Signal level vs normal coil decay

[Goldfinder]

This is a general question on the physics of the signals from the normal Pulse Induction (PI)coil response versus when there is a gold coin or nugget in the ground. Say the coin is a 1/2 oz coin/nugget. How much difference is there between the normal decay of the PI signal return versus the decay with the coin/nugget in the signal. An assume for different depths of coin being buried

For example, if the normal decay response from the background is a 1000 (integrate the signal across some important part of the signal); then, is the response with the 1/2 oz metal a 1001 or 1000.001 or 1010 or what for various depths?

I assume Erik or Reg or Carl could provide a ballpark answer to this or someone who is an similar expert in PI machines.

My reason for this is to make a PI detector with a PIC that has an AD built in or interface it to a stand alone AD converter. I am an expert in digtal signal processing but lousy at analog EE. With DSP I have done programs that could find a signal with a SNR of 1/1,000,000.

Goldfinder

[Carl-NC]

Re: Signal level vs normal coil decay

When you look at just the coil decay on an oscope, you can't see any deflection at all when a target is swept. However, after the preamp gain (& diode limiting) you can see the deflection for a strong target response. I'll guess and say that, after the diode clamps turn off, the difference in the integrated decay areas is a percent or so.

Sounds like an interesting experiment. It also might be useful to divvy up the decay into several time spans, and integrate each time span seperately. Then you can tell which part of the decay curve is moving.

- Carl

[Reg]

Re: Signal level vs normal coil decay

Hi Goldfinder,

I am not sure how to answer you since the answer is more of a variable than set value.

The big difference between something like a half oz nugget and a 1/4 oz nugget is the overall decay time and that varies with the actual structure of the nugget. A 1/4 oz nugget may decay in 50 usec or less, while a 1/2 oz nugget may extend to 80 usec or more. In either case, the signal is very strong when the nugget is close to the coil, but as Carl mentioned, it really can't be see until it has been amplified 500 times or so.

If you take a solid mass type nugget 1/2 oz size, and amplify it a million times or more, then try to measure the signal strength, it will change simply because of the distance from the coil. You may have a nice strong signal at 10 inches, but at 12 inches, the signal quickly becomes lost in the noise base. This is where a DSP might help.

Lets take a look at Carl's PI. Gain of the first stage 1000, gain of the second stage, 100, gain of the 3rd stage, 100, plus there is a gain in the audio circuit but I haven't measured it yet. Now, just the known gains are a total of 10,000,000, if my math is correct. This type of gain is needed to obtain the 12" distance or so of detection.

Once at the output of the 3rd stage, then it is a matter of just how weak of a signal can be separated from the noise and this will depend greatly on the noise base which can vary significantly. I have seen the additional distance vary as much as two inches or more just because of noise.

So, if DSP is to be used, then it is a matter of determining how many stages should be built with analog and then analyzed with the digital. This will depend upon a lot on the resolution of the D/A as well as the overall sensitivity one wants to have on the detector.

A 10 bit would be the least, I would even consider, but prefer 12 bit to 16 bit. As an example, I think the Treasure Baron Cointrax uses 16 bit resolution. This is done by some unique manipulation of 2-8bit channels for each of the two channels of info.

Reg

[Goldfinder]

Re: Signal level vs normal coil decay

Thanks for the interesing info to Reg and Carl.

Looks like we have a good chance to make some improvements. Reg says after stage 3 and looks like Carl is between stage 2 (100000 amplification) and stage 3 (another 100 = 10**7)

So if a sampling A/D with 14 bits was put after stage 2 with sampling at 3 or so points in the decay curve it looks like even some simple DSP could really effect some improvements in detectability.

What say we do a project? Youall define the front end and I'll put my DSP stuff at the end. I'd really prefer a PIC microcontroller driving the whole thing otherwise, timing can be a problem.
Goldfinder