Triangular Wave Technology

[Davor]

Costas loop would do it.

[Aziz]

Nah, forget any balancing loop circuit.
DSP software does it very well.
Aziz

[deemon]

Costas loop would do it.

IMHO , Costas loop is "too much" for this application . Where we use Costas loop ? We use it when we need to demodulate the signal without phase reference . But in metal detector we have this reference - it's our transmitted signal of course . So all we need is to correlate ( multiply and integrate ) received and transmitted signals , and result will be proportional to residual in-phase signal ( including polarity ) , no matter what is it's cause - coil disbalance or maybe some ferrite near .... so we can easily close the feedback loop via op-amp integrator , which will control the balance . Continuous working of this feedback loop will maintain the residual in-phase signal near zero , so we'll have a quite good balance all the time .

[deemon]

Nah, forget any balancing loop circuit.
DSP software does it very well.
Aziz

Of course , DSP can do almost everything ... but don't forget that DSP performance is limited by ADC performance . And if we balance the coil and subtract all unwanted signals ( carrying nothing information ) before ADC , we can push up the gain , and give to ADC more useful signal from the target , rising our signal above ADC's quantization noise . Is it good ? Yes , of course

[Davor]

IMHO , Costas loop is "too much" for this application . Where we use Costas loop ? We use it when we need to demodulate the signal without phase reference . But in metal detector we have this reference - it's our transmitted signal of course . So all we need is to correlate ( multiply and integrate ) received and transmitted signals , and result will be proportional to residual in-phase signal ( including polarity ) , no matter what is it's cause - coil disbalance or maybe some ferrite near .... so we can easily close the feedback loop via op-amp integrator , which will control the balance . Continuous working of this feedback loop will maintain the residual in-phase signal near zero , so we'll have a quite good balance all the time .

But if you pay closer attention, what you just described is in fact a Costas loop. Whether a loop is closed against a self produced floating reference (true Costas loop) or if it is closed against a difference between a known reference + a floating difference it's function is the same. The true Costas must close a loop on unknown frequency and phase (basically a same thing from a loop perspective) or just the phase in a case you just described. It is the very same thing.

And yeah, a Costas loop can also be implemented in DSP.

[Aziz]

Ok, it's time for a blatant advertising again:

Why don't you guys implement the whole thing in the Netbook/Tablet/Laptop PC environment? Its very elegant and you have a very elegant and KISS + simple controller hardware.
Believe me. You can give Mr. Candy a kick in the a.se with the DSP coding and GB algorithm coding as well.

Cheers,
Aziz

[Davor]

Maybe because I wish my solution continues working past the Netbook/Tablet/Laptop PC obsolescence limit, which is, errr... tomorrow?
You see, I have a CRO half my age, and it will be a respectable piece of lab equipment for many more years. In meantime every piece of equipment using Netbook/Tablet/Laptop PC technology became junk beyond repair, and in fact non-existent.
Apart from various FIR signal conditioning, and a few DSP number crunching procedures, everything else is feasible in traditional hardware.

I became utterly disgusted by soft solutions after Altera suddenly (I think it was '99) changed specifications rendering solutions based on a previous version of their chips obsolete. You just can't expect any of the soft solution chips to hang around long enough so that you can be certain about spare parts availability.

[Aziz]

Oh man!, what a luck I didn't code it on the Z1 (Zuse Z1) computer.
And I didn't code it on the Z3 as you can see, what happened to them all.
And I didn't code it on the Cray-1 computer.
And I won't code it on the IBM BlueGene.

Ill wait till there is a computer, which understands me:
<Star Trek mode on>
Computer, start detecting a decent sized gold nugget on the planet X.
Computer, beam the nuggets to Earth.
Computer, - oops - Scotty, beam me up!
<Star Trek mode off>

Comeon guys, just do it!
Aziz