GROUND BALANCE ( without the ground :-) )

[moodz]

Bravo Moodz,
You are a real magician.
I am really glad of your success. I wish you always to be the best and do new and new things.




Indeed, some PI detectors with very early first samples (most notably gold-seeking detectors) and fixed delays for GB samples may not be able to balance certain targets by adjusting the gain in the GB channel/s.
But if we reverse things (fixed the gain and change GB sample/s delay), I think every PI MD could do it.

Doh ... you figured out how I did it ! Now my patent is broken ... I will have to invent something else.
There goes the weekend.

[boilcoil]

Ha, ha, don't joke with me.
I'm sure you're using some kind of magic algorithm.
I'm really happy for you.
Greetings.

[Carl-NC]

Indeed, some PI detectors with very early first samples (most notably gold-seeking detectors) and fixed delays for GB samples may not be able to balance certain targets by adjusting the gain in the GB channel/s.
But if we reverse things (fixed the gain and change GB sample/s delay), I think every PI MD could do it.

The results are basically the same whether you alter the ground channel gain or alter the ground sample. Adjusting the gain offers a wider GB range and if it cannot balance a target it means the range of the gain needs to be expanded. Adjusting the timing can run out of usable balance range because the signal is vanishing quickly, plus the GB becomes very non-linear. When I've used adjustable timing GB I also run the ground channel gain about 5x higher than the target channel.

[boilcoil]

I meant the gain of G channel/s is equal to or greater than the gain of the target channel/s. Then Tgr-Tair must be greater than Ggr-Gair. If the first(T) samples are very early, it may be found, in case of inappropriate disposition of the second samples, the difference in G samples is greater than the difference in the target samples for certain objects/stones.

Otherwise, I also use software amplification in the G channels from 2 to 5 higher.

[Carl-NC]

I'm not sure I understand what you are saying. The GB signal is

GB = Tgt - k*Gnd

where Tgt is the target channel sample and Gnd is the ground channel sample. There are also EFE samples but I'll ignore those. k is adjusted until GB = 0. k can be either the gain of a ground channel stage or the pulse width of the ground sample, or some of both. Regardless of the target (metal or ground) there is always a value of k that with null the response.

[moodz]

I'm not sure I understand what you are saying. The GB signal is

GB = Tgt - k*Gnd

where Tgt is the target channel sample and Gnd is the ground channel sample. There are also EFE samples but I'll ignore those. k is adjusted until GB = 0. k can be either the gain of a ground channel stage or the pulse width of the ground sample, or some of both. Regardless of the target (metal or ground) there is always a value of k that with null the response.

If that was the case ... how do you keep GB constant for varying target signal ? Above is a linear equation. For constant ground and k ... but changing target ( including no target ) GB will appear to vary.

[moodz]

BTW I found this great tool for testing and animating the various math algorithms for metal detectors and anything else( eg exponentials and phasor diagrams ) called GeoGebra
You can watch all the variables changing in simulation time.
Its a math simulator NOT a circuit simulator.

https://www.geogebra.org/calculator

Its free for non commercial use and open source.

Not only that it can export Latex equations and superb diagrams for your patent applications ( which I did very recently ).

Easier than Matlab.
Supports full scripting also. ( javascript ).

You can download the program or run in a browser online.
Anyway have a look.

[Carl-NC]

The variables are samples taken of the decay curve. I call the channels (and their samples) "Tgt" and "Gnd" but in reality they are just 2 samples on the same decay curve. A more precise equation would be

GB = response[t - t_Tgt] - k*response[t - t_Gnd]

This would be for point samples, but in reality we use a sample window and integrate. So the equation should actually be the difference of 2 integrals but I'm lazy and don't want to type all that out.

[moodz]

The variables are samples taken of the decay curve. I call the channels (and their samples) "Tgt" and "Gnd" but in reality they are just 2 samples on the same decay curve. A more precise equation would be

GB = response[t - t_Tgt] - k*response[t - t_Gnd]

This would be for point samples, but in reality we use a sample window and integrate. So the equation should actually be the difference of 2 integrals but I'm lazy and don't want to type all that out.

You are right of course ... just testing.

[Carl-NC]

Well, I'm stumped. Don't know what you could have done that's related to my patent but intuitively I can't see a way to implement a universal GB. I do know that some detectors have used ground blanking whereby the ground is essentially discriminated out but this is horrible at masking targets, and I would assume you know this.

BTW, I like the math simulator. Trying to figure out if there is a hint in that image.

[moodz]

Well, I'm stumped. Don't know what you could have done that's related to my patent but intuitively I can't see a way to implement a universal GB. I do know that some detectors have used ground blanking whereby the ground is essentially discriminated out but this is horrible at masking targets, and I would assume you know this.

BTW, I like the math simulator. Trying to figure out if there is a hint in that image.

Here is my current date legal advice on discussing details in provisional patents ... it is what I expected.
Yes, you can publish details about an invention after filing a provisional patent application, but it's crucial to do so within a 12-month timeframe before filing a full, non-provisional patent application to maintain patent eligibility.
I now have my PP number :
Patent 2025901043 Received 30 March 2025

So I can tell you why patent US 9285496 B1 was cited ... its to do with the terminology used ...." Truncated half-sine methods for metal detectors"
However reading the patent it becomes clear it is referring to truncated half-sine current waveforms. ( bipolar )

My application uses a waveform that is characterised by a half sine voltage waveform ( at both the TX and RX coil ) ... I will let you work out what sort of current waveform in the TX coil generates a half sine bipolar voltage waveform
... and US 9285496 B1 does talk about how to do ground balance ( but strangely not in the claims .... IMHO this is a critical omission ).

However there is a singular key difference between half wave sine current waveforms and half sine voltage waveforms that allows you to determine the ground balance point without reference to the ground.
The method of doing this is the patent ...the particular type of waveform is required to do the method though.

In summary US 9285496 B1 was cited to remove any confusion between current and voltage waveforms ( US 9285496 B1 has a half sine TX current waveform characteristic ... PP 2025901043 does not )

Thats what I can tell you without revealing the full detail which would require an NDA.

The image is just a phasor diagram from the examples on line.

PS I dont use truncation either.

[Orbit]

Hi Modz maybe a waveform like this?

[moodz]

Hi Modz maybe a waveform like this?

Is that the voltage or the current ?

Presuming its the voltage you are saying -pi/4 to +pi/4 wave segment.

[Orbit]

Yes voltage .. There really is a lot to these waveforms..I mean a lot of magic in the pi-ib region..

[moodz]

Yes voltage .. There really is a lot to these waveforms..I mean a lot of magic in the pi-ib region..

True but you have some sharp transitions there .... this is a key problem ... ADCs or sampling circuits have problems with sharp transitions as they are not NTSST ...not the same sample twice at the transition sample points ( due to jitter etc )
The presence of sharp transitions hides critical information in the magnetic circuit and if you are trying to work out a precise variable like GB then sample jitter will hide the required data.