tightly wound copper coil is bad ?

[Tec]

Measure the ringing without damping- good idea

I'm in Australia...hey good idea about Measuring the ringing with no damping resistor it will give me a measure of the whole system..

Another thing about the 'good ' coil..it is made from 'tinned' copper wire..i.e each strand is coated with solder .

On a non-coil topic here is discrimination in action ! (the picture is the 'good' coil made from the loosely wound tinned insulated wire (Jaycar WH3007)

http://www.youtube.com/watch?v=B1wzQQEOEzQ

[Tepco]

There are interesting points in this discussion, and interesting video, so i will add some input here, maybe interesting to someone, regarding coils, discrimination etc.

First , tightly made coil made using enamel-coated wire is worst possible design for PI monocoil. Just because parasitic interwinding capacitance is too high, leading to long decay time of RLC network formed this way. Since target information decays exponentially, whit this "slow" coil less information can be obtained, no matter if you use sampling, AD conversion, time-period measurement or any other means to extract target information. Only way to make it even worst is shielding whit thin alu-foil. Actually, only reason for this shield to work in PI is the fact that thin foil itself is "fast decay" target, so it will not ruin already slow coil performance too much. Will work for VLF, but adding more metal into searchead is not a good idea at all.

Useful measure of coil performance is its series resonant frequency, indicating amount of parasitic capacitance (can be calculated knowing L), can be achieved analyzing ringdown without Rdump, or measured directly using scope and signal source. Anyway, any other coil design, using PVC or teflon insulated wires, flat spiral (my preffered), zig-zag , whatever will reduce parasitic C (increase interwinding space) is better. Flat spirals have major disadvantage, due to large surface area adding shield will slow them down more than other types (but still remain fastest). Recommended shielding is conductive nonmetallic compound (graphite etc) whit conductivity higher than surrounding medium (if it is say salt water)

Considering PI waveform, it consists of 3 different components (ignoring some minor issues like input amp recovery speed, open loop gain limitation etc)

First one is from energy release in flyback "bang", lasting some relatively short period of time before it is dissipated. Second component is eddy current decay of coil itself (it is just bunch of metal), and 3. component is target signal, if present, superimposed on coil decay curve. Ironically, target signal in time domain contains all information about target conductivity, and also magnetic property (ability to store some energy in B field), like in VLF, just this time without "polarity sign", making effective 100% correct Fe-nonFe discrimination impossible. Nearest thing is some sort of calculated guesswork based on decay time analysis to "discriminate", see Chance project here for example, using 28 samples, GS4 analog 2 channel approach, some Bulgarian machines, Infiniuum XL and many others. But this is not "true" discrimination based on magnetic properties, it is impossible without balanced coil system.Obviously using comparator-timer method same effect can be achieved. Leading to interesting conclusion about coil material.

Tinned wire, multistrand insulated litzwire,PCB printed, led or gold or barbwire (or uranium wire, enriched one if possible) is nice idea. (Well, guys from AU, you have 42% of world's uranium reserve, why you are digging gold at all!?!)
Seriously, making coil from different materials make sense, this will change coil's own decay slope, and since target slope is superimposed to it, some variations can be detected and processed. Still not 100% Fe disc. but can be useful feature in some situations and to some degree. Also possible whit regular design Cu coil, this may just save some processing effort.

Finally, SPEED LIMIT consideration. Natural intention for any PI design is "do it as fast as you can" approach, fast coils, sampling, converting, measuring or using any other way, but there are some limits. From my experience, something like 6-8uS is minimal useful time after flyback that can be used in real soil, or maybe sometimes even longer. Making machines faster is possible, i did few below 2uS, performing miracles in air tests, but ending as a test tool for target response evaluation, 100% useless for metal detecting in any real soil. I got splendid success whit fast machine, some 4uS, but just, in soil-fresh water combination not too often can be found, for sure 100% useless in AU for example. Making ultrafast detectors can be challenging, but not always best solution. 6-8uS mentioned (even this may be too fast for some conditions) is easy without many technical and coil related complications and just good enough. No one in industry goes faster is far as i know, and no one will use faster detectors even if they are available.

This is bi longer than intended, but may be useful reading, forums are all about. Best regard.

[golfnut]

I guess your trying to get the most energy from your coil in the shortest time.

Leaving time to one side for a second - no pun, To get the loss down, power coils in Radio eng were often silver plated wire to keep the R value lower. This means you can get a higher drive current - and higher mag field from the same drive volts. At the same L value (as current flow restriction is by L AND R) - So free gain by reducing R for same L.

The interwinding capacitance (capacitance between turns) is a major component and does slew the frequency proportionately. Also having very thin wire - this wire has a higher self inducance per unit length, plus a higher R - So thin is out for fast power systems.

I guess the physics is steering the Tx coil to be only a few turns of thick silver plated on ceramic turn spacers - so you have air between the turns - no enamel - no epoxy etc.
Boom, signal tastic.

This will give the highest bandwith shortest response time Tx coil - with low L, R and C - So you get High I small t, big H.


As with all things - the best Tx coil probably isnt a good Rx coil - few turns - may not Generate enough volts from the decaying H field from target.

Id think having a Tx col as described but with a regular Rx coil - possibly with overlap and partial null of Tx signal.

You may have a separate Tx cct and coil and a separate Rx coil and Rx - using a time synch pulse to gate Rx on. You could have the Tx coil in the centre, and have a larger rx coil around it wound in the opposit direction the give good isolation between tx loop and rx loop.

These are just ramblings - food for thought.

Steve

[Tec]

Hi,

I have attached a picture of my exactly 400uH (403uH) 'park-master' coil, I intend pinpoint $2 coins amongst thousands of pulltabs and bottle tops...so it is small, specifically for parks..
anyway...


Tepco,

Interesting info,
regarding quote:
...making effective 100% correct Fe-nonFe discrimination impossible. Nearest thing is some sort of calculated guesswork based on decay time analysis to "discriminate", see Chance project here for example,

I checked out the chance project, and the Minelab patent US5506506
It seems like actual 'real' discrimination..it uses pulse width.
Different pulse widths affect the iron.

In the patent they are using short pulses and long pulses, but seem to think that it is the different sample delays that are doing the trick...maybe they don't understand their own invention ?

Anyway,
Is it not 'proper' Fe non fe discrimination ?..

Instead of going fishing today, I set up some tests (see graph)
The bottom axis is pulse width/10 (eg 25 = 250uSec)

The y axis is a magic ratio I was measuring.
The DASHED lines are magnetic (fe) targets.
The SOLID lines are not.

What I think is interesting is the nice parallel-ness of the solid lines
Light green = copper coin
Dark blue = 20c
Light blue = old-style pull tab
Dark green = 5c

And look at the bobby-pin (dashed red), bouncing all over the place with different pulse widths I am guessing because it is a complex shape.

The yellow dashed line is a bottle cap, it looks the same as a non-fe target...

But all the others, sag away from the ideal shape mapped out by the non-fe targets..
They are more complex...

[WM6]

Nice work TEC.

I do not understand all your comments.

Can you explain which ratio did you measure?

"The y axis is a magic ratio I was measuring."

[Tec]

ok..

I think I will submit a provisional patent first, in-case Minelab steals it.
It will take a week or two to do this

[Tepco]

Mentioned US5506506 (and dozen others) explicitly require separate TX and RX coils, that is, balanced coil system, DD or any other. Also waveforms cited in this (and other) patents show signal sampled from RX side. In this case balanced coil system will make 100% Fe-nonFe discrimination, detector will only process coil information. MPS (Multiperiod, different pulse width), followed by another MPS (Multipulse, pulse burst followed by single flyback TX, same energy release) DVT (multiperiod again, this time same energy in each pulse, so you need two power sources to store same energy in the coil for different pulse duration) etc, long list of Minielab improvements, not necessarily related to practical implementation in some machines. These methods are in place in order to battle ground effect, and if you observe waveforms and response from different targets, you will instantly realize how this works, without even having to read ML patent paperwork. Chance is story for itself, using MPS (multipulse) waveform in early, and completely specific one in recent version, but not built like MPS machine, instead sampling between 6-28 waveform points whit monocoil in different versions, not designed to sample from balanced coil. Only God, and maybe author know how this is processed later, what i mean saying "calculated guesswork" See its VDI response after all.

I can not fully understand posted waveforms (time scale sampled to 250uS probably, not clear what Y reference is for same targets), and how it is measured, set of decay waveforms without target and whit different ones may provide clearer picture.

[WM6]

Mentioned US5506506 (and dozen others) explicitly require separate TX and RX coils, that is, balanced coil system, DD or any other. Also waveforms cited in this (and other) patents show signal sampled from RX side. In this case balanced coil system will make 100% Fe-nonFe discrimination,

.

Did you know in which ML model is this patent first applied?

[Tec]

Ok balanced

Ok Balanced.

I am going to have a seperate recieve coil.
I want it to be concentric, because it is simple to build, a circle inside a circle.

I don't like the whites one with two rx coils because it is complicated, I don't like the bucking coil, because I don't understand it,
double D is no good either because it requires a former.
oOo , is 50% more complicated.. that o in a O....

No !
A Rx coil in the middle, and a Tx coil on the outside..anyone can point me to a plan with : diameters and inductance's ?

(Not a VLF coil)

So..the way I see it in a perfect world, from reading the forums posts, The balanced coil will just give me target signal (when balanced)
Balanced RX amp, or a (capacitor on the input) So I will see the signal from the switch on and switch off. ..not just the switch off.

I have been listening to the advice to use balanced, but avoiding it because it is a more complicated coil, but maybe it IS the best, if not only, way to get iron discrimination

Thanks

[6666]

Hi
is this the type of coil setup you are looking for ?

[Tec]

Yes Co-centric co-planar balanced

Yes, thanks.

And I found this !!

http://www.geotech1.com/pages/metdet/projects/coplanar/coplanar_150.pdf

[Tec]

Ok, how about this..

Tepco,

What about a transformer (Toroid) on the monoloop line.

This will give response to rate of change in current (Delta I on Delta T)... This is one thing two coil system gives and one coil does not... Rate of change in I

I notice when I hook up a transformer (toroid) in this way, I get a signal that is sensitive to Fe non-fe..Iron flattens out all frequencies, non -iron increases the slope (and rate of change of I). Attached scope.

Anyone doing Fe non-Fe discrimination like this ?, one way of getting the information out of the pulse without using balanced coil..

[WM6]

Anyone doing Fe non-Fe discrimination like this ?, one way of getting the information out of the pulse without using balanced coil..

Ferrite trafo balanced coil circuit was patented decades ago, marketed too, but production discontinued probably due poor depth.

Probably new approach can get us better results. Worth to try.

[Monolith]

For FE discrimination, you can also try, with separate TX and RX coils, is to damp the TX with only 10k. It will now oscillate.

In the oscillations you have a phase shift for FE and an opposite phase shift for non magnetic metals.

You can enhance the oscillations with a small capacitor.

Higher frequency oscillations seem to produce more phase angle.

Monolith

[Tepco]

Tepco,

What about a transformer (Toroid) on the monoloop line.

This will give response to rate of change in current (Delta I on Delta T)... This is one thing two coil system gives and one coil does not... Rate of change in I

I notice when I hook up a transformer (toroid) in this way, I get a signal that is sensitive to Fe non-fe..Iron flattens out all frequencies, non -iron increases the slope (and rate of change of I). Attached scope.

Anyone doing Fe non-Fe discrimination like this ?, one way of getting the information out of the pulse without using balanced coil..

Frankly, i didn't have success whit this. Inserting toroid (basically current transformer) in series whit coil means measuring current thru it, not voltage across it. Changes whit different materials can be observed, but due to coil parameter change (increasing L whit Fe and decreasing whit non-Fe), whit large targets very close, not due to eddy currents, resulting in low sensitivity. As far as i know, no one use this, but maybe something can be done...

And two more words of caution. First, machine like one posted in #25 is not PI, it is induction balance, and you can use any waveform, including PI one for IB (ML does, so what?) Still, all issues whit ferrite and ground balance remains, and must be sorted out some way. (Not going into discussion what is PI, IB, hybrid or whatever, any cat catching mouse is fine, just terminology)

Another is coil behavior, why #25 is not built using DD coil. Well known and probably many times discussed here, coil phasing. Due to it's construction, DD coil exhibits specific response, observing Rx side, signal polarity will change whit object orientation, going in one direction if, say, coin is vertical, and in another if it horizontal to coil plane. This is unimportant in VLF like Tesoro, Fisher etc. typical "two channel second derivative "designs. Simply, both channels will go in same direction, no matter which one, and only difference is measured.OO coils work differently, giving unambiguous response for Fe or colored in definitive direction. This may be important, depending of how signal is later processed. Best way is to build the coil, balance it, (some terminating resistor is needed on RX side too) and observe on scope responses for different targets. This will give you best idea what next step should be.