LOOKING FOR GOLDSCAN 4 Schematic.

[ivconic]

Only things that I have changed from the last time:
1) PPS rate now is 1000 Hz.
2) IRF840 is now with heatsink
3) Damper is now 470/5W
4) previous 2R2 is now also 5W.

Heatsink is lightly warm ... but far from being hot.
470/5W is super cool, previous 2x1k/1W were warm to hot after 10-20 minutes, same as previous 2R2/2W.
Could it be that power consumption dropped so significantly (from 220mA to rock solid 150mA) just because I put much "stronger" resistors which are now totally cool?
This is INTGRIGUE!
Quite opposite from what is to be expected when changing PPS from 500 to 1000Hz... or am I wrong?

[ivconic]

And of course, as an aid in understanding (experts don't need this) timing operation:

[ivconic]

Pito, now let me answer your frequent intrusions regarding the implementation of mcu, which would replace a lot of "hardware" on the detector.
Now that we have a timing "scheme" and a proper schedule of when what works; nothing easier than writing mcu code that will do it without problems.
But that's not the point. The point is to go slower (and the harder way); in order to fully understand the entire process that takes place in such a detector.
Only when we have a full understanding of the processes taking place there; then we can play around and convert the hardware piece by piece into code.
Unfortunately my favorite Atmega328P doesn't have enough resources to generate 4 pulses on 4 separate pins and do it all flawlessly accurately.
By the most stupid method "digitalWrite pin..." and so one after another all 4 pins... is a child's toy, which will not work well.
Because the final goal here is the shortest possible sample delay and the most precise timing.
A single C function can take too much time for such tasks.
Direct port/pin manipulation with (or without) interrupts can be up to 12 times faster.
But still not enough for such a delicate task.
Ok, that's all great when we make a detector that is a child's toy with a delay greater than 50uS and timings that will have a smaller or larger jitter at each clock.
The internet is full of such attempts. And on this forum you have "a dozen" attempts in that style.
It's all child's play, good for educational purposes, elementary education and entry into the world of programming.
But if I want a PI detector with maximum 10uS sample delay and very precise timing; it takes much more than that.
Perhaps something more could be extracted from Atmega at the lowest level, using pure assembler. But even then it wouldn't be good.
Ok, STM32, ARM... that's a better story.
And that can be done.
But we are far from that, this detector has not yet shone with its full glory. There are still a lot of tweaks to come until then.
I will always have a good laugh at one of Carl's "free thoughts" that he unwittingly and unintentionally wrote... and it has something to do with this!
When he was explaining on another topic how easy it is to solve this kind of task; he said (I'll paraphrase, it's not verbatim): "I'll choose a mcu that has at least 8 to 16 timers and that's it..."




Carl the GENIUS!

[Detectorist#1]

The reduced range is due to mixing two channels. = instead of mixing we can calculate the magnitude and phase using stm32

Changing of the tool not change the mats of the process!

[ivconic]

What phase?

[Detectorist#1]

Hi ivconic,

Yes, my question to pito is also " what phase" in PI MD.

[Krzysztof]

ATMega328BP - this one letter "B" gives you an additional 2 x 16b timers with full external access.
Pololu has it on its modules - Extra
!

[pito]

Changing of the tool not change the mats of the process!=
summing
- 5V + 5 V = 0V
magnitude
√ (-5)² + (5)² = 7V

[pito]

" what phase" in PI MD. =
You have two channels in GOLDSCAN 4, so there will be a phase shift between them,.....
Maybe that can be used for discrimination???

[Detectorist#1]

Hi pito,

If using of two channels will gives possibility to stable discrimination in PI MD, do you think that this is not used by Minilab engineers?

[pito]

not used by Minilab engineers? = no idea, I know that phase shift in IB detectors is used for VDI calculation, why not apply that in PI ?

[Detectorist#1]

Hi pito,

If the real world is dual - analog and digital - and metal detecting use VLF and PI - our hope is that have to be way to make stable discrimination in PI MD also. Our attempts continues!

[Detectorist#1]

Hi Krzysztof,

Do you really thing that using of more powerful ( with more timers) 32bit uP for generating of of 6-8 pulses with simple relationships will solve the fundamental problems of PI MD?

[Detectorist#1]

Hi pito,

Yes, phase shift is used in hybrid MD in last years in attempt to combine PI and VLF principle in metal detecting. Good primer is Voodoo project from Carl. This idea is copied from some other hybrid detectors in geotecg1 site.
This idea works, but my hope is that have to be possible to make pure PI with stable discrimination.

[Krzysztof]

When he was explaining on another topic how easy it is to solve this kind of task; he said (I'll paraphrase, it's not verbatim): "I'll choose a mcu that has at least 8 to 16 timers and that's it..."

8b uP ATMega328BP -> 3 x timers 16b (T1,T3,T4)