Latest Minelab Patent - 11th Feb 2016

[6666]

I built a target ID PI back about 1988 that came close to going into production. It was unveiled for the distributors who however said it was too quirky to be marketable (I'd told the boss the same thing but he was in love with the machine and I couldn't dissuade him.) After 2 weeks of deep funk the boss scrapped the project and we went on to do the CZ platform which had already been in the works before we got distracted by target ID PI.

What a pity your project got shelved, maybe you could get somebody interested in today's environment.

[Teleno]

In this example figure 3, ground balance is for log-uniform + log-linear VRM cancel.
...
Hope I help understand why this is not the same as prior art with just simple X balance or no ground balance.

But then:

Australian Patent Application Nos. 2007304831 (frequency or time domain cancel log-uniform + log-linear VRM simultaneously) and 2009243482 (PI time domain cancel saline + log-uniform VRM simultaneously).

These GB techniques disclosed in the ML patent are known from the prior art.

The two-time-constant discrimination technique is also known from the prior art.

GB and discrimination solve different, independent problems. The skilled person is aware of their respective benefits and is always motivated to combine them. The particular combination in the ML patent produces no unexpected effects beyond the predictable simple sum of the known effects of each technique taken separately. Such a combination is, therefore, a mere superposition of previously known features that's obvious to the skilled person and is not inventive.

[daverave]

I built a target ID PI back about 1988 that came close to going into production. It was unveiled for the distributors who however said it was too quirky to be marketable (I'd told the boss the same thing but he was in love with the machine and I couldn't dissuade him.) After 2 weeks of deep funk the boss scrapped the project and we went on to do the CZ platform which had already been in the works before we got distracted by target ID PI.

its a shame that you did not follow through with the ID pulse system...although the fisher CZ is an excellent design with good proformance...i feel now that the market needs a good ID pulse machine which will give people the depth they need with the ID system...im supprised companies like fisher,tesoro,whites never produced a pulse machine that can ID different metals....hopefully my friends machine will be finished by the end of the year.

[clancy]

“Well YOU read carefully, on [0059] and [0063] the authors refer the reader to two PREVIOUS PATENTS for the soil-insensitive channels (saline & GB) which are therefore neither new nor the subject of this application. No alternative GB is disclosed.”
Agree. I do not say otherwise. Agree this patent get target TC spectrum (“at least” 2 values) from properly ground balance channels. But prior art get it from unground balanced channels. Big difference in complex.


O.K. you not think patent why patent is new. I try explain.
Quote:
“In THIS EMBODIMENT, there is a total of seven channels (note 7) which are demodulated with demodulation functions that cancel log-uniform and log-linear frequency dependent signal components from the soil. They could also cancel the conductive signal component from the soil. The demodulation functions are constructed in ways taught by Australian Patent Application Nos. 2007304831 (frequency or time domain cancel log-uniform + log-linear VRM simultaneously) and 2009243482 (PI time domain cancel saline + log-uniform VRM simultaneously).”
First use demodulation that fully ground balance receive signal like Minelab PI (Saline + log-uniform VRM or log-uniform + log-linear VRM cancel) or Eric Foster PI (log-uniform VRM cancel). This cause complicate in function of TC or w. See figure 3. Note nulls and changes sign as function w, where target w = 1/TC on x-axis. In this example figure 3, ground balance is for log-uniform + log-linear VRM cancel. You can see this completely different to, e.g. multifreq with just simple X ground balance for each frequency. Simple X ground balance for Tx frequency wo make well-known approx. wwo/[w^2+wo^2] where w is for single TC target and wo Tx frequency. Very simple, no nulls, no sign changes, therefore no ambiguous. In time domain, same problem as can see in figure 3; nulls and sign changes. This is why Eric Foster PI that ground balance log-uniform change sign for audio for different target TC and can use this for simple discrim.
OK, now with this complicate demodulated log-uniform + log-linear VRM cancel signals in figure 3, some of the 7 receive signals have 2 nulls some 3. Then transform to w or TC spectrum.
Hope I help understand why this is not the same as prior art with just simple X balance or no ground balance.

[Dave J.]

my friend is developing a pulse detector with ID for different metals and will show good ID irrespective of depth...the prototype worked and now he is going to make a working machine for use on the beach.

I built a target ID PI back about 1988 that came close to going into production. It was unveiled for the distributors who however said it was too quirky to be marketable (I'd told the boss the same thing but he was in love with the machine and I couldn't dissuade him.) After 2 weeks of deep funk the boss scrapped the project and we went on to do the CZ platform which had already been in the works before we got distracted by target ID PI.

[daverave]

he has already made a single coil vlf motion machine with discrimination....and now he is going to make this ID display pulse machine...he told me it involves lots of chips...i saw the proto type some time ago which worked well and showed how different metals have their own ID but im not as technical as him and cannot really understand the in's and out's of how it works.

[6666]

my friend is developing a pulse detector with ID for different metals and will show good ID irrespective of depth...the prototype worked and now he is going to make a working machine for use on the beach.

Thats interesting news.

[daverave]

Lots of fancy gobbledegook in the disclosure, but that has very little to do with the claims.

Evidently a newbie "inventor" who has no idea this stuff actually has a history. I can't believe Candy let him get away with this crap, maybe Minelab is shoving Candy out? Most of the claims that can be interpreted as frequency domain read right on top of the Fisher CZ (released November 1991 and still being manufactured) and I would suppose the same is true of the Minelab's multifreakers whose commercial history is also about 25 years old.

Then there's discriminating PI's that have been around probably even longer. Although to my knowledge there's never been a visual target ID PI marketed, it's not because of any shortage of people who know how to do it, it's a matter of how to get a competitive product out of the deal. The knowledge how to do it is obvious to anyone who knows how to construct a discriminating or "ground cancelling" PI, whether or not they've actually built one.

my friend is developing a pulse detector with ID for different metals and will show good ID irrespective of depth...the prototype worked and now he is going to make a working machine for use on the beach.

[Teleno]

Read more careful. Example: 0059 to 0064. Signals are saline (conductive ½space) + VRM ground balance. After ground balance signals then deconvolute.
(Hey my friend Aziz, I miss you here.)

Well YOU read carefully, on [0059] and [0063] the authors refer the reader to two PREVIOUS PATENTS for the soil-insensitive channels (saline & GB) which are therefore neither new nor the subject of this application. No alternative GB is disclosed.

What this patent is about is synthesised in [0066]: fitting two time constants to every target signal. Not new not inventive, plain plagiarism.

[clancy]

Read more careful. Example: 0059 to 0064. Signals are saline (conductive ½space) + VRM ground balance. After ground balance signals then deconvolute.
(Hey my friend Aziz, I miss you here.)

[Teleno]

At last a interesting topic after long time.
I read this patent very careful. It is much more complex than you think.
It decompose target signals that ground balance saline soil with also log-uniform and with also log-linear VRM.
This much more complex than decompose into TC spectrum with no ground balance signals.
I think this not done before.

There's no teaching regarding ground balance or aline sol in the patent, though the former is mentioned in the claim. It just discloses a two TC model of a target and fitting the response to it. It's been done before by multiple people.

[clancy]

At last a interesting topic after long time.
I read this patent very careful. It is much more complex than you think.
It decompose target signals that ground balance saline soil with also log-uniform and with also log-linear VRM.
This much more complex than decompose into TC spectrum with no ground balance signals.
I think this not done before.

[Davor]

I'm at loss why they let multifreakers (thanks Dave, I like this) lapse. None of their multifrequency technologies are under any running patents right now.

[ivconic]

"...The patents are themselves worthless yet ML manages to gather the best existing state-of-the art and publish it in compiled form..."

It makes sense.
I stand corrected.
I'll give it second chance now...

[Teleno]

As i said; reading those patents is HUGE waste of time.

Yes and no. The patents are themselves worthless yet ML manages to gather the best existing state-of-the art and publish it in compiled form.

In this case - for those who actually read the patent and had trouble understanding - it's about decomposing a target signal into its main components.

We usually think of a target as a decaying exponential with a time constant "tau". In reality things are more complex. Targets have shapes that form separate paths for the eddy currents, each path with its own R and L. A target signal is the a sum of multilple decays with multiple taus.

If we manage to find out these taus we get a spectrum of time constants (a signature) characteristic of each kind of target. The problem of finding the taus is called "inversion" and is ill posed because when it comes to exponentials the signal is quickly overcome by noise, giving a disparity of possible solutions for the same data. ML (plagiarizing the UXO papers above) simplifies the problem by attributing only two taus to every target and then finding the best match.

This technique is something I had contemplated a decade ago and even wrote a java program to decompose a sum of exponentials. Nothing new under the sun, but if you follow ML patents you'll be kept up to date on these otherwise arcane developments that are hard to find.