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

[moodz]

I feel that 5v is used to supply power to the coil, its induction distance will be very low, I wonder if it can be increased to 12v, to increase the energy of the coil, 7805... 7909 chip itself is very noisy, must use low noise ldo,

Hi there Jeep ...

The power on this unit cannot be increased as it uses active damping that damps the coil in a bit over 1 microsecond ... if the TX voltage is increased then the coil current will increase and the flyback voltage will increase beyond the avalanche voltage of the TX mosfet.
If the mosfet avalanches the detector sensitivity will not increase.

The EMI and noise in the environment far exceeds the noise contribution of the linear regulators ... so this is not an issue ... If you want change them out ... if it does make a difference ( which I doubt ) then we can roll your suggestion back into the design.

The default unit with a 10 inch monocoil can detect a 0.1 gram gold nugget at 3 - 4 cm or a 10cm X 10cm foil target at 80cm plus. The default settings are good for beach detector and mild mineralisation.

moodz

[jeep6789]

I feel that 5v is used to supply power to the coil, its induction distance will be very low, I wonder if it can be increased to 12v, to increase the energy of the coil, 7805... 7909 chip itself is very noisy, must use low noise ldo,

[jeep6789]

Hello, everyone! Is this detector ready to use? I want to do one to probe, will it work better than delta? Thank you all

[moodz]

I did notice that it tended to damp out targets (like you say "anything that causes a change").
JLK

... hmmm actually now that I think about it its a motion detector .. prob not suitable as a probe.

moodz.

[KingJL]

... Additionally anything ( like the ground ) that causes a change ... even slightly of the L or C of the coil will cause the loop control to move and again warble or false targeting.

I did notice that it tended to damp out targets (like you say "anything that causes a change").
JLK

[moodz]

Paul, the V3 schematic that I have has R20 (sim R2) being 100k, not 1000k ,and connected across (in parallel with) D8 and R1, not connected to GND as in the sim. I will correct that and recheck. Thanks for the update.

JLK

Hi Jim ... there is a reason why the "auto track" option is not the top recommendation ( I use fixed sampling which will work for coils +/- 100 uH from the "standard" 300 uH ).

The resistor values will have to be set based on what range of coils need to be covered .... there are no magic values. Its not a good option for some who dont have a means of testing etc.

The autotrack causes loop "wobble" as the loop tries to stay on track. Additionally anything ( like the ground ) that causes a change ... even slightly of the L or C of the coil will cause the loop control to move and again warble or false targeting.

In principle it works but in practice it is not as good as fixed sampling for a known coil.

Slowing down the loop wobble with additional capacitance / filtering etc causes another problem.

It is easy enough to drive an interrupt to the CPU from the trailing flyback edge which will let the CPU set an accurate damping sampling pulse. I was going to do this but ... lazy ... found the fixed sample pulse worked for all coils that fell in the parameter window. Very easy to do the interrupt method though

moodz

[KingJL]

... but its straight off the V3. R1 R2 and D6 below ( R20 R1 and D8 on the V3 schematic )...

Paul, the V3 schematic that I have has R20 (sim R2) being 100k, not 1000k ,and connected across (in parallel with) D8 and R1, not connected to GND as in the sim. I will correct that and recheck. Thanks for the update.

JLK

[moodz]

Here is proposed a hardware circuit to automatically adjust the timing of the damping loop for idiosyncrasies of the TX coil(e.g. inductance, SRF, target load influence, etc). This circuit is implemented with 4 200V 15pF capacitors, 2 diodes, 4 resistors, a 56 pF capacitor, a 500 pF capacitor (both low voltage), a dual schmitt buffer IC and a quad NOR gate IC.


Referring to the attached partial schematic above: C4, C5, C6, and C8 form a voltage divider network to get the ~400V TX pulse down to something that the rest of the components can tolerate without significantly changing the characteristics of the TX coil circuit. I use 4 200v 15pF capacitors here as these are readily available in a 805 package. D8 is a clamp that presents the TX flyback pulse as a positive across R4. D6 limits the voltage to the 3.3V rail.

The logic circuitry: A1 is a schmitt trigger buffer that creates a nice digital logic signal that is the same width as the flyback pulse. A3 & A4 form a falling edge detector circuit (the RC network R5 & C14 provides a propagation delay to A4 so as to create a suitable pulse from the trailing edge). A6 & A7 comprise a one shot multi-vibrator with the one shot time determined by C10 & R13.

For the simulation, the Q output is fed to S1 (electronic switch in lieu of the voltage (V1)). In the actual circuit implementation, Q# will be fed to the DG411, U4B, pin8 (S1).

I will be posting results of sim runs for a 300 [email protected] ohm, 100pf with original timing control (V1), same coil with timing control via automatic hardware proposal, 800 [email protected] 0hm 150 pF with auto control, and 1500 [email protected] ohm 200 pF with auto control. These sims take a while to run as they are end-to-end complete implementation sims (TX, active damp, noise canceling AFE, integration filters and INA125. I am running these sims for a full second so that the system is well stabilized. The first set of sims are with a target coupling of 0.05.

I will later run a set with a higher level of coupling that shows some pretty interesting dynamics with strong target coupling (that the auto timing hardware option automatically adjusts to).

On edit, I should add: In case you want to implement a sim with the proposed circuit, all of the logic components (except the schmitt buffer) have the spice line parameters Vhigh=3.3 Td=10n, Trise=7n. The schmitt buffer has the parameters Vhigh-3.3, Td=4n, Trise=7n.

Hi Jim,
not quite the same as some of my earlier experiments but deja vu for me ... came up with all sorts of schemes to do what your circuit is doing..... The diode in the loop already serves as a peak detector ( so no exact gating required ) there is only a requirement to have it detect during a window in the RX cycle and the loop will automatically find the optimal coil detect point.

This is already included in the V3 board. It only requires the fitting of the diode and two resistors and a change to the sampling window in the damping loop ( eg make it 0 to 10 microseconds in the RX period ).

This circuit will auto track almost any coil.

Here is the sim circuit ... but its straight off the V3. R1 R2 and D6 below ( R20 R1 and D8 on the V3 schematic )

moodz

[KingJL]

Paul,
I am experimenting with a hardware solution ... for generating the damping loop sample pulse that automatically adjusts the timing of the sample pulse to the coil being used...

Here is proposed a hardware circuit to automatically adjust the timing of the damping loop for idiosyncrasies of the TX coil(e.g. inductance, SRF, target load influence, etc). This circuit is implemented with 4 200V 15pF capacitors, 2 diodes, 4 resistors, a 56 pF capacitor, a 500 pF capacitor (both low voltage), a dual schmitt buffer IC and a quad NOR gate IC.


Referring to the attached partial schematic above: C4, C5, C6, and C8 form a voltage divider network to get the ~400V TX pulse down to something that the rest of the components can tolerate without significantly changing the characteristics of the TX coil circuit. I use 4 200v 15pF capacitors here as these are readily available in a 805 package. D8 is a clamp that presents the TX flyback pulse as a positive across R4. D6 limits the voltage to the 3.3V rail.

The logic circuitry: A1 is a schmitt trigger buffer that creates a nice digital logic signal that is the same width as the flyback pulse. A3 & A4 form a falling edge detector circuit (the RC network R5 & C14 provides a propagation delay to A4 so as to create a suitable pulse from the trailing edge). A6 & A7 comprise a one shot multi-vibrator with the one shot time determined by C10 & R13.

For the simulation, the Q output is fed to S1 (electronic switch in lieu of the voltage (V1)). In the actual circuit implementation, Q# will be fed to the DG411, U4B, pin8 (S1).

I will be posting results of sim runs for a 300 [email protected] ohm, 100pf with original timing control (V1), same coil with timing control via automatic hardware proposal, 800 [email protected] 0hm 150 pF with auto control, and 1500 [email protected] ohm 200 pF with auto control. These sims take a while to run as they are end-to-end complete implementation sims (TX, active damp, noise canceling AFE, integration filters and INA125. I am running these sims for a full second so that the system is well stabilized. The first set of sims are with a target coupling of 0.05.

I will later run a set with a higher level of coupling that shows some pretty interesting dynamics with strong target coupling (that the auto timing hardware option automatically adjusts to).

On edit, I should add: In case you want to implement a sim with the proposed circuit, all of the logic components (except the schmitt buffer) have the spice line parameters Vhigh=3.3 Td=10n, Trise=7n. The schmitt buffer has the parameters Vhigh-3.3, Td=4n, Trise=7n.

[moodz]

Paul,
I am experimenting with a hardware solution (4 capacitors, 2 diodes, and 6 NOR gates) for generating the damping loop sample pulse that automatically adjusts the timing of the sample pulse to the coil being used. Initial Checking out with [email protected] ohm, [email protected] ohm, 1500uh@4 ohm. Initial indications are looking good. The damping loop sample width, of course, remains constant. Is there any need you see to have the sample pulse width variable? Will post solution after a bit more testing.

Good work ... the sample width seems to affect loop response speed but also how noisy it is ( eg jitter ). I just set it for generic monocoils around the 300uh mark ... its pretty forgiving.

For tracking almost any coil I used a single diode peak detector with a "leaky" resistor across the damping gate cap.

moodz

[KingJL]

Paul,
I am experimenting with a hardware solution (4 capacitors, 2 diodes, and 6 NOR gates) for generating the damping loop sample pulse that automatically adjusts the timing of the sample pulse to the coil being used. Initial Checking out with [email protected] ohm, [email protected] ohm, 1500uh@4 ohm. Initial indications are looking good. The damping loop sample width, of course, remains constant. Is there any need you see to have the sample pulse width variable? Will post solution after a bit more testing.

[KingJL]

... You could use 10uF ... for a sharper response esp if you are desiging a probe...

That's what I was thinking... the total sim kinda' indicated that! Thank you for the response!!

[moodz]

100 uF Jim ...

Lowpass filter corner frequency of 1.59 Hz ...

You could use 10uF ... for a sharper response esp if you are desiging a probe.

moodz.

[KingJL]

Thanks Jim .. the circuit is simple enough and could be further trimmed to make a compact PP. I have used a 300 uH ferrite inductor and it works great.

Here is a modified schematic to clarify ... the PCB still works ok. ( ie not modified ). The diodes are rearranged from a inverter to an inverting doubler to cater for lower lithium voltage input ( 8V instead of 12 volts ).

Paul, Is C8 100 uF or 10 uF... just checking?

[Rick S]

I am running ubuntu ... then a virtual windows machine under virtualbox ... where I run the IDE.
You will get this error message if your target is not powered ... you mention you have your power set at 3.25 volts and power is ck'd off ( do you mean the box is checked yes or not checked no ).

The other way this will fail is if your target is draining too much power ( this was my case ) ...and the pickit 3 cannot supply sufficient current.
So for reliable programming ... I have the pickit 3 power box option unchecked and the target board is powered - not by the picket 3.

Also I seem to remember the logic gets confused on linux ... and this is the fix.

moodz

Thanks for the help. In IPE settings/advance/power settings/voltage options "power target circuit from..." then the check box. With the check present when connecting with pickit 3 "unable to connect to target device failed ID" error
message. With the check not seen the error notice no voltage at vdd is given. So, what I am understanding is to have a separate power supply(3.3v) to the PIC and not from pickit 3. I think there are 3 pins to connect to the 3.3v. Would
you have a schematic or site to visit how this is done without frying the PIC? Also, I measured the voltage at pin 2 and 3 of pickit 3 and read about 1v. Not sure what that means. PIC taking to much or not enough voltage?


Ricks