How does ground balance work in GPX 4500?

[belax2018]

Can anybody give a clue on how ground balance works in gpx 4500 machine?
I am interested in theory behind it and a practical realization.

I am building my own hobby PI machine with MCU, FPGA, ADC, and fully differential front end. My coil is a two coil concentric type. TX is a single MOSFET – classical tx design – not a bipolar.
RX consists of two ad8139 + adc 40 msps + fpga.
In fpga I can do any integrations. I mean I can set any delays after TX and any t1-t2 where I do integration.
In MCU a do a configuration in FPGA (set delay and integration windows), also I do their oversampling, average values calculations and so on.
The device works well in the air. However, when it comes to ground I do not get any decent results.

I have looked through lots of pages in this forum studying the methods and theory behind Ground balance in monopole (single mosfet) PI machines. But cannot find a good working practical solution.

Right now, my ground balance works like this:

TX length is T=100 us.
Delay 1 (t0 - t1): 8 us.
Target Sample (t1-t2): 8 us
Delay 2 (t2-t3), where t3=3*t1: 8 us
Ground sample (t3-t4), where t3=3*t1, t4=3*t2.
EF sample: 800 us after TX end, length of EF sample is Ground Sample length minus Target Sample Length.
(P.S.: Sample means - I do integration of ADC values from RX coil in this sample window)
Final signal for processing: X = Target Sample – Ground Sample + EF Sample
Delay 2 is adjustable: So I do up and down with coil and adjust this delay until no change in signal is noticeable.
However, as I said, that method does not work well. It is not very stable.

Thank you.

[kt315]

we must rebirth QED anyway and give gold nugget seeking winter project for every australian.

[green]

Don't know about GPX4500. This is what I use. First delay is adjustable(6-20usec), first sample(target sample adjustable, used as ground balance adjustment), second delay(short 4 to 5usec), second sample(GB sample, fixed at about 100usec), EF sample time(second-first sample time)I take it just before Tx on. GEB timing.png shows settings to balance ground when changing first delay. Tx time, constant current or constant rate and probably others effect ground slope so changing either one could effect first sample time needed to balance ground.

[Qiaozhi]

Read this thread -> http://www.geotech1.com/forums/showt...ground+balance
It should answer your question.

[kt315]

https://en.wikipedia.org/wiki/Walsh_function

Applications of the Walsh functions can be found wherever digit representations are used, including speech recognition, medical and biological image processing, and digital holography.
For example, the fast Walsh–Hadamard transform (FWHT) may be used in the analysis of digital quasi-Monte Carlo methods. In radio astronomy, Walsh functions can help reduce the effects of electrical crosstalk between antenna signals. They are also used in passive LCD panels as X and Y binary driving waveforms where the autocorrelation between X and Y can be made minimal for pixels that are off.

[belax2018]

Don't know about GPX4500. This is what I use. First delay is adjustable(6-20usec), first sample(target sample adjustable, used as ground balance adjustment), second delay(short 4 to 5usec), second sample(GB sample, fixed at about 100usec), EF sample time(second-first sample time)I take it just before Tx on. GEB timing.png shows settings to balance ground when changing first delay. Tx time, constant current or constant rate and probably others effect ground slope so changing either one could effect first sample time needed to balance ground.

Thank you green for real numbers.

Would be interesting to know how much depth you loose with "your type of GB" in a real environment?

I know there can be two factors of loosing depth (sensetivity).
So cold "hole" in response - because of a curtain TC value (or a combination of TCs) of a target. I am not interested in this, as I am usually interested in low TC targets and here it is usually not the case.
What I am interested is the following. GB sample does not equally eliminates ground part in Target sample. I mean ground balance is not uniform over a slightly varying coil-soil distance or it is not uniform over soils with different density of minerals and so on. I don't know how to say it better. In other words how much do you need to decrease your "sensitivity pot" in order to work over the ground without constant false beeeps?


1) You say you adjust target sample width. How much (approximately in %) do you usually change it from the values stated in PNG file to reach GB in California? I will try with your method soon when the whether let me.
2) You mention ground slope depends on constant rate of TX current change. Here it would be very interesting to know what part of TX coil current slope Minelabbers use in GPX 4500. I have read somewhere that they prefer to use only first part of a coil current curve ( Icoil = V/R (1-exp(-Rt/L)) ), I mean the part which looks almost like a straight line.

[green]

Thank you green for real numbers.

Would be interesting to know how much depth you loose with "your type of GB" in a real environment?

I know there can be two factors of loosing depth (sensetivity).
1) So cold "hole" in response - because of a curtain TC value (or a combination of TCs) of a target. I am not interested in this, as I am usually interested in low TC targets and here it is usually not the case.
2) GB sample does not equally eliminates ground part in Target sample. I mean balance is not uniform over a slightly varying coil-soil distance or it is not uniform over soils with different density of minerals and so on. I don't know how to say it better. In other words how much do you need to decrease your "sensitivity pot" in order to work over the ground without constant false beeeps?

1) You say you adjust target sample width. How much (approximately in %) do you usually change it from the values stated in PNG file to reach GB in California? I will try with your method soon when the whether let me.
2) You mention ground slope depends on constant rate of TX current change. Here it would be very interesting to know what part of TX coil current slope Minelabbers use in GPX 4500. I have read somewhere that they prefer to use only first part of a coil current curve ( Icoil = V/R (1-exp(-Rt/L)) ), I mean the part which looks almost like a straight line.

The numbers in the PNG file are what was required to balance California ground. Ground from the back yard balances very close to if not the same settings. I'll try to answer some of your other questions later, need to do some yard work.

[belax2018]

https://en.wikipedia.org/wiki/Walsh_function

Applications of the Walsh functions can be found wherever digit representations are used, including speech recognition, medical and biological image processing, and digital holography.
For example, the fast Walsh–Hadamard transform (FWHT) may be used in the analysis of digital quasi-Monte Carlo methods. In radio astronomy, Walsh functions can help reduce the effects of electrical crosstalk between antenna signals. They are also used in passive LCD panels as X and Y binary driving waveforms where the autocorrelation between X and Y can be made minimal for pixels that are off.

Thank you KT315.
I can calculate an integral of f(t)=1/t^a. or understand why Minelab version of ground slope (in the pdf on their web site) is time invariant.
But I am not a professor. That is new for me and I will need months to understand it.

Have you ever done a repair of GPX series of minelab MDs? Maybe you can tell us how waveforms look like? What TX current slope looks like?.....

Thank you again.

[kt315]

read also https://en.wikipedia.org/wiki/Convolution

[green]

The numbers in the PNG file are what was required to balance California ground. Ground from the back yard balances very close to if not the same settings. I'll try to answer some of your other questions later, need to do some yard work.

Couple charts I posted in other threads. TRT_21, example of ground slope changing with Tx on time with constant current Tx vs constant rate Tx, each would require a different balance setting to cancel ground signal. target delay_1, some data taken awhile back showing detection distance and integrator out volts with some different targets. GEB on causes distance loss because noise increases when the 100usec GEB sample is added. I use a(Rx, two 8inch round connected figure eight)(Tx surrounds Rx)reduces noise and ground signal. Don't think it would have different timings vs mono coil. I did bury a target in back yard 10inches deep to see if integrator out voltage was the same as air with GEB on adjusted to cancel ground signal. Little or no difference.

[Qiaozhi]

Also, read this thread from post #69 onwards -> http://www.geotech1.com/forums/showt...ead#post208718
It describes a GB method proposed by Teleno that doesn't generate a hole in the detector's response.

[waltr]

All very good threads listed and what I read to learn how to GEB.

Also check my HH2 build thread from post #14.
http://www.geotech1.com/forums/showt...ake-on-the-HH2

I use the three sample method and a dual (plus/minus) integrator like most of the simpler PI detectors use.
I fix the first sample time and second delay time then adjust the second sample width to GEB.
My thread has discussion, links to other threads and scope pictures.

[belax2018]

Thank you waltr and Qiaozhi.
I have read all these threads before.

Will try soon method from Qiaozhi link. As it looks promising.
But I honestly disagree with assumption Teleno does: That for T1>T2 (where T1 and T2 is a TX length) target signal is the same

But anyway, does anybody know the exact method GPX implements for their GB.
I saw (youtube Woody DetectorMods) a GPX waveform.
It has 6 small peridos TX and one long T2. It looks like T2=6*T1.

I guess they use the following math.

If the signal from soil governs by: G=T/t - ln((T+t)/t)

So integrating it we have: Integral of G = IntG = (T+t)*ln(t/(t+T))

Lets consider two cases.
1) T=T1, Sample is from t1 to t2
2) T=T2=6*T1, Sample is from t3 to t4, where t3=6*t1 and t2=6*t2

For the 1'st case: G1=IntG (from t1 to t2) = (T1+t2)*ln(t2/(T1+t2)) - (T1+t1)*ln(t1/(T1+t1))
For the 2'nd case: G2=IntG (from 6*t1 to 6*t2) = (6*T1+6*t2)*ln(6*t2/(6*T1+6*t2)) - (6*T1+6*t1)*ln(6*t1/(6*T1+6*t1))= 6*((T1+t2)*ln(t2/(T1+t2)) - (T1+t1)*ln(t1/(T1+t1))) = 6*G1

So we have G2=6*G1

Now we can eliminate ground signal from T1 periods without a "hole" based on ground signal from T2 periods.

But it is OK here on paper. How it works in practice is a big question.

Does it really nobody knows how GB works in GPX series?

[waltr]

But anyway, does anybody know the exact method GPX implements for their GB.
I saw (youtube Woody DetectorMods) a GPX waveform.
It has 6 small peridos TX and one long T2. It looks like T2=6*T1.

Does it really nobody knows how GB works in GPX series?

Not sure if anyone really knows exact details of how GB is done on the GPX series.
The TX Wave forms you describe does sound like the way the SD2000 does GB. This is described in the MineLab patent US5506506.

Also Mine Lab has a list of patents per detector model on their web site:
https://www.minelab.com/usa/patents

These would be a good reads for how they may be doing GB. They also include other patents which are related so look these up.

Did notice that for the SDC, GPX & GPZ detector this patent is common.
https://patents.google.com/patent/US6636044

The SD2000 schematics are posted in this forum so a study will help show the weighing factors.
http://www.geotech1.com/forums/showt...000-schematics

The GPX & GPZ use a processor instead of analog circuits for integration and weighting which no one has the code for.

[kt315]

QED sent. check your box.