VLF driving waveforms

[permute]

I've finally gotten around to starting my hobby metal detector design. I'm really more interested in the design and construction aspects than the actual outdoor all day in the summer digging into dry dirt aspect. I'm trying to get some idea on the scale of some things, as well as why some of my ideas will either work, not work, or etc...

I've been debating between PI/VLF mainly, but I do wonder if there isn't a way to modernize the BFO concept.

For the VLF front, I understand that the idea is to transmit a waveform that has generally lower frequency content in the 10-50khz range. A second coil is placed in what is nominally a dead zone for the tx coil. It only receives a signal when this balance is disrupted. My main questions are:
1.) for a reasonable tx coil, such as a 20 cm diameter, what is a good range for the number of turns and for the approximate load impedance when driven? I know it would depend on other factors, but what would some low/high examples?
2.) What would a reasonable drive level be for the coil if the goal was 30cm? (or maybe 20 cm, in the area where this will actually be used the amount of digging that can be done unassisted is fairly small). The location would be in the Arizona desert, USA. I'm not sure on the mineralization of the soil.
3.) What is a reasonable rx coil and rx signal strength?
4.) What about bandwidth? can the same rx/tx pair work equally well at 10khz and 40khz? or would it be expected to work well only at one or the other with near zero output away from resonance?
5.) Finally, what is the preferred driver? Something like an H-Bridge where the coil is switched between 2 (or 3) discrete levels, or something like a linear amplifier, where the output is much more linear?


For BFO-inspired ideas, I understand that BFO works by using an oscillator that is pulled when the inductance/resistance/capacitance of the coil changes due to the presence of various metals.
6.) Would this idea also work with a single tx/rx coil where a voltage is driven onto the coil and a current is measured?

My thought there is that it might be possible to make a wideband detector by measuring the coil impedance over a wide bandwidth, and extracting some of the parameters. This would probably linear amplifiers for the driver and current amplifier.

[Carl-NC]

1.) for a reasonable tx coil, such as a 20 cm diameter, what is a good range for the number of turns and for the approximate load impedance when driven? I know it would depend on other factors, but what would some low/high examples?

Most detectors have a TX inductance between 0.5mH and 6mH. See the chart here. Transmit frequency is a factor on choosing the inductance.

2.) What would a reasonable drive level be for the coil if the goal was 30cm? (or maybe 20 cm, in the area where this will actually be used the amount of digging that can be done unassisted is fairly small). The location would be in the Arizona desert, USA. I'm not sure on the mineralization of the soil.

Usually the drive level is as high as possible, limited by practical power supply design, power consumption, and the ability to maintain enough nulling, which can depend on mineralization.

3.) What is a reasonable rx coil and rx signal strength?

Low enough not to saturate the circuitry, with headroom left over detection. Which depends on the circuit design.

4.) What about bandwidth? can the same rx/tx pair work equally well at 10khz and 40khz? or would it be expected to work well only at one or the other with near zero output away from resonance?

You can design for narrow band or wide band. The latter tends to be noisier. If you design a coil for 40kHz then getting it to work at 10kHz is not too difficult, but power consumption will be higher.

5.) Finally, what is the preferred driver? Something like an H-Bridge where the coil is switched between 2 (or 3) discrete levels, or something like a linear amplifier, where the output is much more linear?

Depends on whether you want narrow-band efficiency or wide-band flexibility.

For BFO-inspired ideas, I understand that BFO works by using an oscillator that is pulled when the inductance/resistance/capacitance of the coil changes due to the presence of various metals.
6.) Would this idea also work with a single tx/rx coil where a voltage is driven onto the coil and a current is measured?

This is how a loaded-loop design works, which is a variation of BFO. Some pinpointers use this method.

Your questions are good ones, but are too general for anything better than vague answers. There are many many design trade-offs, and no one best approach.

- Carl

[permute]

Thanks. Like I've said I am still a bit in the research phase. I'm looking at using either a beagle-bone or a de0-nano for the core processing platform. More likely the ARM based beagle bone. A secondary goal is to gain some experience with ARM-based designs.

See the chart here.

The info on coils is very helpful. I'm basically trying to determine if I need 5V power, +-5V, +12V, etc... and what amount of current. I'm not sure if the coils are normally driven with, say 2W or 1W, or 0.1W or etc... Just trying to find a ballpark number that isn't way too high or way too low. If I had an estimate of tx power, I'd be able to determine what my options are for the supply voltage.

Usually the drive level is as high as possible

Low enough not to saturate the circuitry

For #2/#3, I guess a better question would be what the dynamic range would be -- eg, a "nulled" coil would pick up ? much of the Tx signal vs ? for ground mineralized vs ? for a steel plate/metal ring. This would give me a ballpark gain-bandwidth product for the front end processing.

Depends on whether you want narrow-band efficiency or wide-band flexibility.

Because of my personal interest, I'm looking more into wideband designs. I've been wondering about using an H-bridge for 3PAM (+V,0,-V) with a chipping sequence to generate the wide bandwidth signal. Perhaps with the addition of an output filter to limit the harmonic content transmitted to a reasonable level. However, if the drive strength didn't need to be excessive to meet a 1-3hour usage, I could implement a linear amplifier+DAC that would allow for more flexibility, at the cost of more losses.

This is how a loaded-loop design works,

A follow on to #6, is there any advantage to sensing Rx current as opposed to Rx voltage for the VLF designs?

There are many many design trade-offs

True, I'm just trying to find a feasible starting point and then determine how to optimize the design toward my goals afterwards.

[Skippy]

Search-coil building is difficult, especially if you haven't got a working detector to try it on. I suggest using a commercial one. The coils fitted to the Teknetics 'Greek series' (omega, delta etc) and Fisher F5 are reasonably broadband, being also used on the Tek G2 and new Fisher Goldbug. Also coils designed for Minelabs FBS machines should be wideband, but some have a pre-amp built in, so this may not suit your needs.
A TX drive level of +/- 5V should be adequate, so use of a single +5V supply is possible if tuned single-frequency operation is used. Power needed to drive the coil is typically below 50 milliWatt.
Regarding dynamic range, you will need 18 bit resolution, I think. Some commercial machines will do some pre-processing in analogue, such as compression, which allows 12 bit conversion, some just use a 24 bit ADC.

[Davor]

A follow on to #6, is there any advantage to sensing Rx current as opposed to Rx voltage for the VLF designs?

It comes to the same arena because of the coil impedance and the ways to transform it. In case you are using voltage approach, you'll have Rx impedance of several kohms, (off resonance) parallel tank, and due to the coil loading there will be some phase shift. Same goes for current sensing, series tank, and again some phase shift due to loading. Both approaches are in use.