EXTRA_EXTRA_DEEP DISCRIMINATING DETECTOR

[Carl-NC]

I wonder why we can't set a goal and then define what might be required to accomplish the goal. Do we need GEB? Is #4 lead shot a good test target? What distance do we want to detect the test target(GEB on and GEB off)?

Probably you will need ground balance in most gold-hunting areas. For 0.3g #4 or maybe even #2 shot is a good test target.

A good discharge slope mental model is formed when you take the coil value in uH and divide this value by Rd or damping resistor value. The 1.39 uS target would need an Rd of 1.39 divided by 5 or .278 uS with a 300 uH coil needing a 1079 ohm damping resistor to fully stimulate the 1.39 uS target. 300 uH coil divided by a Rd of 1079 is .278 uS.

Couple of problems with this. First, the coil turn-off is not a simple RL circuit with tau=L/R. It is an RLC circuit with a damping tau=0.5*L/R for critically damped. Per the equation for turn-off current, the settling is about 40% longer than an RL circuit (to equivalent 99.5% settling).

Second, even this doesn't matter. The raw coil signal is applied to a preamp with a gain of, say, 500 so when the coil voltage is 99.5% settled the preamp is still overloaded. Typically it takes 15-20 taus of settling before the preamp is in a happy spot where you can sample.

The actual position of the target hole will depend somewhat on the soil conditions, and consequently the TC you're trying to ignore. The grounds I've measured have the same R decay slope. Tx time and profile effect decay slope, maybe others. Don't think soil condition effects hole location. Tx time and profile + delay and sample times do.

Ground viscosity varies which is why we have a variable GB control. This alters the target hole position so, yes, it is fair to say that ground affects the target hole. However, it typically doesn't vary a whole lot so the target hole won't move a whole lot, either.

The surface area delivers most of the target response amplitude. Take a lead shot and knock it flat and see how the target response amplitude increases. Still the same mass of lead, but much more amplitude.
Gold nuggets are nearly always more flat than round. Often like flakes for very small nuggets.

This is why I say if you can detect a #9 shot you can detect most 1/2 grain nuggets. Because most will be flattened. However, some may be elongated or porous enough that they are actually more difficult.

[green]

Couple tests with #4 lead shot. Increasing delay by .7*target TC should reduce signal in half for straight line decay linear log(targets with a TC<10us, most targets I've measured with TC's>10us decay start straight line log log before finally decaying straight line linear log). Qiaozhi reply#55 calculated 1.39us for #4 shot TC(.7*1.39us=1us)close to chart data. Other line is amplitude vs distance in inches.

X axis is distance in inches or delay in us.

[green]

A start at what might be required for a gold detector(test target_#4 lead shot)

[bbsailor]

Probably you will need ground balance in most gold-hunting areas. For 0.3g #4 or maybe even #2 shot is a good test target.



Couple of problems with this. First, the coil turn-off is not a simple RL circuit with tau=L/R. It is an RLC circuit with a damping tau=0.5*L/R for critically damped. Per the equation for turn-off current, the settling is about 40% longer than an RL circuit (to equivalent 99.5% settling).

Second, even this doesn't matter. The raw coil signal is applied to a preamp with a gain of, say, 500 so when the coil voltage is 99.5% settled the preamp is still overloaded. Typically it takes 15-20 taus of settling before the preamp is in a happy spot where you can sample.



Ground viscosity varies which is why we have a variable GB control. This alters the target hole position so, yes, it is fair to say that ground affects the target hole. However, it typically doesn't vary a whole lot so the target hole won't move a whole lot, either.



This is why I say if you can detect a #9 shot you can detect most 1/2 grain nuggets. Because most will be flattened. However, some may be elongated or porous enough that they are actually more difficult.

Carl, Green, George

Sorry for using the wrong precise wording. I should have said. An idealized discharge slope mental model for a low TC target might be at odds with the reality of properly damping a coil where capacitance seen by the coil tends to lower the Rd value to about half the idealized value to fully charge the desired target of 1.39 uS. Thanks Carl.

We need to find creative ways to reduce the capacitance as seen by the coil so higher Rd values can be used to obtain a more vertical discharge slope to better stimulate small TC targets. Things like: coil size optimized for target size and TC; pulse power that can be quickly damped; TX speed to sample multiple RX samples to integrate to improve the signal to noise ratio and unique coil design requirements that are needed to sample in the sub 7 uS region.

When we start to get the delays below about 7 uS to detect a 1.39 uS target we must ensure the no parts of the coil design or component parts respond like a target. We often do not talk about this and I was attempting to introduce it as something to think about so Green could improve his detector design.

Right now I think one of the best design opportunities is to reduce any capacitance as seen by the coil to: 1. Reduce the delay and 2. Steepen the TX pulse discharge slope to better stimulate smaller targets. One big capacitance removal or reduction opportunity is the coax cable that adds about 20pf to 30pf per foot.

I hope this clarifies my former post.

Joseph J. Rogowski

[Qiaozhi]

An interesting point that Carl made is: "the settling is about 40% longer than an RL circuit".

I've noticed this as well, where certain targets appear to be detectable at sample delays longer than you might expect from calculation. This happens (I think) because we make the assumption that the eddy currents in the target begin to decay from the point of mosfet switch-off. When in fact the eddy currents need to build up in the target first (as the magnetic field collapses) before reaching a peak, and then starting to decay. Hence the the time to total eddy current decay appears to be longer than expected.

[green]

An interesting point that Carl made is: "the settling is about 40% longer than an RL circuit".

I've noticed this as well, where certain targets appear to be detectable at sample delays longer than you might expect from calculation. This happens (I think) because we make the assumption that the eddy currents in the target begin to decay from the point of mosfet switch-off. When in fact the eddy currents need to build up in the target first (as the magnetic field collapses) before reaching a peak, and then starting to decay. Hence the the time to total eddy current decay appears to be longer than expected.

Not understanding what 40% longer than an RL circuit means. Why does it take longer if critical damped tau is 1/2 an RL circuit?

Tried a spice circuit with a snubber at 200, 400 and 600V. Appears target signal(#4 shot)starts decaying as soon as coil comes out of avalanche. Appears could sample sooner with a separate Rx coil. Tried snubbing my real circuit but didn't see an improvement, maybe need to try again.

[Carl-NC]

Not understanding what 40% longer than an RL circuit means. Why does it take longer if critical damped tau is 1/2 an RL circuit?

Because the equations are different. For an RL circuit the current is of the form



The flyback response is an RLC circuit with the form



and the coil current has the form



All this reminds me of where I intended to take that other thread.

[Tinkerer]

Our present discussion is about the possibility of tweaking the Arduino PI, to make it into a small nugget detector.
Maybe Admin would be so kind and move the pertinent posts to a new thread that serves the purpose?

[green]

I've recorded amplifier out decay for a 10x10mm regular strength aluminum foil target to determine tau(about .7us). Tried this morning recording integrator out at different delay times(tau about .7us). I've read, target not detectable after 5 target tau. The foil is easily detected after 10 tau. Circuit avalanches for about 1us so target decay might start 1us after Tx off. What determines where certain targets appear to be detectable at sample delays longer than you might expect from calculation. What is the calculation?

target was at 1inch distance

[green]

Our present discussion is about the possibility of tweaking the Arduino PI, to make it into a small nugget detector.
Maybe Admin would be so kind and move the pertinent posts to a new thread that serves the purpose?

Are you thinking tweaking Arduino Nano Pulse Induction Metal Detector Project - Published March 2021 or starting another Arduino project?

[Tinkerer]

Are you thinking tweaking Arduino Nano Pulse Induction Metal Detector Project - Published March 2021 or starting another Arduino project?

The Arduino is a very good learning platform. Too bad that I have not enough time to learn it. toolnuts suggested to use it for a nugget detector, so I assumed that he was familiar with the Arduino. I am not.

[Tinkerer]

I've recorded amplifier out decay for a 10x10mm regular strength aluminum foil target to determine tau(about .7us). Tried this morning recording integrator out at different delay times(tau about .7us). I've read, target not detectable after 5 target tau. The foil is easily detected after 10 tau. Circuit avalanches for about 1us so target decay might start 1us after Tx off. What determines where certain targets appear to be detectable at sample delays longer than you might expect from calculation. What is the calculation?

target was at 1inch distance

How is it possible to detect a target with a Tau of 0.7us after 5us?
As a "rule of thumb" we say that a target signal is gone after 5 x Tau. If we want to look at high precision, then it depends on when the signal disappears into the noise.
For example, if you have a target signal of 1000mV, even 1% signal leftover is still 10mV.
The 10mmx10mm foil target has high amplitude and a very short Tau.
There is a way to "stretch the short Tau target signal by adding a small capacitor in the preamp feedback. we often add such a capacitor for stability and to limit the bandwidth of the opamp.
When experimenting with different opamps and feedback capacitor values, I noticed that sometimes the target signal was stretched a little.

The damping:
When we look at the damped Flyback, we see a very steep slope and then it turns into something like a foot. At this point there is still some current flowing in the coil, that helps maintaining the eddy currents in the target.
When you look at simulations, you see that the eddy current target signal starts to raise some time after the Mosfet has been switched OFF. The way to look at the target signal decay is the beginning of the decay of the target, just after it has peaked.

[Tinkerer]

Not understanding what 40% longer than an RL circuit means. Why does it take longer if critical damped tau is 1/2 an RL circuit?

Tried a spice circuit with a snubber at 200, 400 and 600V. Appears target signal(#4 shot)starts decaying as soon as coil comes out of avalanche. Appears could sample sooner with a separate Rx coil. Tried snubbing my real circuit but didn't see an improvement, maybe need to try again.

Your snubber is a very interesting feature. It needs to be done just right, then it is very useful. The choice of the diode is very important.

[Qiaozhi]

When you look at simulations, you see that the eddy current target signal starts to raise some time after the Mosfet has been switched OFF. The way to look at the target signal decay is the beginning of the decay of the target, just after it has peaked.

That's what I was hinting at in post #65.