Moodz' Awesome Gold Pulse Induction Version 3 - MAGPI V3 Project

[Gunghouk]

Nice. How does the hardware compare with the mods we've been discussing eg. the steeper low pass front end hardware, new op-amps or TX fundamentals?

[moodz]

Added target modulated volume control TMVC. The MAGPI uses a software VCO that responds to target voltage. The only problem with that is you have to listen to the tone when there is no target. When there is a target the tone frequency rises proportionaly to the target signal strenth the volume stays the same.

The TMVC now drops the no target volume to around 5% .... when a target is near the coil the volume will increase proportional to the target signal strength in addition to the VCO tone changing.

Much nicer ... like changing from a B&W TV to color.

Heres the video

https://youtu.be/X7eVNNC9yq4?si=BFzb0bdBHND8VilW

[moodz]

I guess in air the frequency change has no discernible effect with the small target but in ground who knows. Was the 1.1m range for the 10x10 target unchanged aswell?

The previous range was just on 1 meter ... the range increased by 10 cm to 1.1 meters at 20 Khz. ( up from 16 Khz )

I have also been considering the reason for this and its related to the maths behind PI target detection which is proportional to SAMPLERATE x Integral ( K X 10-6 )

Where K is a value dependant on coil inductance / area / current / target size / target decay time etc etc.

The main thing is the 10-6 ... the target response is a 6th power down.

Now if we log that or voltage we get 20log(10-6) = -120db. WIth the latest change to the preamp the preamp gain the overall system gain in the MAGPI is now +135db.
Because this exceeds the target path loss from TX to RX we now have above break even target detection. ( break even means the target signal = the noise floor in the amplified demodulated target signal)

Generally speaking unless the ground is highly reactive the ground depth wil exceed the air tests due to halo effect.

[Gunghouk]

I guess in air the frequency change has no discernible effect with the small target but in ground who knows. My understanding is that lower frequencies propagate through ground better anyway though I suspect the optimum frequency will vary depending on multiple variables the major one being soil type/composition.

Was the 1.1m range for the 10x10 target unchanged aswell?

My bare bones 26 cm dia. homemade coil is 320uH and 1 Ohm so it should perform reasonably well.

[moodz]

Target saturation ? You could try reducing the pulsewidth until the sensitivity noticably drops then crank back to where the sensitivity is 'normal' then alter the freq for that PW. Maybe there's a sweetspot for PW and/or freq for that target size/material in air/ground.

Of course it's going to be different for every coil !?

BTW whats the lower freq breakpoint for your RX amp, higher than the 1kHz TX freq ?

I tried your suggestion and the highest stable TX frequency I got was 25 Khz ... but I backed it off to 20.3 Khz and the range on the 10cm x 10cm foil was 1.1 meters ( more than 2.5 coil diameters ). The sensitivity to the 10 x 8 mm foil remained the same ( approx 20 cm )

The coil is a Minelab commander mono 300 uH 0.5 ohm.

[moodz]

In the meantime JLKing has made the following suggested change to improve the preamp low frequency rolloff.
A second order filter in the feedback element. Requires minimal components.
Thanks JLKing !

Picture says it all.

[moodz]

Reduce F to 500Hz and the harmonic 'nit' comb gets finer

Its just the mathematics of time domain pulses to frequency domain.

For instance here is a spectrum from DC to 1 GHz. (In the time domain its a single pulse of 1 million volts / amps that is 1 nanosecond wide )
The individual spectral peaks cant be seen as they occur every 1 nanohertz so they merge to form a continuous line.

However no matter how "exciting" these spectrums look for metal detecting there is no time domain in FFT and nuggets are not frequency selective or they would be used in radios

[Gunghouk]

Reduce F to 500Hz and the harmonic 'nit' comb gets finer

[Gunghouk]

Yes that would seem to be the ideal. What are the conditions here ?

[moodz]

Like this ?

[Gunghouk]

Yeah once you get to 50% duty cycle it's the fundamental and only the odd harmonics by amplitude ratio 1/3 + 1/5 etc. Once again is there an optimum duty cycle where the power & harmonic distribution is evenly spread ?

[moodz]

Target saturation ? You could try reducing the pulsewidth until the sensitivity noticably drops then crank back to where the sensitivity is 'normal' then alter the freq for that PW. Maybe there's a sweetspot for PW and/or freq for that target size/material in air/ground.

Of course it's going to be different for every coil !?

BTW whats the lower freq breakpoint for your RX amp, higher than the 1kHz TX freq ?

No I meant coil saturation as in the LM7805 current limiting kicking in. But then again maybe not I just checked it and the limit is around 2.4 amps.

What is interesting is that despite the coil current change there was not much change in sensitivity ... just response time. Not enough testing though.
The lag in response time would be due to less samples per second moving from 14 Khz to 1 Khz. - filters take longer to charge.

The breakpoint is 10 Khz so about 20 db down at 1 Khz ... but all the harmonics have the same target info modulated onto them and the 1 Khz signal has more harmonics as they are spaced at 1 Khz instead of 14 Khz in the passband.


spectrum of 14 khz square wave vs 100u / 1000u tx

[Gunghouk]

Target saturation ? You could try reducing the pulsewidth until the sensitivity noticably drops then crank back to where the sensitivity is 'normal' then alter the freq for that PW. Maybe there's a sweetspot for PW and/or freq for that target size/material in air/ground.

Of course it's going to be different for every coil !?

BTW whats the lower freq breakpoint for your RX amp, higher than the 1kHz TX freq ?

[moodz]

That would be great ... I noticed that the detector seems to have more lag at the lower TX frequency even though the TX current has tripled ... the sensitivity has not increased.
But it is still detecting the 1 microsecond target.

[Gunghouk]

I would love to bury a sensor and see what amplitude of signal it actually picks up for varying freq & pulse width. The results would make a great waterfall diagram and FFT the results also.