THREE COINS IN THE P.I. FOUNTAIN

[Tepco]

Coin rule?

Very interesting, and probably correct explanation, in situation when TX pulse is long enough, longer than object TC. I played with this time ago, and noticed that actual coin separation is not that critical, very little change from thin tape insulating layer to 1mm plastic, comparable to coin thickness, TC “superposition” takes place. However I find something like “coin rule”, that is, maximal rise in amplitude occurs when spacing between coins is about coin diameter, (18mm for 18mm coin etc) now amplitude superposition takes place, very little impact on TC. Normal coil is fine for this test, tried with few different coin types with about same result. With short TX pulse, slightly shorter than TC, situation is different. Now, stacking coins will increase amplitude but TC will change very little. This on picture below is taken with fast setup and 10uS TX, with empty, one and 3 coins, clearly visible early time response (this remain with longer pulse too) but almost no change in TC, change appears when TX width approach TC. Actually, with very short pulses TC is not constant at all, it varies with pulse width. (sorry for picture quality, scope is rubbish, camera too, scale is 5uS\div.).

[Ferric Toes]

As I understand, each single eddy current space (on coin surface) act as separate inductance. What single coin has, can be only resultant of many (depend on frequency) "eddy current" inductances and not fixed coin inductance per se (cause resultant of coin inductance changes according to coin orientation). Reason that stacked coin has resultant similar ti single coin, is up to shielding effect that upper coin has to other two bottom coin.

At the frequency components we are dealing with in PI, a coin does have at least two, and possibly three eddy current modes. If the coin is thick and/or is made of a very conductive metal such as copper or silver then then at early delays and the coin normal to the magnetic field, you could see an early time skin effect which causes an upward lift to the signal prior to settling down at the main fundamental exponential at 1 tau and onwards. This is inductance no. 1. Inductance no.2 is the fundamental mode which is operative for the rest of the decay and gives a straight line negative slope on a log lin plot. Inductance no. 3 is if there is a component of the field at an angle to the plane of the coin, or if the coin is on edge. This has a considerably faster decay and smaller amplitude than the fundamental decay, but nevertheless it is there and will sum in at the early part of the decay. With most modern coins and especially cupro-nickel, these additional early decays are long gone within the first 10-15uS, so we are just left with the fundamental single inductance decay for each coin.

Now with stacked cupro-nickel coins that are insulated from each other, each coin can be regarded as a single inductance in parallel with a resistance and that has a single TC. Stacking one on top of the other means that they will have a mutual inductance and be tightly coupled magnetically. Have a look at series and parallel coupled inductances and you will see that for parallel the total inductance is not far off that of a single inductance. Once you have more than two it become tricky to calculated, at least for me, and then I don't know the coupling factor. What is observable in the tests I did is that the TC increases with each coin, so either the inductance has stayed the same, or gone up, and/or the current path resistance has gone down. Because of the mutual coupling, you can regard the current paths as being in parallel, which will push the TC up. My view is that both effects are taking place. Later, I will try the same tests with coins that are more conductive, such as quarters or copper pennies, and I am sure the results will be quite different. Don't anyone ask me to do plated steel coins.

Eric

[WM6]

The stacked coins are acting as mutually coupled inductances where the total inductance is similar to that of a single coin.

As I understand, each single eddy current space (on coin surface) act as separate inductance. What single coin has, can be only resultant of many (depend on frequency) "eddy current" inductances and not fixed coin inductance per se (cause resultant of coin inductance changes according to coin orientation). Reason that stacked coin has resultant similar ti single coin, is up to shielding effect that upper coin has to other two bottom coin.

[Ferric Toes]

Yes, if the gaps between the 3 coins is negligible.

I think the difference between your plots and green's are due to the proportional TX. For this experiment, I think you want to hold the pulse width constant, and large enough that the thickest target has died out before turn-off. Your plots suggest 100us is about right.

For a nickel, and looking at the linear plot in post 27, you can see that the start value of 1000 is the star at the top of the plot. The single exponential fit did not reach star because there is a faster exponential superimposed due to the skin effect of the single coin. This is with a TX pulse of 40uS (TX is always 4 x delay). At 20uS delay and 80uS TX, and from then on, the single exponential fit is good, indicating that the coin is fully energised for all TX pulses >80uS. It is important that the coil current flat tops at the same value for each TX pulse width i.e. is not inductance limited. Even for the shortest TX pulse (40uS), you have 35uS of constant current. A more conductive coin such as a quarter, would need a much longer TX pulse and reach a single exponential decay considerably later. All this can also be checked on a scope as the delay/TX/sample control is rotated. The decay waveform fills out and then becomes invariant at the point where the TX width is fully energising the target.

I did some tests today with a UK 5 pence coin which is smaller than a nickel but still a cu/ni alloy. Two coins side by side give 2x the signal of 1 coin at any delay. This indicates that the TC of two coins is the same as for one, and just the amplitude is affected. Conversely, two or three coins stacked gives similar results to those above i.e. the TC increases as coins are added.

It's now fairly clear to me what is happening. The stacked coins are acting as mutually coupled inductances where the total inductance is similar to that of a single coin. However, the small inductance of each coin is associated with a parallel resistance (resistivity of the metal) which is relatively high for cu/ni, and even if the coins are insulated, these resistive losses are effectively in parallel. The loss is less, the current is shared, and the TC gets longer.

Basically, 100 individual coins scattered on the surface, even if quite close to one another, will need different detector settings to 100 coins packed into a jar, even if not electrically touching. Of course we all knew that already, but it's nice to know why.

Eric.

[Carl-NC]

Are not the three insulated coins acting similar to a laminated transformer core, preserving energy loss i.e. our signal?

Transformer laminations are aligned with the magnetic flux lines, so that eddy currents attempt to transverse the laminations, which they can't. In this experiment, the insulated gaps are normal to the flux lines, and eddy currents are unaffected. The results are the same with or without the insulated gap, as there should be no vertical current flow.

[Carl-NC]

The question I am wondering about is, if I had a piece of identical solid metal equal to the thickness of three coins and the same diameter, would I get the same decay curve as that of three insulated coins?

Yes, if the gaps between the 3 coins is negligible.

I think the difference between your plots and green's are due to the proportional TX. For this experiment, I think you want to hold the pulse width constant, and large enough that the thickest target has died out before turn-off. Your plots suggest 100us is about right.

[Kev]

The question I am wondering about is, if I had a piece of identical solid metal equal to the thickness of three coins and the same diameter, would I get the same decay curve as that of three insulated coins?

Eric.

Hi Eric,
Should you not get a stronger response?
Are not the three insulated coins acting similar to a laminated transformer core, preserving energy loss i.e. our signal?

Cheers
Kev.

[green]

Eric
Your plots with nickels look closer to being the same at the start. I averaged my readings from 10usecto 12usec to see the difference. (54usec pulse) 1 coin .0378v,
2 coins .0381v, 3 coins .0403v. (145 usec pulse) 1 coin .0625v, 2 coins .0705v, 3 coins .0691v. Mine were lower at the start. Thanks for all the testing you have been doing with targets and ground response.

[Ferric Toes]

(Carl got it right from the very start, working with frequency domain discriminators teaches a person these things.

Anyone who might wish to disagree with Carl and myself on this, don't bother arguing with us, I recommend first collect your own data and argue with that.)

I posted my plots since they were different from Erics. Why I don't know. Maybe neither test relates to measuring the coins at one and two times the coil radius. I got very little difference in amplitude at 10usec when adding the coins. At 40 usec adding nickels increased the amplitude. Adding quarters decreased the amplitude. We have a sign at work (One test is worth a thousand expert opinions). Only true if the test is valid and relates to what you want know. Measuring the temperature of a cup of coffee with a large metal dial thermometer gives the right answer but not what the temperature was before measuring it. My test is simple. Anyone with a DSO could do the test. If it doesn't relate to PI metal detecting I would be better off not doing them. I've worked in measurement for many years and for me it's fun. I'm not trying to disagree with anyone. When it comes to metal detectors I don't know enough to agree or disagree. Just wanting to know what testing methods might be valid.

We are all learning, whether we have been in the game for a few years or many. There are many variables which can alter results, but experimenting can be extremely valuable and as you say, Green, it is fun. By experimenting, I have come across things which work, that no amount of time doing calculations would have done. When it works, then see how the theory fits.

I have just done another plot for fun. Using our plentiful 10p coins, which are a similar material to a nickel, I set my instrument to 80uS delay where 1 coin does not quite give a reading. I then stacked up one at a time 15 coins and noted the reading increase. This is still my 30mm diameter solenoid coil, so the coins are getting further away as they stack up. However, the amplitude increase at 80uS shows the lengthening decay time. X axis is number of coins.

Eric.

[green]

(Carl got it right from the very start, working with frequency domain discriminators teaches a person these things.

Anyone who might wish to disagree with Carl and myself on this, don't bother arguing with us, I recommend first collect your own data and argue with that.)

I posted my plots since they were different from Erics. Why I don't know. Maybe neither test relates to measuring the coins at one and two times the coil radius. I got very little difference in amplitude at 10usec when adding the coins. At 40 usec adding nickels increased the amplitude. Adding quarters decreased the amplitude. We have a sign at work (One test is worth a thousand expert opinions). Only true if the test is valid and relates to what you want know. Measuring the temperature of a cup of coffee with a large metal dial thermometer gives the right answer but not what the temperature was before measuring it. My test is simple. Anyone with a DSO could do the test. If it doesn't relate to PI metal detecting I would be better off not doing them. I've worked in measurement for many years and for me it's fun. I'm not trying to disagree with anyone. When it comes to metal detectors I don't know enough to agree or disagree. Just wanting to know what testing methods might be valid.

[Ferric Toes]

Here are the plots for the 3 Nickels test. As each coin is added it is insulated from the previous one by double sided sellotape. This prevents relative movement of the coins while the measurements are taken. To answer Carl's question about the TX pulse, all my tests are taken with proportional pulsing. This means that the TX width increases proportionally to the delay time and sample width. i.e. 40uS TX for 10uS delay and sample pulse width - 80uS TX for 20uS delay and 20uS sample width - and so on such that at 100uS delay we have 400uS TX and 100uS sample. Constant current TX circuit keeps current the same for all pulse widths. With the Nickels I still see the drop in initial amplitude as each coin is added. This shows up more on the linear plot than it does on the log linear as the top end is obviously more compressed on the log. Actual data is in the data table.

The question I am wondering about is, if I had a piece of identical solid metal equal to the thickness of three coins and the same diameter, would I get the same decay curve as that of three insulated coins?

Eric.

[Davor]

I think these experiments speak volumes about target identification. Similar experiment cleared the ferrous materials orientation dependence. With targets buried in ground you can't get clear idea why they appear like this or like that, because in a process of digging you disturb the original orientation and there is no way of learning much.

Shovel is the ultimate discriminator...

[Dave J.]

Carl got it right from the very start, working with frequency domain discriminators teaches a person these things.

Anyone who might wish to disagree with Carl and myself on this, don't bother arguing with us, I recommend first collect your own data and argue with that.

--Dave J.

[Ferric Toes]

I didn't get a difference in amplitude at 10 usec when stacking the coins the other day. I tried again today. I measured 1 coin and 1 coin stacked on a wood spacer = to 2 coins to see if position mattered. Plotted the same. I tested 1,2,and three nickels again with different charge times, 54usec and 145usec. Very little change in TC.

I'll give it a try with nickels starting at 10uS and see what results I get.

Eric.

[green]

I didn't get a difference in amplitude at 10 usec when stacking the coins the other day. I tried again today. I measured 1 coin and 1 coin stacked on a wood spacer = to 2 coins to see if position mattered. Plotted the same. I tested 1,2,and three nickels again with different charge times, 54usec and 145usec. Very little change in TC.