gold nugget simulation?

[green]

Replies #68, 69 and 71 I used the 100 x 265 mm rectangle. I don't know the answers to your other questions. It looks like if it plots straight on the linear log plot, on time doesn't make a difference. But then ground doesn't so I'm wondering if changing on time changes the slope. The nickel and $1 chip plotted close to straight on these charts.

Charted more targets including ground with 62 and 260 usec coil on times.The ground slope did change. I'm thinking it makes more sense to plot the charts log log if doing ground balance by subtracting a second sample from the first sample. Ground is a straight line on a log log chart so it is easy to add a line or a band of lines(slope -1 to -1.27) for ground. If the target line is parallel to the ground line where the two samples are taken the target and ground get rejected. A nickel, foil, small nuggets, etc plot a straight line on a lin log chart. Most other targets I've tested don't until latter in the decay curve. Since we are talking about small nuggets the samples are going to be taken early in the decay curve. Looking at nugget log log chart it makes no sense to sample sooner than 8 usec for the nuggets I tested. Might gain some signal strength for the 4 grain nugget on edge, but would loose for the other nuggets. Looking at rings log log chart with a 62usec on time the washer would come close to being rejected. I've been trying to get the test sample delay as short as I could, but it looks like 8 usec where Dan is taking his with the chance is a good place to start.
I've been using small samples for ground slope, Maybe ground has a slope outside -1 to -1.27.

[Chet]

Charted more targets including ground with 62 and 260 usec coil on times.The ground slope did change. I'm thinking it makes more sense to plot the charts log log if doing ground balance by subtracting a second sample from the first sample. Ground is a straight line on a log log chart so it is easy to add a line or a band of lines(slope -1 to -1.27) for ground. If the target line is parallel to the ground line where the two samples are taken the target and ground get rejected. A nickel, foil, small nuggets, etc plot a straight line on a lin log chart. Most other targets I've tested don't until latter in the decay curve. Since we are talking about small nuggets the samples are going to be taken early in the decay curve. Looking at nugget log log chart it makes no sense to sample sooner than 8 usec for the nuggets I tested. Might gain some signal strength for the 4 grain nugget on edge, but would loose for the other nuggets. Looking at rings log log chart with a 62usec on time the washer would come close to being rejected. I've been trying to get the test sample delay as short as I could, but it looks like 8 usec where Dan is taking his with the chance is a good place to start.
I've been using small samples for ground slope, Maybe ground has a slope outside -1 to -1.27.

Hi green


Another set of interesting charts. I agree with you on 8us being well enough for early sampling. The log-log chart with the gold nuggets is very interesting. I can envision software using this to get around the hole normally caused by ground balancing. It might also be used for discrimination.


If possible could you plot your charts both ways? I think the lin-log may be more useful for eliminating or identifying ferrous objects. From a software standpoint linear amplitude and linear time spread provides more separation of the curved time constants of the ferrous vs. gold time constants. The final software scheme might need to incorporate both log-log and linear processing.


Approximately 95% of the targets dug are nails, wire, large pieces of cans or rusted band pieces off of wooden barrels. Lead and very small ferrous pieces make up most of the rest. Identifying and eliminating digging for large ferrous junk will produce more useful hunting hours and more nuggets.


The big concern is that whatever discrimination scheme is used that it doesn't identify a large nugget as ferrous.


Thank you,
Chet

[Chet]

For a target with longer than 62us TC, say 100us, the eddy currents are still increasing. In other words they have not yet reached their peak value. At switch OFF, these eddy currents must be eliminated before the new eddy currents, of opposite polarity are generated.

With 260us TX pulse, the eddy currents in a 100us target are already partly dissipated after the flat top TX current time. Less eddy currents to be eliminated, so more new eddy currents generated.

Hi Tinkerer


Thank you for the excellent explanation. It means a lot coming from you. I have read and reread almost all of your posts. I have gained much knowledge and many ideas from them. I am currently putting much of what I have learned throughout the many forums and still learning into a PIC32 based detector.


Regards,
Chet

[green]

Added non ferrous gold platted coin to see if platting made a difference to the decay curve. More amplitude, small change in slope. Coil on time 62 usec.

[Chet]

Added non ferrous gold platted coin to see if platting made a difference to the decay curve. More amplitude, small change in slope. Coil on time 62 usec.

Hi green

Looks like the gold plating is too thin to mask the base metal. But interesting results.

Thank you for plotting both ways, it helps,
Chet

[green]

Hi green

Looks like the 260 us pulse provides fairly consistent data. Is it possible to determine if there is a critical pulse width between 62 us and 260 us at where the results abruptly change? Or is it a gradual change as pulse width changes?

Thanks again for some interesting data,
Chet

More charts. If I did it again, think I would position the targets in the same location and divide by peak current before charting for each of the coil on times. Shows log log decay slope steeper for shorter coil on times. Some targets more than others. Added a decay slope for Ca clay for 31 and 250 usec to each of the log log charts at 10 usec for the 31 and 250 usec target. The 31 usec on time comes close to balancing some of the targets if you do ground balance.

[Chet]

More charts. If I did it again, think I would position the targets in the same location and divide by peak current before charting for each of the coil on times. Shows log log decay slope steeper for shorter coil on times. Some targets more than others. Added a decay slope for Ca clay for 31 and 250 usec to each of the log log charts at 10 usec for the 31 and 250 usec target. The 31 usec on time comes close to balancing some of the targets if you do ground balance.

Hi green

More interesting charts; most interesting is the slope differences of the aluminum can and the aluminum ball versus the steel ball. It creates a puzzle on how to discriminate between ferrous and nonferrous when the shape and thickness varies the slope greatly.

The concern is to be able to discriminate ferrous from both small flat gold and large lumpy gold with accuracy.

baum754 just posted a successful discrimination test with a 10kt gold 200 grain ring in the Chance PI build thread.

I don't know if a gold ring responds similar to a large solid nugget. Maybe a similar test could be done with an aluminum ring for comparison to these charts. Aluminum wire or a small aluminum nail could be formed into a ring that would be somewhat equivalent size to the aluminum ball and can targets.

Otherwise or in addition a similar gold ring chart would be interesting.

Thank you again,
Chet

[landman]

Hi Funfinder
Just for your information. I tired this with wood objects and covered them
with two sheets of leaf and it does not work over wood.
Not enough metal apparently to give off a signal.

Why not buy a few mini-pages of leaf gold (its not that expensive because its extremly thin)
and wrap it over very small nuggets-shaped alu-balls? Would create pretty authentic results.
The main signal-creation work gets done by skin-effect, so 2 layers of leaf gold should work.

[green]

Charted the nuggets varying the delay time and sample time. With fast time constant targets I thought the signal might be less with longer sample times. The increase in integrator gain more than made up for the loss in signal average.

[Chet]

Hi Green

Great work! Shows the importance of short delay time for small nuggets. Can you post the measurement data for the 18 grain nugget.

Thank you,
Chet

[green]

18 grain data

[Chet]

Hi Green

Thank you for the data. When you say gain=300, 100 uv resolution are the charts reflecting the output of the integrator or the voltage computed back to the input across the input diodes?
Also do you have very much interference when taking your measurements?

Have another great day,
Chet

[detectormods]

I myself had a lot of trouble trying to obtain a Gold nugget test standard, I had some alloy made up that has the exact same electrical conductivity as 95% Gold 3% silver and 2% Copper. This is an average of the type of Gold found around the World. There are some places where Gold can contain 40% Silver and other different metals and it is too difficult to cover every combination so I settled on the 95% Gold as a fair representation on a Gold target. These were primarily developed for testing detector coil performance in different ground types but now have become detector performance test pieces. Currently they are modelled on 20 40 and 60 Gram Gold equivalent. Pics are on http://detectormods.com/product/55

[green]

Hi Green

Thank you for the data. When you say gain=300, 100 uv resolution are the charts reflecting the output of the integrator or the voltage computed back to the input across the input diodes?
Also do you have very much interference when taking your measurements?

Have another great day,
Chet

The data is volts out of the integrator. Two stage amplifier 15 x 20 for a gain of 300. Integrator gain 250 x percent sample time. 250 x 4usec/1000usec=1, 250 x 8usec/1000usec=2. Some times I divide integrator volts by total gain for coil volts. Stability varies, some times 1 count(100 uvolts), some times more. I try to record the difference between target and no target signal.

[Chet]

Hi Green

Thank you for the quick reply.

Chet

PS please check your private messages.