Multi-Frequency

[WM6]

Even an analogue solution may be reasonably simple, with proper design even simpler than a classic VLF. Any takers?

40% more money and 40% more free time.

Take this as a hobby and life become up to 40% more colorful.

I pledge for analogue solution as first step (to get takers to take main idea) - then we (someone) can follow with uC solution.

So, can you post here block scheme of your improved FBS-like technology?

[Skippy]

Might I suggest an alternative? Rather than the BBS approach, would it be easier to 'copy' the Fisher 5/15KHz CZ-series dual frequency approach? The circuit of the CZ-5 is available online, and working out the way it functions is probably easier than a ML machine. Use of more modern opamps (eg to replace the discrete/opamp front-end amplifier), stripping out some of the features, like "Nickel Hi-tone", "salt-beach mode" (though that is a good feature), a more modern oscillator / divider circuit, and other changes, might simplify it.
It would also be interesting to make a higher frequency version of this machine, say 15 / 45 KHz for better detection of the lower-conductivity targets that us Europeans hunt for.

[WM6]

Intuitive contribution, must say, cause, as far I know, Fisher CZ series is taken (from Davor) in high consideration in this project.

[golfnut]

The ML Exp II has 3 frequencies see the scope screenshot. I think the freqs are not harminically related - so uses non resonant Coil. It Txes freq A and listens, then B then C in sequence and compares results - It just does 3 freqs in rotation, sequentially.

The fisher (2 freqs)and the whites (3 freqs) do it with non linear Tx'es doing the fundamental plus harmonics into a resonant coil at same time - with 2 and 3 Rx ch respectively, all on at the same time and comparing the op level of the 2 or 3 chs respectively.

[Davor]

Skippy, you wish to rip the best parts out of CZ

There are many ways to skin a cat and you may achieve proper multifrequency in many ways. At the end it all boils down to a proper choice of Tx. As for Rx, it may be very simple. CZ is not that simple on Rx. It can get much simpler.

As for the choice of frequencies, you may use whatever you want, but with a few caveats. Whatever you know about GB in PI applies here as well. If frequencies are too close, there is more attenuation. If they are too far, I and Q components get increasingly separated. It may be compensated with a 3rd frequency. There is also a hole as a response to combining 2 frequencies. It can also be helped with the 3rd.

Discrimination ... gets somewhat more complex.

[Carl-NC]

The BBS approach is much simpler than the CZ approach. BBS is a sequential 2-frequency design and since the frequencies are channelized in time, there is no need for complicated channel filters. Ground balance is easier as well, and you get more accurate target ID because there is no channel bleed-through. Overall, it makes for a much simpler project.

[golfnut]

If the Tx were to be sourced with many pulse widths in a long sequence, maybe never repeated? To my mind this would be more like a real spectrum of freqs used.

A simple pseudo random number generator doing the Tx slot width.


Im getting the feel that the gnd may average-out somehow - I dont have the relevant system knowledge, only a hunch. Bit like CDMA can see stuff below the noise floor.

If the gnd was like the noise in a CDMA set up,

Plus, all sizes of target would get a chance to shout as a range of freqs

[scrungy_doolittle]

The basics of the Tx drive look fairly simple on exp2 i.e. fast, not so fast and slow and back.

[ATTACH]25324[/ATTACH]

S

I see. What it looks like they are doing there is issuing a string of 8 pulses and integrating them together to get the equivalent of 1 pulse for a PI type detector.

[Carl-NC]

I see. What it looks like they are doing there is issuing a string of 8 pulses and integrating them together to get the equivalent of 1 pulse for a PI type detector.

When you look at the TX voltage it doesn't make sense. When you look at the TX current it becomes crystal clear.

[Skippy]

I'm not sure about 'crystal clear', but it does give an important insight. The fact that the receive coil then differentiates the resulting target response rather complicates matters.
Here's a thread on Dankowski's forum (with contributions from Carl and myself) that looked at the BBS signal. It starts off a bit weak, but gets more informative.
http://www.dankowskidetectors.com/di...2,24272,page=1

[golfnut]

The current is something like this then,

what is odd and looks like a third frequency is probably a 180 deg phase reversal

so freq A for a while then one cycle of freq B then phase flip and one more cycle of freq B

[Carl-NC]

Here's the voltage & current:

[punagold]

Look at this video from Garrett. The tech gives a good explanation of using multi-freq (2) in a PI. https://www.youtube.com/watch?v=XMzwi2FeIBM Starts at around 3:30.

[Davor]

Here's the voltage & current:

[ATTACH]32502[/ATTACH]

That's precisely the way frequencies are seamlessly tailored together. With aperiodic coil and no flim-flam with Tx switching, you get any frequency you wish in a sequence. Sadly, the market leader did the very flim-flam with compensating coil resistance (L/R of a coil) to improve wave shapes, and lost the significant portion of energy conservation ability of a simple design - the way Fisher did before them.

A big number of frequencies announced by marketing was backed by the ability to run any combination of frequencies this way. But two frequencies in sequential repetition was all you got. The Candy gospel says you got all those frequencies simultaneously, which is misleading and a fine example of defunct marketing.

Anyway, this may start to take shape. Discrimination-wise this is a bit more interesting, but solvable even in analogue solution.

[Carl-NC]

I suggested this approach because it is super-simple; simple transmitter drive, and simple receiver design. And easy to change frequencies as you like. Yeah, not especially fuel-efficient, but that's secondary to performance.