Automatic ground tracking and auto-tuning

[JoyJo]

Greetings, dear ones! I ask for advice in the implementation of the function of ground tracking and auto-tuning.
Now we are trying to run such an algorithm in the metal detector (stalker ib):
1. We assume that the ground signal is near the frequency of 3 Hz.
2.The received signal passes through digital high-pass filters (5 Hz cutoff frequency) and through a low-pass filter (1-2 Hz cutoff frequency). It turns out that the signal passes through a bandpass filter.
3. At the moments of the maximum signal (the maximum is monitored by the microcontroller), the calculation of the soil VDI is performed (arctg Y/X)
4. After that, we add 90 degrees to the VDI of the soil (VDI of ferrite) and get the angle of the soil.
5. The data obtained from point 4, in fact the angle of the ground, we take into account when turning the axes X & Y. To achieve the absence of a ground signal in the channel Y.

But this approach works crookedly. But, to be more precise, it does not work on the ground))
With this method, we have the following disadvantages:
а. Big iron targets (the type of which have a "minus" sign) lead to a greater displacement of the ground angle: 20-30
b. when the axes of the X&Y channels are rotated by at least 0.1 degrees, false alarms appear.

Can you recommend any articles or a book to read in more detail on this topic?
How can this algorithm be implemented optimally?

[pito]

0.1 drg ? It is your VDI so accurate ?

I think you should start from checking accuracy of phase detectors, make a pendulum from a coin on a string and check if the numbers are constant and from what distance.
3 and 4 are not clear, in 3 you have soil VDI = angle of soil, make this as reference number = Ref, when you have signal from target you have target VDI total = VDI Ref + VDI target, from there you have VDI target = VDI total - VDI Ref.
4 - ferrite ?

[Carl-NC]

I can't think of any literature or even patents that cover auto tracking. It seems to be mostly a trade secret with companies.

Ground phase is usually limited to 0-15°. So with any response outside that range (like iron) you want to throw that data out. Next, ground does not usually change by a large amount over a small area so you also want to limit the speed at which you change the track phase. It might be useful to impose a track window on the algorithm, which limits how much phase change you allow on the incoming data. For example, suppose the ground phase is currently tracking at 5°. You impose a limit of ±2° so that any incoming data beyond that range is thrown out. As the tracking algorithm alters the GB phase, you also alter the track window so that it's always centered on the current GB phase. This will effectively limit track speed (widen the window to increase speed) and it also throws out any potentially bad data.

Another thing to consider is to create a separate G channel, independent of the X & R channels. The G channel is your ground-free target channel and X & R never get rotated, therefore target phases remain accurate. You can see a G channel in many VLF schematics, sometimes created with a separate demod channel and sometimes by a simple addition of the X & R signals.

[JoyJo]

Thank you for your help. Let's try to implement this. I will unsubscribe about the result.

[JoyJo]

And in general, for tracking the ground and to compensate / exclude the influence of the ground, will multi frequency help: simultaneous transmission of no more than 3 frequenci? Do I understand that based on data at 3 different frequencies, it is possible to obtain more reliable information about the magnetic susceptibility of the soil. As far as I can understand, it is the magnetic viscosity that has a significant effect on the response rather than the conductivity of the soil. Or am I wrong?
As it were, the greater the number of transmitted frequencies, the more accurately it is possible to estimate the soil parameter and more "qualitatively" exclude it from the print signal, but the transmission of a large number of frequencies should have some difficulties. For example, a decrease in the strength of the transmitted field with an increase in the frequency of the TX signal.

[JoyJo]

I found a similar document. I'll study it, maybe something will become clear.

[pito]

Pay attention to page 3 and 4 = I was thinking to use it for ground balance problem, I did some testings and for target was 100ms and for ground 200ms.




in my tests ; 0 = violet line = ground, rest lines are from target