Amplifier frequency response

[green]

http://www.geotech1.com/forums/showthread.php?22564-felezjoo-PI(the-best-pulse-induction-metal-detector-that-I-made-until-now)&p=216493#post216493

Some of the discussion was if felezjoo-PI could detect a .25x.25 inch piece cut from the side of an aluminum can. Charted the decay for the can, US nickel and some California clay. The slower amplifier changes the decay. Trying to understand what it all means.

[Carl-NC]

I haven't kept up with that thread but, yes, a modest limit on the preamp bandwidth will spread the fast response of low conductors out in time. 80kHz is 2us, which is about as much as I like to push it on a gold machine.

[Teleno]

I haven't kept up with that thread but, yes, a modest limit on the preamp bandwidth will spread the fast response of low conductors out in time. 80kHz is 2us, which is about as much as I like to push it on a gold machine.

Great idea!

The band-limited amplifier works as an integrator with time constant T. Targets with tau < T produce a response decaying as T and become measurable at a later time.
The catch: slow opamps won't cut it, you need an opamp fast enough to follow the shorter tau and integrate it.

[green]

Charted three targets with four different amplifier filter RC time constants. R=20k, C=3.3p(.066usec), 10p(.2usec), 33p(.66usec) and100p(2usec). Error,I see I labeled the capacitors uH instead of pf on the charts.

[Davor]

Well... there is a small problem with this approach - the coil decay gets extended by the same amount as well

[green]

[QUOTE=Davor;216686]Well... there is a small problem with this approach - the coil decay gets extended by the same amount as well [/QUOTE

Chart trace (Recorded target signal minus recorded no target signal). The RC filter does slow the decay. I could post scope pictures of the .2 and 2usec decays with and without the target. Jladre is detecting a 5x5mm piece cut from a soda can at 4 inches with the felezjoo PI which has a amplifier feedback time constant of 5 usec(1meg, 5pf). Don't see how he can unless the filter is extending the decay or it's not aluminum.

[Carl-NC]

Well... there is a small problem with this approach - the coil decay gets extended by the same amount as well

So does the ground decay. You want some filtering on the preamp just to reduce noise, typically I like to run a TC of 0.5-1 us. For a security walk-through PI I've run it up even higher.

[green]

The .2 and 2usec recordings for the soda can.

[baum7154]

[QUOTE=green;216699]

Well... there is a small problem with this approach - the coil decay gets extended by the same amount as well [/QUOTE

Chart trace (Recorded target signal minus recorded no target signal). The RC filter does slow the decay. I could post scope pictures of the .2 and 2usec decays with and without the target. Jladre is detecting a 5x5mm piece cut from a soda can at 4 inches with the felezjoo PI which has a amplifier feedback time constant of 5 usec(1meg, 5pf). Don't see how he can unless the filter is extending the decay or it's not aluminum.

--------------------------------------
Davor,

At least for me the jury is still out on the detection of a 5mm X 5mm square of soda can at 4". In one case it is 6mm X 6mm and in another comment it is 5mm X 5mm. If I look closely at the video zoomed up, the piece appears to be rectangular 7mm X ?mm and is not aligned with the scale divisions and is seen skewed and from a high parallax angle. There are some other questions in this test to be resolved. When we are testing the rarified limits of these kinds of targets, precision makes all the difference in the results. Just my thoughts.

Regards,

Dan

[Davor]

I guess if it would be practical to take many more samples, and instead of integration we use sum, and if every sample could be compensated individually, we could sample deeply into a coil response. Because coil response is cyclostationary, there for sure exists a method of eliminating it from samples by means of compensation. However, in lieu of a proper compensation we must sample somewhat later. Regardless of coil response being more or less cyclostationary we may expect to detect only objects that have response stronger than a coil response, and that is a limiting factor.

[mickstv]

I had a look at that youtube from the other thread, the coke can target is 6x7mm (after scaling a screenshot).

The detector must be sampling up the decay curve, not after the decay settles back to ground. If this is the case it isn't going to work very well detecting over mineralized ground.

[6666]

Who's detector can get that 5mmx5mm square of coke can ? its a tough target, tried with my TDI, no signal with 14x9 coil maxed out.

[mickstv]

The one at the link below.

http://www.geotech1.com/forums/showthread.php?22564-felezjoo-PI(the-best-pulse-induction-metal-detector-that-I-made-until-now)&p=216676#post216676

[6666]

The one at the link below.

http://www.geotech1.com/forums/showt...676#post216676

Thanks .

[baum7154]

Who's detector can get that 5mmx5mm square of coke can ? its a tough target, tried with my TDI, no signal with 14x9 coil maxed out.

----------------------------------------------------

Hi 6666,

First of all it is quite difficult to accurately fabricate a 5mm X 5mm piece of Aluminum can side wall. I was able to make a 4.96mm X 4.99mm square, .10mm (.00393") thick target for this test. I had to hand file to the final 5mm dimensions using a fine cut jewelers file. I used a digital caliper to measure this target.

By the way the use of the thicker soda can top or bottom makes a sample target that is much easier to detect because it is 2 to 3 times thicker than the sidewall. To show its effect, I made a sample 6mm X 6mm target of the top of the can and it is repeatably 'first detected' at 4.5 inches using the test method below. To be clear I did not use these thicker parts of the soda can for the 5mm X 5mm test.

The test I did was on a quiet section of my basement floor that I use for other testing. This is the test 6666 told me about that more closely simulates 'in ground testing' and I have been using it for some time now. Test result distances are reduced compared to air tests which most know to be inconsistent with in ground tests. There is more than a foot of snow again from the past few days so I can't test outdoors.

The coil I used is the 3DSS 4" X 12.5" Racetrack coil. This coil is placed on the carpet about 3/4"over a quiet section of the concrete basement floor. The 5mm X 5mm X .1mm target was taped to a wooden yardstick and 2 corrugated cardboard spacers (small boxes) placed on the floor were bridged with the ruler to present the target at a consistent and repeatable distance over the coil. A clear plastic ruler was placed vertically inside the coil to measure detection distance from the top surface of the coil. The modified CHANCE PI detector with 2 stage amp and 14.31818mHz clocking was warmed up for 7 minutes prior to this test. The detector menu selections were Guard Interval 10 (7us), Barrier 1, Battery voltage 12.6, Sample width is 9.625us due to the clocking.

First detection of this target, repeatable for 3 consecutive passes minimum, was at a measured distance of 2.25". I have not yet tried to find the absolute maximum detection distance for this target but I don't think it can be much greater.

Here is a photo of the test setup on the concrete floor.


Regards,

Dan