with the receive at .33 tx with the result at 30" 1.034. Then in post #149 you posted the data for the same coil except with an AI coil mounted 10" above the coil assembly. The result is substantially less than the non AI variant. My question is, would there still be enough target response received by the AI coil to subtract that much response from the target? I would have thought that another 10" above the primary receive with the target at 30" that there would be little effect on the target response.

Cheers Mick

Hi Mick,

don't forget, that the comparison is referring to the equivalent 10" diameter mono coil (TX=RX) regards to EMI noise induction (SNR).
The N(RX)*A(RX) is kept constant in all RX coils (RX, RX-, RX+).

Let's calculate the N(RX)*A(RX) product.

Our reference coil:
10 inch mono loop round coil, N=20 (let assume, it has 20 turns only).
Note: TX = RX (mono coil)
N(RX) = 20
A(RX) = pi*R² = 3.14 * (12.7 cm)*(12.7 cm) = 506,7 cm² (flux area for one turn)
N(RX)*A(RX) = 20 * 506,7 cm² = 10134 cm² (total flux area)
That's our ultimate reference in our comparison.

Our small RX coil (0.33):
R=(10"/2) / 3 = 1.67" = 4.2 cm (radius)
A(RX) = pi*R² = 3.14 * (4.2 cm)*(4.2 cm) = 56.3 cm² (flux area for one turn)
As we make the N(RX)*A(RX) = 10134 cm² for all RX coils, we have to increase the turns count of the small RX coil so that it gets total 10134 cm².
N(RX) = 10134 cm²/56.3 cm² = 180 turns
Our small 10"/3 diameter RX coil must have 180 turns count to induce the same amount of EMI noise as the 10" diameter reference RX coil.

For a radius/diameter factor of 0.33 (1/3 diameter) you have 9 times less flux area and thus less EMI noise induction (and less target response induction as well). You could ramp up the gain. In my coil comparison, the "gain" is made via more turns count of the small RX coil so that a 10" large mono RX coil is picking up the same EMI noise as the 3.3" small RX coil.
The shown figures tell you, how the actual coil performs at a given detection depth regards to the reference coil (10" mono coil). BTW, the TX coil is normalized to 300 µH, regardless of coil type.

One important note to the small RX coil with high turns count (N=180). It is obvious, that it would have a higher coil capacitance and higher coil inductance. You might not able to make it as much as required regards to fast coil condition (high bandwidth). In this case, you make a low turns count RX and compensate the induction losses with a higher gain in the low noise amplifier. Low turns count RX would induce low EMI noise of course and you can increase the gain provided that, you use a low noise amplifier.

Anti-interference (AI) mode subtracts a considerable amount of target response (dependent on the co-axial distance in this example). But it removes EMI noise to some degree and you can crank up the gain on the other hand to compensate the target response losses. The goal is to increase the target response (SNR).

In our AI example at 30" depth (10 inch co-axial distance, ratio = 0.597):
We would require additional gain of 1/0.597 = 1.7 times to be equal to the reference coil at 30" depth. Considering the fact, that EMI cancellation factor is much higher, we can easily compensate the losses and add more gain to get more target response. More target response -> more detection depth.

The get the benefit of AI modes, one needs an ultra low noise amplifier of course as the amplifier noise is getting critical. Same applies to the small RX coil, which might not be made with as much turns as required to be equivalent to the reference coil and therefore it would require more gain too (but the EMI noise induction is less - should not be a problem with more gain).

Let's make an example for the normal mode (not AI):
N(RX)=180 (optimal to be equal to mono coil)
But we make it 60 turns (1/3 less turns count) and would get three times less target response and three times less EMI noise. We would require 3 times more gain in the amplifier to compensate for the induction losses.

Let's make an another example: AI mode (at 30 inch target depth)
N(RX)=60 turns (not optimal, less capacitance, less inductance)
We would require additional gain of 3*1.7 = 5.1 to be equal to the mono coil. If we get an EMI noise rejection ratio of 10 times (due to AI mode), we could increase the gain to 10 times and can calculate the gain benefit.
Gain benefit = 10/5.1 = 1.96
Our target response would be 1.96 times more compared to the mono coil at 30 inch depth. It would give us some detection depth increase. But don't forget, that the detection depth and target response signal level isn't a linear function. So the detection depth increase isn't much.

I hope, this clarifies the whole analysis a bit. If not, ask further please.

Cheers,
Aziz

[Aziz]

Ground Noise Considerations in a PI Coil

Hi all,

I could (or maybe should) show you the benefits of the small separate RX coil in a PI configuration and its benefits regards to ground noise pickup on high susceptible mineralized grounds.

This has been rised by Eric in this thread earlier but we didn't analyse or prove its benefits in detail.

Ready for another shock?

Aziz

[Mechanic]

Hi Aziz,

Thanks for the clarification, that makes sense now. The other thing worth considering is for a deeply buried target that might only give a small response when detected with a mono coil, that the AI coil mounted 10" above the coil assembly would not pick up any response from the target, thus little to no target signal would be subtracted by the AI coil, so there would be no depth loss. On a target that gives a much bigger response, it is likely that some target signal would be subtracted by the AI coil, however it would never be enough to make the target un-detectable. Even if the distance for the AI coil was reduced from 10" to 5", a target, though the response would be reduced, once again, it should remain detectable, especially if the target is deep and "just" detectable.

Now with the winding of the receive coils, rather than using a single strand @160 turns, can say a fine litz with 10 strands be used @ 16 turns? Also winding the receive coils in a basket weave will go a long way to reducing the inter-winding capacitance. Then the question of whats a good low C shielded cable that can be used between the coils and the detector? The other option is to mount the preamp(s) closer to the coil, however it would need to be shielded with a shielding that does not support eddy currents for very long.

I while ago I did some tests increasing the gain of the preamp and testing for noise. From memory I think I was able to double the gain without any noticeable increase in noise(without the coil, preamp connected to coil ground) However, when connected to a coil(not AI) the detector was practically un-usable. I also tried using 2 coils, one small and the other large and tried increasing the gain on one preamp to cancel emi, but I was not happy with the results, though this was at my house which has heaps of emi so I will have to re-visit this one later...

Cheers Mick

[Aziz]

..
Now with the winding of the receive coils, rather than using a single strand @160 turns, can say a fine litz with 10 strands be used @ 16 turns? Also winding the receive coils in a basket weave will go a long way to reducing the inter-winding capacitance. Then the question of whats a good low C shielded cable that can be used between the coils and the detector? The other option is to mount the preamp(s) closer to the coil, however it would need to be shielded with a shielding that does not support eddy currents for very long.
..

Hi Mick,

as the RX coil does not carry any significant current (voltage measuring), the RX coil wire can be thin. Any winding technique or wire, that will help reduce the parasitic capacitances, is ok. If you don't have much loop turns for the RX, you know, you have to amplify more and this could be critical due to amplifier noise. 16 turns is low. Keep it at least to a RX inductance of 300 µH if possible.

..
I while ago I did some tests increasing the gain of the preamp and testing for noise. From memory I think I was able to double the gain without any noticeable increase in noise(without the coil, preamp connected to coil ground) However, when connected to a coil(not AI) the detector was practically un-usable. I also tried using 2 coils, one small and the other large and tried increasing the gain on one preamp to cancel emi, but I was not happy with the results, though this was at my house which has heaps of emi so I will have to re-visit this one later...
..

Pay attention to the phase orientation of the RX+ and RX- coil! One can make easily mistakes here.

My workbench even has lots more EMI noise. And the AI configuration reduces it significantly. It is very very important, that both RX+ and RX- coil parts are equal (same geometry, same size, same turns, same L, same R, same C, same ... ) .
That's the reason, why I build my RX+ and RX- coils out of ribbon cable. You get almost perfect identical RX coils.

Cheers,
Aziz

[Aziz]

Magnetic Flux Linkage Relation of TX/RX PI Coil

This is the relation of magnetic flux of the 10 inch TX coil and the various separate concentric co-planar RX PI coil. The magnetic flux is the integrated magnetic field density over the RX coil flux area.

Simplified:
Magnetic flux = B*A,
B = magnetic field
A = RX coil flux area
(see http://en.wikipedia.org/wiki/Magnetic_flux . S ::= A)


The magnetic flux is calculated on the compensated RX coils (the smaller, the more loop turns, the higher the inductivity to get the same level of EMI noise as the 10" TX=RX coil).

The relation is mag flux (RX)/mag flux (TX).

See, what comes out:

(Pay attention: The RX diameter factor isn't linear, only calculated for given values.)

This is the reason, why the TX/RX (small & separate RX coil) will see less ground noise (not confuse with EMI noise).

Cheers,
Aziz

[Sergey_P]

Some words

Noise from the ground is determined smoothness of the surface of the soil.
Magnetic fields TX1 and TX2 are compensated at a distance from the area of the coil, H= height search.
E(target)=E(t.rx)+E(t.cx)
My tester told me that the best results were obtained when H~7cm

[Aziz]

A better magnetic flux linkage relation RX/TX of the graphics above:



The RX factors between 0.2 .. 0.3 are hypothetical and aren't convenient for us as you would require more RX loop turns (=more coil capacitance). I would even go for 0.4 and up.

Cheers,
Aziz

[Davor]

You often mention "ultra low noise preamplifier" lately, but to what end? We may assume an amplifier to be very low noise at 1nV/sqrt(Hz) and that's equivalent to ~60ohm source resistance. Going lower than that is not so practical because all designs with less noise draw enormous currents. Going along with a Muhammad and the hill proverb, it is a better compromise to reach the noise limit from the coil side, rather than the preamp side. A practical PI with first sample being later than 10us will be bandwidth limited to below ~100kHz, which would not be a significant limiting factor for coils up to maybe 10mH, and resistances ~30ohm, which is close to a noise limiting resistance. I'd say this separate Rx approach is much closer to the ideal S/N in terms of a possible turns ratio gain.

[Aziz]

You often mention "ultra low noise preamplifier" lately, but to what end? We may assume an amplifier to be very low noise at 1nV/sqrt(Hz) and that's equivalent to ~60ohm source resistance. Going lower than that is not so practical because all designs with less noise draw enormous currents. Going along with a Muhammad and the hill proverb, it is a better compromise to reach the noise limit from the coil side, rather than the preamp side. A practical PI with first sample being later than 10us will be bandwidth limited to below ~100kHz, which would not be a significant limiting factor for coils up to maybe 10mH, and resistances ~30ohm, which is close to a noise limiting resistance. I'd say this separate Rx approach is much closer to the ideal S/N in terms of a possible turns ratio gain.

Davor,

Ultra low noise amplifier noise density := 1nV/sqrt(Hz)
(This is enough for us and we need some noise for the Gaussian distribution.)

The ultra low noise amplifier is important for AI configurations, where you have to compensate for the response losses. You can't beat the best without considering the amplifier noise. But we are going to beat the best.

Aziz

[Davor]

I have no problem with that. My point is that even with a real low noise pre you may beef up the Rx coil a little bit so that S/N becomes better than monocoil, and you actually can do that without much effort. 10dB is 10dB. It is not free, but it is well worth the effort and the extra wire.