TARGET TIME CONSTANT

[Monolith]

( One square inch of household aluminum foil has a total discharge time of 50 uS and a 10 uS TC. Eric Foster mentioned, in one of his posts, that his Goldqust PI can detect this 1 sq inch foil at 12" using an 11" coil at 10 uS.)
(So, even though gold is the primary target it is still desirable to find coins. Here is the design compromise. Any transmit pulse width beyond what fully stimulates a gold ring is wasted power from that ring's perspective but is under powered from the perspective of a copper penny that has a total discharge time of 350 uS (reported by Eric Foster on his web site). The copper penny TC is one fifth of the total decay time or 70 uS for the copper penny. A nickel is a good contrast as it has a total decay time of 100 uS with a TC of 20uS. Nickels are a substitute and a good indicator of a PI design/control settings to gold jewlery.)
I tried measuring some (Target Signals) awhile back. The time constants weren't correct. I don't have access to the equipment used then. I've been trying again at home. I get different answers but are still different than the time constants listed in the above posts. Can some one suggest what might cause my measurements to be different? I am buying (Inside the Metal Detector) hoping to answer some questions.

The shape of the TX curve has some influence on the results. With a "flattop" TX, the ON eddy currents in the target have decayed to some degree. With a near linear TX current curve, the eddy currents in longer TC targets are strong at switch OFF. With the reversal of the coil current, these eddy currents must be first eliminated before the new eddy currents can build up.

[baum7154]

Thanks for the replies. I just got back from vacation so I'm just now catching up on the forum.

So what I get from Green's post is that doubling thickness approximates doubling the TC. This gets back to the issue of aluminum foil in .001" and .002" thicknesses. Is it correct then that the .001" thick 1"x1" square would have about 1/2 the TC of the .002" thick 1"x1" square?

"Tinkerer quote:

One square inch of household aluminum foil has a total discharge time of 50 uS and a 10 uS TC. Eric Foster mentioned, in one of his posts, that his Goldqust PI can detect this 1 sq inch foil at 12" using an 11" coil at 10 uS."

I'm a little confused by the statement above in that his 1"x1" foil has a 10 us TC. We really don't know how thick this sample was. Greens Heavy Duty Aluminum foil (.002" thick) sample 1"x1" measured a TC of 2.2 us. This would lead me to think that the 10us sample was about 4.5 times thicker.

To eliminate this ambiguity shouldn't we always specify the thickness of the sample material in addition to it's area? I now understand the relevance of area to signal strength and not to TC.

Thanks,

Dan

[baum7154]

Ran a quick test with a 1"x1" piece of .001 aluminum foil as measured with my 'tenths' reading micrometer. It actually measured 0.0011" thick. In an air test in my basement the first detection of this target using my 8" fast nugget coil was at a distance of 8". The detector I used is a CHANCE PI. My basement is not the best environment for these kinds of tests but was the best I could do at the time.

[Tepco]

This test can be highly problematic. What software version of Chance you use, what coil type, original coil is not too fast at all, and what guard interval value? Chance is completely software driven detector, with one interesting peculiarity (more or less present in different software versions), it will tend to completely reject objects below some time constant, ignoring them no matter how large or close they are or how strong signal they may produce, results can be misleading.

[baum7154]

My detector is running Version 1.2.1 software. This is the latest version. I am not using the original coil. My coil is an 8", 335uh 'spider wound' coil on a machined toroidal form 1.050" wide x .094" thick Lexan form with 37 slots on the inner and outer edges using no coax and the feed line is a continuation of the coil wire itself twisted at 3 turns per inch with no solder joints in the coil. The self resonance of this coil using BB Sailor's setup is 1.042mhz.

My Guard Interval for this coil will run at a setting of 16 but for this test it was set at 17. I don't know of a Chance unit and coil that has operated below a Guard Interval of 20 on this forum. This unit was recently tested using a 128 grain placer nugget and was able to see the nugget at 9". I am trying to borrow a nugget in the range of 15 grains for further testing. This unit does detect a small collection of placer flakes in a small glass vial equivalent to the volume of a match head at a distance of 2.5 to 3". However as I posted in the Chance PI Build thread the discrimination of this unit does not appear to be functioning/trustworthy below a signal level of 6 increments. This may be due to my coil's significant departure from the designer's original recommended coil. I did build an accurate copy of his coil but could not get it to operate below a Guard Interval of 20.

I have not seen this detector ignore any metal with my fastest coil and the guard interval set at 17. The 8" coil will see a pop can at about 20" and it sees a stainless serving spoon at about 15"...Huge targets...that is why I am interested in this thread so that I can use repeatable, standardized small targets to help determine performance.

Hope this answered your questions,

Dan

[Tepco]

Original coil tend to be weakest link in entire detector design. I built chance long time ago, immediately after it was introduced, and maybe few comments can be useful to someone. First version was v.8 software, with fixed guard time and original coil, and only version actually field tested on bad ground, performed perfectly, just performance for very short TC object was poor. Making any faster or better coil was pointless, due to fixed delay. Interestingly, this responded to some small gold, but I noticed something strange: relatively large pieces of specifically shaped jewelery, not even small (over 20g), otherwise very “fast”, problematic for ordinary analog PI too, gets completely undetected, even when placed right in the center of the coil, not even interfering with detector operation, everything else, like coins and rings are detected normally (?!?). Later I upgraded software to newer versions, but only bench tested, so no idea how this can behave in real world. Also tried different coils, and by far best results I get with 10in flat spiral speaker wire coil. Inductance is somewhat lower, around 270uH, original is 400, no problem with that, just current consumption will be bit higher. Whit this, guard time can be set right down to 10, eventually 12, depending how and if coil is shielded. Effect on performance is really significant, well wort trying, no more phantom invisible objects, far better range, only no guarantee what can happen in soil. One remark here, with any coil change, very precise dumping resistor adjustment is needed, most detectors are somewhat tolerant to this, but Chance is not, even slightest misadjustment can significantly affect performance. Probably reason why original coil design is considered important. Fact that proper VDI reading kicks in when signal is sufficiently strong, not at maximal detection depth is normal, happens with just about any coil, some amount of information must be collected before algorithm can make correct calculation. One more notice on flat spiral coils, measured self resonant freq. interestingly can be lower compared to some other types (this was 870kHz, or 780, cant remember exactly) but still capable to achieve faster sampling due to some other reasons.

[Ferric Toes]

bbsailor,
thanks for the feedback. I will try the alu foils with holes.

There are some corrections: The foil I am using has been measured by a watchmaker, using a micrometer. He says that it is 0.022mm thick. this corresponds to 22 micrometers.
A foil that is 0.02" (Inches) thick, would correspond to 0.05mm thickness.

Also I made a mistake with the sampling time. The actual time as shown by the scope, is 2.6uS, (two point six microseconds) after switch OFF.

This might present a record. I have not heard of such a short delay. It also might be somewhat academic, I can not think offhand of any use for so much detecting sensitivity.

Tinkerer

Hi Tinkerer,

You will have to halve that figure and even then you will have 0.3uS to go to equal my best. I designed a hand held detector for Eriez Magnetics for detecting broken off needle tips in fabric manufacture. That samples at 1.5uS to give a bit of safety factor. Also, there was a machine mounted zoned coil that could scan large widths of fabric while still on the rollers. It used printed circuit coils and operated at 1uS delay. By the way, this was in the 1990's and there is no reason today why 0.5uS should not be achievable. There are lots of uses if you can do it, although mainly in industry. I had one enquiry though from a university here who wanted to attach a tiny metal tag to the backs of beetles and use a detector to track them through leaf mould in a forest. I didn't have anything suitable at the time but lent them a Goldbug 2 which they found reasonably good.

Eric.

[baum7154]

Original coil tend to be weakest link in entire detector design. I built chance long time ago, immediately after it was introduced, and maybe few comments can be useful to someone. First version was v.8 software, with fixed guard time and original coil, and only version actually field tested on bad ground, performed perfectly, just performance for very short TC object was poor. Making any faster or better coil was pointless, due to fixed delay. Interestingly, this responded to some small gold, but I noticed something strange: relatively large pieces of specifically shaped jewelery, not even small (over 20g), otherwise very “fast”, problematic for ordinary analog PI too, gets completely undetected, even when placed right in the center of the coil, not even interfering with detector operation, everything else, like coins and rings are detected normally (?!?). Later I upgraded software to newer versions, but only bench tested, so no idea how this can behave in real world. Also tried different coils, and by far best results I get with 10in flat spiral speaker wire coil. Inductance is somewhat lower, around 270uH, original is 400, no problem with that, just current consumption will be bit higher. Whit this, guard time can be set right down to 10, eventually 12, depending how and if coil is shielded. Effect on performance is really significant, well wort trying, no more phantom invisible objects, far better range, only no guarantee what can happen in soil. One remark here, with any coil change, very precise dumping resistor adjustment is needed, most detectors are somewhat tolerant to this, but Chance is not, even slightest misadjustment can significantly affect performance. Probably reason why original coil design is considered important. Fact that proper VDI reading kicks in when signal is sufficiently strong, not at maximal detection depth is normal, happens with just about any coil, some amount of information must be collected before algorithm can make correct calculation. One more notice on flat spiral coils, measured self resonant freq. interestingly can be lower compared to some other types (this was 870kHz, or 780, cant remember exactly) but still capable to achieve faster sampling due to some other reasons.

Yes I too started with Version 0.8.4 and there were significant upgrades to the software.

I'll have to try a spiral coil design sometime. Impressive that it will allow the Guard Interval to operate at 10. I would expect that fine small gold would be much easier to detect with that coil. Did you shield the coil and if so how and with what? What gauge of wire did you use for that coil? Was it teflon/PTFE insulated? Is there a thread in the COILS forum on your coil?

I tried to optimize the damping resistor from the original 390 ohm value in the original CHANCE, using the potentiometer method in BB Sailors paper, but was unable to get any improvement using the chance to excite the coil. Will have to revisit this with a pulse gen for excitation later on. It appears that CHANCE does active coil damping with a MOSFET and it will be interestng how much benefit there will be in chasing the damping thing.

Regards,

Dan

[Tepco]

Guard time change (or sampling delay to nearest sample) is not too dramatic all, original corresponding to 30 setting is around 17uS, at 10 around 12uS or somewhat less as far as I can remember. May be different in some versions, new software uses almost two dozen samples, this is shortest one, from TX end to ADC start command, so no problem to achieve. Coil is mentioned in some unrelated topic, probably surf related, never separately in coils, time ago, unable to find it now. In essence, flat spiral made of cheapest possible thin speaker wire, 2x .25mm, 12.5cm inner, 25cm outer radius, around 30 turns, both wires in parallel. Shield for this is graphite only, forget foils etc, can be connected wit that same, or any other wire, or coax. Next time I made one, will photograph and document everything in “coils”. For dumping resistor, if not sure, use somewhat lower value, then increase, monitoring minimal guard time, after some point with coil ringing, ADC will measure nonsense and everything will fail miserably. Don’t use pulse generator, but chance original power stage. It is not designed for active dumping, dumping resistor does this, instead it short circuits the coil between short pulses in one timing period to prevent energy release and flyback, it will happen only at the end of cycle.


Now, one more issue, many times discussed here, about small gold time constant etc, sometimes ending with absurd conclusion that gold is low conductivity material etc. Not willing to start discussion at large here (maybe in some other dedicated topic), but this is plain wrong. Actually gold is too good conductor, then small piece may have trouble to dissipate energy stored in form of eddy current, exactly because good conductivity, nowhere to be burned out, not enough resistance. So it will remain stored for longer period of time, increasing TC. I tried cooling it with liquid nitrogen to increase conductivity even more, naturally, time constant increased even more, not fallen as simple explanation may suggest. So, if you find 20g jewelery made from 14 or 18K gold having 5 times shorter time constant compared to .5g gold, this is just normal, amplitude will be higher but time constant slightly shorter, this can bamboozle some detector designs. Not logical maybe, but physics is not logical either, just you cant change it. This is why i'm bit skeptical about foil testing.

[green]

I was playing with spice to see how fast I could dump a coil. Looked like 3/coil resonance was a close approximation. Is this right or should I be looking at something other than resonance?

[baum7154]

Thanks for the description. I look forward to future coil construction details and will rerun my coil damping, monitoring the Chance Guard Interval to get it minimized. I am familiar with graphite shielding a la Satdaveuk and can implement it on a spiral coil. I may try to build one with teflon/PTFE insulated small diameter wire.

Regards,

Dan

[Tepco]

I was playing with spice to see how fast I could dump a coil. Looked like 3/coil resonance was a close approximation. Is this right or should I be looking at something other than resonance?

In most cases yes, around .35 or so of coil resonance.There are some exceptions, flat spiral for example, acting like distributed transmission line with continuous impedance change is more complicated to model and analyze, not nearly simple like ordinary coil with distributed capacitance.



For object to coil interaction however, or object TC analysis, spice modeling is about useless, with too many things going on simultaneously, some sort of finite element analysis method is needed, better just measure it, leave simulations.

[Tepco]

Thanks for the description. I look forward to future coil construction details and will rerun my coil damping, monitoring the Chance Guard Interval to get it minimized. I am familiar with graphite shielding a la Satdaveuk and can implement it on a spiral coil. I may try to build one with teflon/PTFE insulated small diameter wire.

Regards,

Dan

No need to waste too much resources and time on this, ordinary cheap plastic insulation tween lead speaker wire will do the job, coil physical layout is important, however, all improvements are welcomed and very likely possible.

[Tepco]

I had one enquiry though from a university here who wanted to attach a tiny metal tag to the backs of beetles and use a detector to track them through leaf mould in a forest.
Eric.

I’m afraid this can seriously compromise results, beetles nervous system may not be happy with applied pulses, or currents in tagging metal. Even with humans some methods, like TMS ( Trans-cranial magnetic Stimulation) may open field for more than just industrial use. It need just to pulse, not to detect anything, this is even FDA approved. Some commercially available units actually release almost 100J per pulse (well, this is not approved, certainly)


Never mid, this is my “writing day” obviously, this “burst mode” will continue for some time, so instead of this pointless comment, I have to post something far more meaningful in another topic, dealing with ground behavior in presence of static fields.


Another interesting question is what front end was used for 1uS delay, can it beat 40+year old 733, with 250nS response, when overdriven. PCB coil was most probably multilayer, very thin and fine tracks, very high resistance, to minimize coil decay time (i'm I correct?)

[Ferric Toes]

.

Another interesting question is what front end was used for 1uS delay, can it beat 40+year old 733, with 250nS response, when overdriven. PCB coil was most probably multilayer, very thin and fine tracks, very high resistance, to minimize coil decay time (i'm I correct?)

OPA620 for front end. Hand held unit used a large half pot core with 50uH winding, centre tapped. Wide unit used 4 layer pcb coil with two screen layers.
http://en-gb.eriez.com/Products/Inde...dmetaldetector

Eric.