How does ground balance work in GPX 4500?

[belax2018]

Thanks Mick.
At the end of long TX pulse of Normal mode I have expected to see more flat part of a current curve.

Thank you again.
I am still working on replicating these patterns you provided for us.

[Mechanic]

Rx pictures taken from front end output.
Current was measured from the 0.033r current sense resistor in the coil cct.
I have included the earliest sample in the trace for reference.
I didn't take direct coil voltage shots. Flyback is clamped at 180v.








Cheers Mick

[Mechanic]

Will do. I am away at the moment so either tomorrow or Wednesday as I need it off my bench.

Cheers

[belax2018]

So for each channel, unlike the sd2000 where the signal goes through the 10k resistor on its input to the integrator, there is 10k and 20k path to provide different levels of gain for the samples depending if the 10k or 20k or both are on together.

The cycle time is about 775us.
I will post scope shots of the tx voltage and current next week.

[ATTACH]44844[/ATTACH]
[ATTACH]44845[/ATTACH]

Mechanic, Please send us scopes of the TX coil/voltage next week or whatever time is convenient for you. That will make drawing the picture "GPX-4500" almost done.

[waltr]

Thanks for posting the timings. I saved these locally and now only need to analyze them.

[belax2018]

And ch2 for Normal and enhance.
I didn't bother with the SETA channel or Discrimination as it is not really relevant.
[ATTACH]44849[/ATTACH]
[ATTACH]44850[/ATTACH]

Cheers Mick

Great Job. Thank you.

You have moved all PI members of this forum to a new level.
All it left is to digest the waveforms you have just provided.
Thank you again.

[Mechanic]

And ch2 for Normal and enhance.
I didn't bother with the SETA channel or Discrimination as it is not really relevant.



Cheers Mick

[Mechanic]

So for each channel, unlike the sd2000 where the signal goes through the 10k resistor on its input to the integrator, there is 10k and 20k path to provide different levels of gain for the samples depending if the 10k or 20k or both are on together.

The cycle time is about 775us.
I will post scope shots of the tx voltage and current next week.

[Mechanic]

You will have to keep waiting!

It is a fair bit of work to extract the timings. I first had to figure out which logic lines were for what samples, which is done now.

Hopefully I will get enough time tomorrow to finish otherwise it will be next week.

Cheers Mick

[belax2018]

Very interested in seeing these.

Ok these are the timnigs I mentioned before.
Machine GPX-4500. Mode: Standard timing. Source: Woody DetectorMods.
All my comments on the picture and comments below need to be verified.

My guess: During three short samples GPX takes only target integration samples. During long TX pulse GPX takes only ground integration sample. It uses some kind of weighting coefficients to subtract Long_pulse_ground from Short_pulse_target.

Ground balance switch works as follows: Turn on the switch. You pump the coil up and down - GPX changes weighting coefficients until GPX sees no change in the signal. After that you switch off GB switch and start your prospecting.



P.S. Mechanic - we are all waiting for your signal analyzer waveforms. Thank you.

[Mechanic]

Thank you for answering directly to subject of the thread.

Could you please draw on paper timings of two modes of GPX-4500 - Enhanced and Standard? It is a confusion in my understanding of enhanced mode timings you explained above.

Below are my waveforms.

[ATTACH]44796[/ATTACH]

From the left to the right: Coil TX current, Flyback voltage, magnified bottom part of a flyback voltage. Ground contact of oscilloscope attached to Coil_VCC=15.5 V.
(R_d_coil=415 ohms. Coil current was measure across 0.3/4=0.075 ohm resistor. Coil voltage - along part of R_d_coil which is 84 ohms (415/84=4.95) )

I will get a logic analyzer hooked up in the next few days and post a couple of piccys.
There are 16 logic channels split between 5 channels, seta, ch1, ch2, gb, discrimination, -15v tx , -5v tx.

Cheers Mick

[belax2018]

http://www.geotech1.com/forums/attac...7&d=1545255530 The US Quarter has 3 times the signal in the 100usec ground sample as it does in the 10usec target sample. Don't know why 7A would be different than 1A other than coil current decay time, slightly longer delay time needed. Maybe someone could explain if it would effect.

You are right green. I think now I understand.

That is because I monitor the change of (short_average_of_signal - long_average_of_signal). If the difference is positive - I beep. If it is negative - I don't. In the case when target signal is higher in a ground region - my difference becomes negative - and I just ignore it. Need to rethink my detection scheme inside MCU.

[green]

It is simply too much of a target signal left in the region of 20-100 us (I confirmed it by just setting target time start to 20 us and target sample width to 80 us). I think because my coil current is around 7A - is way bigger than yours.

I haven't tried this type of GB with 1.0A current.

http://www.geotech1.com/forums/attac...7&d=1545255530 The US Quarter has 3 times the signal in the 100usec ground sample as it does in the 10usec target sample. Don't know why 7A would be different than 1A other than coil current decay time, slightly longer delay time needed. Maybe someone could explain if it would effect.

[waltr]

P.S.: I actually have found the oscilloscope waveforms for "standard" timings of GPX-4500. Will share soon. It is from one Australian man who does modding of Mlab machines.

Very interested in seeing these.

[belax2018]

Hi belax,

Have you found why you loose so much distance when ground balancing. Doesn't make sense to me with your timings unless targets are in GB hole.

It is simply too much of a target signal left in the region of 20-100 us (I confirmed it by just setting target time start to 20 us and target sample width to 80 us). I think because my coil current is around 7A - is way bigger than yours.

I haven't tried this type of GB with 1.0A current.

Currently I am changing my programs (FPGA and MCU) for a different methos of GB - two TX pulses - short and long - GB will be (I hope so) similar to QED/GPX.

P.S.: I actually have found the oscilloscope waveforms for "standard" timings of GPX-4500. Will share soon. It is from one Australian man who does modding of Mlab machines.