DEEPER PI DETECTION DEPTH

[simonbaker]

Target tau is directly related to VLF phase.

Then is it also true that VLF phase is subject to both target metal (conductivity) and target size/shape?

If so, how do we interpret our discriminators when they say "foil" or "penny" or "nickel" or "gold" or "silver"? Is it only accurate for coins of known size? Will a large zinc plate have the same phase/tau as a small silver coin, or something like that?

Regards,

-SB

[Davor]

It would relate only in cases of thin structures where thickness is close to skin depth. E.g. foil. In such cases you'll get response as if it was a target of lower conductivity. Target mass is irrelevant for phase.

You may extend this idea to Tx coil wire gauge. Inductance will be the same for any coil of similar number of turns on a same size form. However, the coil Q will degrade if wire gauge is too thin.

By little mind stretch it is very same for PI as well.

[simonbaker]

It would relate only in cases of thin structures where thickness is close to skin depth. E.g. foil. In such cases you'll get response as if it was a target of lower conductivity. Target mass is irrelevant for phase.

You may extend this idea to Tx coil wire gauge. Inductance will be the same for any coil of similar number of turns on a same size form. However, the coil Q will degrade if wire gauge is too thin.

By little mind stretch it is very same for PI as well.

That was what I previously assumed, but Qiaozhi and Carl seemed to indicate that target size is very important for "tau". This seems backed up by the huge range of "tau"s mentioned, much larger than the range of conductivities of typical targets.

So I'm still wondering....

-SB

[Davor]

Of course, things in nature come with obstacles not present in simulations. Also with oxidation and rugged surfaces, difficult soils, odd orientations etc. However, even with a flattened target response distribution curve, you'll get taus neatly grouped at distribution maximum for a given metal.

Given a choice between discrimination and no discrimination, I'd go with discrimination.

[simonbaker]

Of course, things in nature come with obstacles not present in simulations. Also with oxidation and rugged surfaces, difficult soils, odd orientations etc. However, even with a flattened target response distribution curve, you'll get taus neatly grouped at distribution maximum for a given metal.

Given a choice between discrimination and no discrimination, I'd go with discrimination.

Ok. Maybe Tinkerer et al will be able to show us some data on real targets that will illustrate this.

-SB

[Aziz]

Aziz,

Like everyone else I'm perplexed about exactly what the question is your asking but to keep the ball rolling I'll attempt to draw some (though quite probably not the ones your driving at) conclusions from your plot.
* Long time constant targets seem to respond maximally to a broader range of frequencies.
* With shorter time constants the maximum target response occurs at a higher frequency.
* As the frequency increases into the Mhz range target time constants starts to become irrelevant and all the target responses diminish together.

Midas

Excellent Midas!


Why was it so much difficult to see the obvious behaviour?
We should interpret the findings further.
Any ideas?

Cheers,
Aziz

[simonbaker]

Excellent Midas!


Why was it so much difficult to see the obvious behaviour?
We should interpret the findings further.
Any ideas?

Cheers,
Aziz

Actually, looking at the graphs makes me want to do this:

Design a continuous wave MD (can't call it VLF, but same design) with two frequencies:

1. 200 kHz - this gives high amplitude response for all targets, but not much phase differentiation.

2. 5 kHz - this has differing amplitude responses, but good phase differentiation for discrimination.

If you ignore the longest time constant target, you could raise the second frequency to 50 to 100 kHz and get better discrimination depth. Or just compromise and use 100 kHz for both, but some trouble detecting short TC targets.

I'd still like to see some real data on how TC ("tau", time constant) is a function of target conductivity and size/shape.

-SB

[Aziz]

Actually, looking at the graphs makes me want to do this:

Design a continuous wave MD (can't call it VLF, but same design) with two frequencies:

1. 200 kHz - this gives high amplitude response for all targets, but not much phase differentiation.

2. 5 kHz - this has differing amplitude responses, but good phase differentiation for discrimination.

If you ignore the longest time constant target, you could raise the second frequency to 50 to 100 kHz and get better discrimination depth. Or just compromise and use 100 kHz for both, but some trouble detecting short TC targets.

I'd still like to see some real data on how TC ("tau", time constant) is a function of target conductivity and size/shape.

-SB

Hi SB,

good idea. I think 20-100 kHz would be enough too.

Regarding the phase change:
This will happen always due to coil coupling/inductivity change due to mineralisation of the ground/ferrous targets either. Hey, don't look at the f...ing discrimination at this stage. We should look at it, when we have tamed the ground balance first. We are light years away from it.

Power efficiency (worth to look at it):
If you change the spice circuit I have provided and run a transient analysis (commenting the .ac analysis and uncommenting the .trans command) and would compare both frequencies (high and low), you will find, that the low frequency stimulation cost's enormous battery power compared to the high frequency stimulation.
This is really an eye opener!
That's the reason, why low frequency VLF detectors have high inductivity TX coils. Well, I have taken the standard PI coil (300µH / 0.4 Ohm) in the circuit simulation.

Well the target stimulation response level tends to go to the same level for all targets (low, mid, high TC targets) if we increase our operating frequency. But the high frequency region is more power efficient (saturation of the TX coil current does not happen in the high frequency region).


Aziz

[simonbaker]

Hi SB,

good idea. I think 20-100 kHz would be enough too.

Regarding the phase change:
This will happen always due to coil coupling/inductivity change due to mineralisation of the ground/ferrous targets either. Hey, don't look at the f...ing discrimination at this stage. We should look at it, when we have tamed the ground balance first. We are light years away from it.

But if a target pops out of the background, maybe the info is readable?

Power efficiency (worth to look at it):
If you change the spice circuit I have provided and run a transient analysis (commenting the .ac analysis and uncommenting the .trans command) and would compare both frequencies (high and low), you will find, that the low frequency stimulation cost's enormous battery power compared to the high frequency stimulation.
This is really an eye opener!
That's the reason, why low frequency VLF detectors have high inductivity TX coils. Well, I have taken the standard PI coil (300µH / 0.4 Ohm) in the circuit simulation.

Good observation. Would be hard to have optimal efficiency at two frequencies, unless double resonant coils (messy).

Well the target stimulation response level tends to go to the same level for all targets (low, mid, high TC targets) if we increase our operating frequency. But the high frequency region is more power efficient (saturation of the TX coil current does not happen in the high frequency region).

Interesting...

Of course if we made the TX circuit resonant at some frequency(s), that relationship could be changed.

-SB

-SB

[Midas]

So what your sim needs now Aziz is a realistic non-constant and unpredictable ground effect component, then you can play around trying to remove it. From what I gather Minelab are doing this already by using some clever maths to combine the low frequency response with the high frequency response. They are actually use 28 frequencies which might be more to increase the wank factor than because its necessary or who knows perhaps it really does help.

Here's Minelabs consumer level explanation of their technology:

Find Every Target Type & Size with Every Sweep Generally, high transmit frequencies are more sensitive to small targets and low transmit frequencies give more depth on large deep targets. FBS simultaneously transmits and analyses a full band of multiple frequencies from 1.5 kHz to 100 kHz and is therefore sensitive to both very small and large deep targets at the same time. This means you only need to cover the ground once and can be confident you’re not leaving ANY valuable treasure behind.

If that's to be taken literally and isn't just sales patter then once your model is complete you should see some improved sensitivity to long TC targets by using a lower frequency that isn't captured in your current model. Perhaps as a result of the increased attenuation effect of the ground at high frequencies.

[Aziz]

But if a target pops out of the background, maybe the info is readable?

Good observation. Would be hard to have optimal efficiency at two frequencies, unless double resonant coils (messy).

Interesting...

Of course if we made the TX circuit resonant at some frequency(s), that relationship could be changed.

-SB

-SB

Well, the real reason, why the standard VLF coils have a high inductivity should be due to to get the resonant capacitor small enough, which is placed in the coil of course. And they don't require a high current flow through the TX coil so a thin wire can be used for it without becoming the search head to much heavy.

Even a 100µH coil could be used as a TX coil. That's convenient, if the upper frequency of a dual-frequency VLF is high enough.

The efficiency is very high, if the most of the TX energy is held locally in the coil's resonant tank.

Aziz

[Aziz]

So what your sim needs now Aziz is a realistic non-constant and unpredictable ground effect component, then you can play around trying to remove it. From what I gather Minelab are doing this already by using some clever maths to combine the low frequency response with the high frequency response. They are actually use 28 frequencies which might be more to increase the wank factor than because its necessary or who knows perhaps it really does help.

Here's Minelabs consumer level explanation of their technology:

Find Every Target Type & Size with Every Sweep Generally, high transmit frequencies are more sensitive to small targets and low transmit frequencies give more depth on large deep targets. FBS simultaneously transmits and analyses a full band of multiple frequencies from 1.5 kHz to 100 kHz and is therefore sensitive to both very small and large deep targets at the same time. This means you only need to cover the ground once and can be confident you’re not leaving ANY valuable treasure behind.

If that's to be taken literally and isn't just sales patter then once your model is complete you should see some improved sensitivity to long TC targets by using a lower frequency that isn't captured in your current model. Perhaps as a result of the increased attenuation effect of the ground at high frequencies.

Ground effect circuit simulation is a non-trivial task. It is even not understood well enough in the MD business.

It's way easier, to solve the ground effect problem in the firmware (software solution).

Regarding the ML's 28 FBS:
The clever math is called FFT (Fast-Fourier-Transform). If they really use 28 frequencies, they must use a 64-point FFT, which delivers 32 frequency demodulation components. The 0 - operating frequency range can't be used and very likely some of the upper frequency ranges as well. So 28 out of the 32 frequency components were used at the end.

No problem, you could have more frequencies if you have enough processing power. But you could do it with less frequency components too.

The Minelab's consumer level of explation is correct. To see the proof, just have a look at the bode plot I'm talking about for a long time now.

Well, we didn't talk about the frequency response of the transmitter. Every transmitter type has it's own frequency response. A single frequency VLF uses a single frequency and the total TX energy is focused at this frequency.
A PI transmit pulse is a wide band pulse, which has different content of energy in each frequency spectrum.
If I compare VLF with PI, the PI type isn't efficient as most of the emitted TX energy isn't processed at the end.

Some food so far to think about.

Cheers,
Aziz

[Davor]

But the high frequency region is more power efficient (saturation of the TX coil current does not happen in the high frequency region).

That would depend on pulse duration only. You get a true impulse response with step voltage source. E.g. like with switching power supplies.

System response would still be highly dependent upon coils loading, and I see even this going into favor of step voltage supply: cold driving transistors, flat frequency response, preserved phase response.

OT:
I worked as an RF engineer at a transmitters factory, doing mostly MW/SW designs, and the most ingenious thing at the time was a Harris transmitter which had a copper rod as a summing device for a multitude of driving elements, and operating as multiple transformers in series. (see picture) In case any of the drivers broke, it was SHORTED automatically to avoid damage to the device, and maintain continuous operation of the whole transmitter. So the broken driver of a coil shorted it in order not to hamper current flow through the rod. Just brilliant. Mind you, it was a well above 80% efficient - even with broken devices.
If I stretch the principle just a tiny bit, when I want to detect a tiny response from something deep in the ground, the last thing I wish to do is couple any kind of resistance to it.

[Davor]

It just hit me

Using a thick metal ring, copper or aluminum, and a ferrite-current pliers-like-transformer, just like in the above example, you could forget about the capacity problem in PI coils forever. Such contraption would act as a coil of the ring diameter, but with number of turns equivalent to the number of turns on a ferrite thingy. In case of aluminum it could be self supported as well, and as a bonus you'll be able to forget about shielding alltogether.

Did I earn a lollipop or what?

A step forward... I could have a pair of separated search rings (with no IB overlapping) and supply them with in phase for Tx and counter phase for Rx balance. Using aluminum rings it could make a very lightweight construction, with 100% metal used for both Tx and Rx, and in perfect balance as well.

As Aziz said before, patent hogs beware! This idea is NOT for commercial use.

[Aziz]

Hi all,

what's the reason for the increased interest in this topic thread?
The number of viewers went up.

I rather would like to see more contributions. This ain't a team work otherwise.
And:
Patent Trolls! Keep Out!

Aziz