PI metal detector for really small nuggets

[eclipse]

Lately I'm thinking to visit some local creeks in search of really small nuggets.
So I'm putting some specs of a PI metal detector here that I may construct to help me in this particular quest, so what do you think:

Pulse width: 60uS
Mono coil: 150u-220uH (20 centimeters)
Power: 10-12V power pack size doesn't matter I get lazy quickly
Frequency: ~10Khz
GB: maybe, I don't know what the mineralization in the area is
Sample width: 10uS
Sample Delay (S1) 5-6uS

Bipolar (maybe ?) - this will help me bring the frequency very high, probably +20% higher (12Khz)

Mosfet with low Coss. Flyback voltage - 150V-200V

Low capacitance shielding (nice shielding spacer).

Maybe try to put the preamp in the coil to avoid cable capacitance?

[green]

What do you consider really small nuggets? https://www.geotech1.com/forums/atta...2&d=1547586775 some I charted awhile back. Between 1 and 2us TC.

[eclipse]

Well I don't know how small but probably the only ones I can find will be tiny nuggets and that's my main focus.
I have bunch of metal detectors that I can use but it's definitely more fun with a DIY and I'll focus on really small - for example 2uS-3uS TC for this DIY PI.

[waltr]

I think you are on the right track with the parameters stated.

Since the target TC will be very short you do need to ensure coil decay and pre-amp settling are in the 1-2usec range.

You probably could decrease the TX pulse time to 5Tau or 15-20usec.

Here is some forum posts I collected on this:

There is another I can not find. It is by Ferric Toes about his design of a PI detector to detect very small broke needle tips. Maybe a search can find this.

[Carl-NC]

A preamp in the coil won't help unless you have a separate RX coil. Even then, the cable C is still affecting the TX settling so the separate RX coil will also need to be IB.

I would suggest bipolar to make high pulse rates easier, and maybe less pulse width to keep power consumption down. Pay tremendous attention in building the coil, that's where 90% of the payback is at.

[surfdetector]

@ Waltr
Thank you for sharing your Faster Sampling document. The collection of information is extremely educational.

[Ferric Toes]

I looked up some of the information I have on the development of small object PI detectors in 1996.

As already mentioned the coil is very important. I made several different types of coil i.e. ferrite cored, printed circuit, and standard wire. The details of a standard wire coil (this was a rectangular figure eight with overall dimensions of 160 x 220mm. The winding were 2 off 80mm x 220mm side by side. Each had an inductance of 30uH and a resistance of 2.3ohms. Wire was 30AWG wire wrapping wire with Kynar insulation. Screening plates were made with 1mm Tufnol sheet painted with two coats of conductive paint. RG179 coax was used and the maximum length possible was 1m. 0.5m was used in practice. I also made a 100mm diameter round coil that worked OK.

TX pulse width was very short at 7.5uS at a very low pulse current (39ohms in series with coil) but very high rep rate. Preamp IC was OPA620 with a gain of 25. Today I would use a OPA1611. Standard diferential integrator and following circuit.

Preamp response was clean and sampling started at 2.5uS with a sample width of 2uS. You can just see the glitches in the response at 2.5 and 4.5uS at the start and end of the sample.

Eric.

[eclipse]

Thanks folks!
Eric I have one of your own - Goldquest metal detector.

I've updated specs sheet. Right now the current draft is:

15uS TX pulse width
DD COIL: TX (50uH) / RX (50uH) ~20centimeters
Power: 10-14V
Frequency: 20kHz
GB: no
Sample width: 3uS
Sample delay: 2uS
Mosfet with low Coss, flyback 60V-120V
Preamp, MCU, integrators - built in coil (I will use larger coil housing to put the miniature board directly in the coil with a reasonably large distance from the wire bundles)
Bipolar pulse; Center Tapped RX
Resistor parallel with mosfet 10R-40R

[Ferric Toes]

Thanks folks!
Eric I have one of your own - Goldquest metal detector.

I've updated specs sheet. Right now the current draft is:

15uS TX pulse width
DD COIL: TX (50uH) / RX (50uH) ~20centimeters
Power: 10-14V
Frequency: 20kHz
GB: no
Sample width: 3uS
Sample delay: 2uS
Mosfet with low Coss, flyback 60V-120V
Preamp, MCU, integrators - built in coil (I will use larger coil housing to put the miniature board directly in the coil with a reasonably large distance from the wire bundles)
Bipolar pulse; Center Tapped RX
Resistor parallel with mosfet 10R-40R

What model Goldquest do you have? Does it have a 4017 decade counter on the pcb and a IRF112 or 113 as the Mosfet? If so, that is the basic arrangement for a detector with a short delay. The clock frequency control alters TX width, delay, sample pulse width and pulse rate all with one control. If you want longer pulses, all changes in proportion and current draw stays the same.

It would be good to check the area you want to search with a standard PI to see if the ground suffers from iron mineralisation. If it is, then very short delays and short pulses will still be a problem.

If I were making such a detector, I would house the TX with reservoir capacitor and preamp in a small waterproof case about 30cm up the shaft from the coil. Supplies, pulse and sample generators, display, audio etc in separate box with controls at the top of the shaft with a multicore cable feed to the TX/RX box.

Here is the TX/RX and integrator circuit I used 24 years ago with pseudo figure eight coil. Single mono coil could run from just one Mosfet.

Eric.

[Ferric Toes]

A preamp in the coil won't help unless you have a separate RX coil. Even then, the cable C is still affecting the TX settling so the separate RX coil will also need to be IB.

I would suggest bipolar to make high pulse rates easier, and maybe less pulse width to keep power consumption down. Pay tremendous attention in building the coil, that's where 90% of the payback is at.

Yes, in any new design I would go bipolar pulses now. I was reluctant in the past because it was almost impossible to get P channel Mosfets to match N channel ones. Bipolar has the benefit of 'natural' EF cancellation and power line noise too.

Getting coil shielding right is difficult. To much destroys the abillity to use short delays and too little brings in hand capacity and wet ground effect.

Eric.

[6666]

What model Goldquest do you have? Does it have a 4017 decade counter on the pcb and a IRF112 or 113 as the Mosfet? If so, that is the basic arrangement for a detector with a short delay. The clock frequency control alters TX width, delay, sample pulse width and pulse rate all with one control. If you want longer pulses, all changes in proportion and current draw stays the same.

It would be good to check the area you want to search with a standard PI to see if the ground suffers from iron mineralisation. If it is, then very short delays and short pulses will still be a problem.

If I were making such a detector, I would house the TX with reservoir capacitor and preamp in a small waterproof case about 30cm up the shaft from the coil. Supplies, pulse and sample generators, display, audio etc in separate box with controls at the top of the shaft with a multicore cable feed to the TX/RX box.

Here is the TX/RX and integrator circuit I used 24 years ago with pseudo figure eight coil. Single mono coil could run from just one Mosfet. [ATTACH]50852[/ATTACH]

Eric.

Thats very smart Eric, do you recall what voltage you would set TP1 to was it zero volts ?

[Elliot]

.... Here is some forum posts I collected on this:.....

Can you please clarify point 4 of the Faster Sampling attachment: Where is the recommended placement of the 56k resistor - ground at one end, and the other end is where ?

[waltr]

Can you please clarify point 4 of the Faster Sampling attachment: Where is the recommended placement of the 56k resistor - ground at one end, and the other end is where ?

That post was originally by Carl.
The 56k resistor should be from the MOSFET drain (and the HV diode) to system ground (or maybe should be across MOSFET's Drain to Source).
It is needed since the diode blocks discharging the MOSFET's Drain charge.

[eclipse]

What model Goldquest do you have? Does it have a 4017 decade counter on the pcb and a IRF112 or 113 as the Mosfet? [ATTACH]50852[/ATTACH]

Eric.

The board label says "PPD3 Issue 2" it has the IRF740 mosfet and the board looks alot like Goldscan 4 stuffing.

[Ferric Toes]

Thats very smart Eric, do you recall what voltage you would set TP1 to was it zero volts ?

Yes, that is correct. I will have to look at a pcb to see if there was an offset trimmer. Make take a while to find a board as they are in a box somewhere. The trimmer shown injected an attenuated 1st sample pulse to back off surrounding metal in one application. I seem to remember selecting IC's that needed little if any trimming.

Eric.