Standard targets

[Altra]

Standard targets are a good way for us to compare the performance of various circuits and coils. There are many threads using lead shot and foil squares as standards. Green, Carl and others have posted these ideas.
Sometimes it was unclear how the time constants of the targets were determined? So hopefully this can be described.

My first attempt was to cut mm squares and circles using my cnc. In the process, I discovered my machine is losing steps in the y-axis. So the geometries are slightly oblong. On my next JLPCB order, I plan on having a board made. Along with repairing the cnc.

Some photos of the idea

[Carl-NC]

Green has posted a lot on measuring tau. The easiest way is to use a digital oscope that can save raw waveform data. Use a pulse width that is at least 5x longer than the expected tau. Measure the preamp response with no target, save the waveform. Do it again with the target. Import both data into Excel, subtract the no-target data from the target data, then plot with a log-linear axes. Measure the linear slope to get the tau. It takes a few tries to figure out the best oscope settings, which varies with tau.

If you don't have a digital oscope, Green also made a log preamp which allows meausring tau on an analog scope. See this thread. It also discusses digital measurement. Green had another thread on measuring tau, I don't have it bookmarked so you'll have to search.

[Altra]

That's a good thread, especially with Eric's input.

[Skippy]

Myself and Mr.Green did a lot of work on target time-constants , mainly in the year or two before the pandemic.
Everything from drinks-can aluminium squares/circles, to 'World coins', up to 'large targets' like 80mm diameter lead discs.
Not forgetting the coil of wire with a resistor across its ends, which is actually a very useful one - not only is it accurate and adjustable, it can cover the wide range from f_c below 1000 Hz up to as high a freq as you want.
I suppose you're wanting a wide range of TC's , and a wide range of 'detectability' , so you're going to need all sorts of weapons in your armoury.

[Skippy]

"Sometimes it was unclear how the time constants of the targets were determined?"
My measurements were done on a Fisher F75 VLF. This has X & R demodulated outputs available , all that's needed is a DMM, then simple maths to convert the two values into a frequency / time-constant. For quicker testing, the '88' ID display can be calibrated, then all that is needed is to sweep the target over the coil, read the ID, and refer to a lookup chart/table.

[Skippy]

Regarding squares / circles cut from thin materials, eg drinks can aluminium:
The time-constant is proportional to the edge length of the square ( and obviously the diameter of a circle ). So if you test a selection of 'evenly spaced' different size samples, you'll get a straight-line graph of TC vs. length, with an accurate gradient, that is better than taking just one measurement of a 'larger' sample. After that, the TC of smaller targets, or larger ones, can easily be calculated. Specific TC targets can be made, eg. a 5us / 10us / 15us / 20us etc series of targets.
Regarding the circle/square relationship:
somewhere in a thread I posted the results of a comparison ( done with drinks can metal ) between square/circle. They were surprisingly close, about 3% (from memory) different.

The 'coil + resistor' targets I made used drive coils from a flat brushless motor as commonly used in floppy disk drives. There are 6 identical coils, self-bonded ECW construction, roughly circular ( sort-of rounded triangular ). They can easily be levered off their PCB to re-use. I also have a couple of hand-wound coils about 60mm diameter, tightly-bound and consistent. Low enough resistance to make 1kHz targets.

I will have to track down the relevant threads.

[Altra]

Skippy, sorry I didn't give you credit. I hadn't seen that thread since 2018, until yesterday when Carl linked it.

I made an adjustable target years ago using a 20R pot and about 6 turns of #32 awg. It could trick the detector's vdi for most non-ferrous targets.

My goal here is not to repeat all the experiments. But establish a list of common targets and their tau's. The pcb geometries are repeatable and once the tau's are established anyone can use them as a reference.

Thanks for your input

[Skippy]

Re: not giving credit: I failed to credit Eric for his inputs, and others who gave valuable contributions to numerous threads, even Dave Johnson made some posts.
"My goal here is to .... establish a list of common targets and their tau's."
One problem is repeatability, especially worldwide. I never felt confident that "0.10mm drinks can aluminium" was global. I'm in Europe ( U.K ) and it's probably true they are consistent across Europe, but the U.S.A guys don't have to use the same can 'recipe' , for example.
Coins ARE repeatable, and cheap, easy to obtain. But don't cover every TC, or size. There are no 80mm coins. PCB foil thickness is going to have a 20% tolerance. Copper wire is repeatable, easy to get and cheap ... but the test loops that we made( about 20mm ) were not a universal success. Skin effect hindered the way thick wire ( 2mm plus ) behaved. Very thin wire is near impossible to join accurately.
Are you primarily interested in very short TC targets? Eg. grain size nugget stuff?

[Skippy]

Here are some other target modelling etc threads:

detection depth - Geotech Forums (geotech1.com)

test targets - Geotech Forums (geotech1.com)

MD Physics - Geotech Forums (geotech1.com)

gold nugget simulation? - Geotech Forums (geotech1.com)

Tests of big depth metal detectors ( TR and PI systems ) - Geotech Forums (geotech1.com)

[Altra]

Mostly short tc's. That seems to be of the interest. See Tinkerers elusive 1us target thread. Also see moodz's new design which is pushing some new limits. As you say coins serve as good standards for long tc's. Everything will have some tolerances. The cola can squares work well, but I found it difficult to cut the smaller squares accurately. These cans are stamped from a flat piece of Al and probably have inconsistent wall thickness on top of that the origin sheet tolerance. I think pcb copper foil is close enough for general testing. On my next try, I'm going to add closed rings.

[Carl-NC]

Here are some other target modelling etc threads:

Thanks, I've added those to my "Best Of" bookmarks.

[Skippy]

"As you say, coins serve as good standards for long tc's"

Well that depends on what's 'long' , obviously. I know of 4 short TC coins that are common and cheap. They would serve as a 'standard reference' , as they are pretty precisely made ; you could then calibrate drinks-can aluminium squares to that, and subsequently make smaller shorter-TC squares ( or longer ) of known value. As you say, the smaller ones, sub-6mm , are hard to make accurately.
What is really needed is a 'standard' low-conductivity metal / alloy , then larger targets with short TC's can be created. Unfortunately coin CuNi ( eg US 5c , or plenty of other foreign ones ) is hard to machine. Lead, and tin ( lead-free solder ) can be successfully squashed in a sturdy bench vice to make thin samples, that's a viable method. I've made 'fake US gold dollar' coins from tin ( it's a close conductivity match for 0.900 gold alloy ) , I posted the details on Tom Dankowski's forum.
The four CuNi coins in question are:

Norway / Sweden 10 Ore from the 1980's
New Zealand threepence from 1950's 60's
UK ( and New Zealand ) sixpence from the 50's / 60's
UK currently-circulating 5 pence ( pre-2011 when they changed to steel junk )

All are common, cheaply available from eBay etc. Their TC's are in the 5 - 9 microsec range, I don't have figures to hand at the moment. From memory, a 22mm square from drinks can was 6 usec TC

And: drinks cans are extruded - a flat sheet is squished between a solid round billet of tungsten carbide and a circular ring on the same metal. There must eventually be some wear on the WC bits, before they are replaced, but the cans I've tested all fall in the 0.095mm to 0.105mm range. However, even if they are 0.12mm in your country, you should still be able to calibrate them to a known reference.

[Skippy]

The Dankowski Forum thread on making the 'fake' gold US Dollar coins:
Make your own fake gold dollars! (dankowskidetectors.com)

They were 0.7mm thick, but with some effort on the vice lever, I think 0.5mm could be achieved, less if the sample was small.

[Altra]

Skippy after reading through the various threads. You seem to be the dominant contributor on this subject. Maybe you should compile and write a article on targets. It's hard to Search out the real information spread all over the place. Also do you think the Dankowski forum will be around in two years?

I seem to have a misunderstanding of the stated target tau's. As a builder, I see tau's as measured from TX cut off. Above you mentioned that a 22mmX22mm square is approximately 6us. Assume your memory is correct, this a huge signal for a sensitive pi. I would expect to be able sense this target at 12us or greater after Tx cut off. This is where Green lost me early on when he stated he measured the tc after the flyback settled. Now if l took a test circuit where the flyback settled at 5 or 6us on the pre-amp output. Adding the settling time and target tau, I would get the observed 12us target sensing. So It must be me not understanding the difference between pure target tau and circuit sample timing?

[Carl-NC]

A target's tau determines its decay rate. It is the slope of the decay, which becomes a straight line when plotted log-linear. You can measure it anywhere (any two points on the slope), you don't have to start exactly at the TX cut-off. Usually you can't start there because there is an initial diffusion error in the target response that must settle out. This normally takes 1 tau of time, so for a 100us target you have to wait at least 100us before its decay settles in to a straight line.

Here is my approach. I use a monopolar TX (NMOS) and a preamp. You can't start measuring the target tau until the preamp comes out of overload but you also don't need to wait until the flyback is "settled." I subtract a no-target reading from a target reading to eliminate the effects of flyback settling, preamp settling, eddies in the coil wire, etc. That should leave just the target response.