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

[moodz]

see this post .. the actual circuit ... https://www.geotech1.com/forums/foru...560#post408560
post #11

[lucifer]

Thanks. Is that what happens in practice too? I don't always trust simulations. Is your preamp's output nice and clean?

[moodz]

Hi moodz,
Is that loop worth it? Is it really reducing noise? Isn't more noise (generated by the DG411 switching + noise generated by NE5534) injected into the LM6171? I think I've seen this in other PI circuits and always wondered if it really does any good.

Hiya Lucifer ... refer the two circuits below ... identical except one amp is only referred to ground. There are white noise sources in each coil.

The top blue trace is the grounded amp. :-)

moodz

[lucifer]

Loop2 ... noise and offset reduction loop
The TX pulse is used as a control sample DG411 during the RX period to feed a noise / DC offset error correction via the NE5534 to the "ground" reference of the LM6171. This removes substantially all DC offset errors and low frequency mains / magnetic field influences etc in the signal.

[ATTACH]n409755[/ATTACH]

Hi moodz,
Is that loop worth it? Is it really reducing noise? Isn't more noise (generated by the DG411 switching + noise generated by NE5534) injected into the LM6171? I think I've seen this in other PI circuits and always wondered if it really does any good.

[moodz]

Looks great and a completed project.

Can you please post Timing of all the control signals and then a flow of the signal processing.

Also, the PIC32 code.

thanks

Thanks waltr ....

Attached is the source code ....

Theres stuff in there to do LCD / Serial and Rotary Encoder .... WIP.

The timing is set up for a 30us TX followed by a 30us RX. Why 30us ? ... because otherwise the TX fet will approach avalanche and the linear 5 volt reg will current limit and/or overheat.
The damping method will achieve the same timings with higher TX currents / voltages ... but obviously with a TX circuit and damping that can handle the higher currents and voltages.

.... built with NMOS / or SIC is the bees knees.

There are 2 feedback loops

Loop 1 ... the damping loop
The BSP230 mosfet performs the active damping and has a capacitor at its gate that stores the bias level fed by a sampling feedback loop. The sampling feedback loop takes a sample from the flyback error amplified by the preamp LM6171 ... this is the X sample.
Typically the X sample is taken just after the optimal flyback damping should hit 0 volts and 0 amps in the TX coil this the zero damping point ZDP ( use the SFR of the coil to get this ). The feedback loop sampling either can use a simple resistor ( eg 1k ) OR a peak diode detector & resistor ( eg 1k ) and leaky resistor ( eg 100k ).
If a simple resistor is used then the X sample can be set to say 100 nanoseconds at around at about 100 - 200 nanoseconds past the ZDP. If the diode peak sampling is used then the X sample window can be quite wide ( eg 10 us ) as the diode optimally samples the error peak.

Loop2 ... noise and offset reduction loop
The TX pulse is used as a control sample DG411 during the RX period to feed a noise / DC offset error correction via the NE5534 to the "ground" reference of the LM6171. This removes substantially all DC offset errors and low frequency mains / magnetic field influences etc in the signal.

Target Detection ...
Two samples DG411 drive the integrators feeding each input of the differential amp INA125 the output going the ADC on PIC32 via a simple low pass filter. As a baseline setting ... the RX time ( in this case 30 us ) is simply divided in half so that the first sample A is taken for 15 us .. then the second sample B is taken for 15 us where the total sample time is 30 us.
Surprisingly ... for a PI this "baseline" gives very good performance. This is the baseline setting for V3 with no controls fitted. ie basic detector. There is code in there to operate an LCD / Serial and Rot Enc.

Sound Generation ...
A single cycle of sine waveform is calculated and stored in an array at a particular volume. ( amplitude ) when the code boots. The cycle is played of an OC port pin as a PWM waveform by DMA using the repeating the single cycle sine over and over. The OC PWM timing is driven by clock timer 4. A DMA complete interrrupt is generated at the end every sine cycle "played" out.
The DMA interrupt handler basically reads the current averaged ADC target value and sets the frequency of timer 4 according to the value of the average target value. In this way a variable sine wave VCO is generated with relatively little CPU overhead. The volume can be adjusted by recalculating the amplitude of values stored in the single cycle array.

thats about it ..

PS ... the attached code is for a PIC32MX150 not 250 variant chip ... the MX150 is a little cheaper ( no USB ) ... but I dont use USB ... so etc .... ordered by mistake ... works the same otherwise.

PPS .. I use the v2.15 compiler ;-)

moodz

MAGPI008_2023_01_28_MX150.zip

[moodz]

Pcb as pdf for printed please

Here are your pdfs ....

MAGPIV3TOP.pdf
MAGPIV3BOTTOM.pdf


moodz

[moodz]

DO YOU HAVE A GERBER FİLE FOR Moodz' Awesome Gold Pulse Induction Version 3 - MAGPI V3 Project

gerbers and drill files are here ...

PRODUCTION_V3.zip

[waltr]

Looks great and a completed project.

Can you please post Timing of all the control signals and then a flow of the signal processing.

Also, the PIC32 code.

thanks

[samir3]

Pcb as pdf for printed please

DO YOU HAVE A GERBER FİLE FOR Moodz' Awesome Gold Pulse Induction Version 3 - MAGPI V3 Project

[moodz]

How this detector works, can you please describe it ?

For example here if you using PI detectors there are lots of mineral voices to rocks and stones. Are there any detectors non effective from the minerals ? Is this useable ?

Hi ... the detector is a pulse induction detector which are generally more resistant to mineralisation. This detector uses some ideas to increase the sensitivity to smaller targets and also to reduce noise.

It is mainly designed for conditions in Australia ( and may suit other places ) where there is a high degree of mineralisation in the ground but relatively little trash ( compared to Europe ).
The main target of interest in Australia is gold nuggets which forces the detector designers to detect smaller nuggets or deeper nuggets.

By using active damping the target sampling can be done earlier and if you are aware of how a pulse induction detector works then you can detect smaller metallic objects.
Noise reduction is also important particularly near power lines. I think this detector will work OK on the beach also .... I didnt find any gold during testing but did find alot of junk in a very short time.

moodz

[wtrk]

How this detector works, can you please describe it ?

For example here if you using PI detectors there are lots of mineral voices to rocks and stones. Are there any detectors non effective from the minerals ? Is this useable ?

[Mdtoday]

I like the way you have assembled it like that and got it out and tested in the field.
Having it out in the wild early is worth just as much as tons of time on the bench I think.
Will be interesting to see what you come up with re charging circuit 2 to 3 spare pins should do it if on chip comparator is available…

cheers

[moodz]

...the case fitted with a handmade 18650 battery pack ( 2 cells in series - 8.4 volts ) which is good for hours of detecting joy @ 230 mA draw.

I am just figuring out if I can make a QADCC ( quick and dirty charging function) with some spare ports on the CPU.

moodz

[moodz]

Nice work as usual Moodz.
Thanks for sharing.

Thanks ...

Below is the board installed in a shielded case

... yup shielded case is essential as the front end is sensitive to capacitive pickup and noise ...

The early X sample that controls the damping loop samples the "flyback" error which is largely dependant on the the value of the inductance ( 300 uH ) and capacitance of the coil ( ~ 200 pF ) so it only takes a few puff of capacitance to change the 1/(2pi root LC) SRF of the the coil. Hence the shield.

I used PCB inside a plastic case but a plastic case with copper EMI tape would suit.

moodz