HV temporal sampling ?

[boilcoil]

That will work ...

OK, thanks,
When I have more free time, I can try it.

[Detectorist#1]

This is the timing of old good SD 2000 Minelab MD. Now is 21-st century! Let to try to do some new in this area!

[Detectorist#1]

Hi boilcoil,
It will be useful information how to implement in PI MD active samples for more real Fe/non-Fe discrimination. Maybe the samples during TX-ON pulse will be more useful for this discrimination. But in this case, I have a question - how to do ground balance in this case. Any MD without GB is not very useful in real searching. The voltage on the RX DD coil during TX pulse is very strong (linearly increasing voltage during all TX pulse if not exist saturation). In ITMD3 I haven't find the answer for this. Maybe it this case, the help of Carl will be useful for us.

[Op04]

One more thing to think is why ML put 240V fet on their gpz because in that case coil discharge current is lower vs if we put 500V fet

[Detectorist#1]

Hi Op04,
To answer your question, we need exact schematics of the TX part of the ML GPZ solution (if that's not secret yet).

[Op04]

Hi Op04,
To answer your question, we need exact schematics of the TX part of the ML GPZ solution (if that's not secret yet).

Unfortunately I don’t have it, I saw that fet on gpz board

[Detectorist#1]

Explanation for how ML GPZ idea works in the patent is too complicate and not clear to be useful for design of the TX part of GPZ7000.

[Op04]

Must check old post by Kev, daemon, bklein

[Detectorist#1]

Yes, I have some little parts of information - but not from sure sources. Last week I read post for end stage of GPZ TX stage and in this post was mentioned half-bridge end TX stage but without serious information.

[NigelLeeming]

After my last post about my sample time of 40microseconds, I began to look at a function called adc_continuous on the ESP32 in the hope of bringing forward sampling time and getting a few more readings per pulse. I'm pleased to report some progress despite adc_continuous's attempts to overload the few working brain cells I have left. Here is a sine wave from one adc input and the response from a square wave pulse of 100microseconds. I'm currently asking it to sample at 40kHz but I only get 2 maxed out data points where the pulse it, which makes me think it's a bit slower. The documentation says it samples up to 1MHz or so, which I'll believe when I see it.

[moodz]

Explanation for how ML GPZ idea works in the patent is too complicate and not clear to be useful for design of the TX part of GPZ7000.

I have met the guys who designed the GPZ7000 .... they spent arount 10 million AUD to develop it and went to a real lot of trouble to keep the current constant during the pulses ...
The module is shrouded in mystery .. because there is no mystery. In Australia we call that a Furphy .... its looks like one thing but is really something else and so the "copiers" go crazy trying to figure it out.
Keeping the current constant during TX pulses ( ie down to uA level ) is the key and that inlcludes compensating for EF as you swing the coil.
You cant patent the method of keeping a constant current pulse as it has already be been done ( eg barringer patent ) ... so the method is shrouded in security through obscurity.

Having said that though ... the GPZ has no discrim .. and this is a PIA as it finds every bit of buckshot and rusty wire thats out there.

[moodz]

This is the timing of old good SD 2000 Minelab MD. Now is 21-st century! Let to try to do some new in this area!

Boilcoil is on the right track .. the method I used and patented uses "interrupted TX" to obtain ground information whilst still utilising IQ information from the sine components.
Whereas that older method uses time domain multiplexing ( long pulses / short pulses ) I use "overlap multiplexing".

This system is superior to that used by the GPZ since it allows discrim as well.

eg below .... works with sine or rectangular.

[boilcoil]

Hello Detectorist#1,
Sorry for the delay, but I just sat down in front of the computer.
I see that Moodz has answered you very accurately, and has also given a lot of useful information.
So I'll just try to add something on your question.

...how to do ground balance in this case......

If you divide the detector operation into two channels - one for Sens, GB, EFE, etc. with samples taken by classical way, and second one, with active samples only for discrimination FE/COLOR, it is not difficult to make a ground balance.
But then there is another problem. With a balanced detector and a conductive target near the object against which it is grounded, the result of the active sample will be influenced by the ratio of two effects - decrease of the total magnetic resistance from the ironized ground object, and an increase in the magnetic resistance of the system, from the highly conductive target.
So in this case, proper discrimination is not guaranteed.
This the reason why I am looking for some way to add an active nuance to passive samples.

BTW I'm still a modest hobbyist, so it's quite possible that I can give the wrong answer to many of the questions or write some nonsense. I think it is best to ask questions to people who really understand metal detectors and the processes in them.

[Detectorist#1]

Hello boilcoil,
Thank you for your answer. I agree with you - we needs advises from more experienced people.

[boilcoil]

Boilcoil is on the right track ..

Hi Paul,

I've done a few experiments and there really is a distinct difference between ferro targets and conductive targets.
I determined what was the optimal ratio of the two pulses and the corresponding delays of the samples (as far as I could, of course), but...

But there is no goal, again a beam.

Samples from ferro targets are very close to the results of low-conductive targets (this could be assumed because in collapse time they are have a similar nature).
Later, maybe next month, when I really have enough free time, I will change a little of the detector code to look for a combination between these samples and samples after the long pulse (I think four additional samples will suffice).
If I manage to achieve some satisfactory results, I will post them on the forum.

Hey Paul, thank you for the help you provide and for the extremely useful things you share in the forum.
Thank you again.

Regards,
Vangel