NEW TECH PHOTOS

[Tinkerer]

My quick reply wasn't ample, huh?



My suggestion would be to initially design a light-weight unit for coin/relic hunting utilizing an 8" coil. (Hays has a couple of 8" housings that would work well.)

The TX power should be determined as the design progresses, due to all of the trade-offs between TX power, battery consumption, battery size, unit weight, etc., which are all dependent on the pulse width, pulse frequency, coil parameters and driver circuitry.

A primary goal should be to achieve some form of ferrous/non-ferrous discrimination. Your IB-PI experiments should obviously be the stepping stone for this goal.

Very good observations. Could you post a drawing of the Hays 8" coil housing, so that we can design a coil that fits?

Yes, everything is interdependent. This is why I want to take the decisions now, taking into consideration the wishes of the team members.
As a starting point, lets use a middle ground. Medium everything.
But lets define the numbers.
Say: TX 100uS, max 2A coil current, 12V, 600 PPS?

Tinkerer

[Tinkerer]

Tinkerer, I was looking at the graphs and I am puzzled why there is such a big difference between the cannon ball and the steel lugs and steel nails. Isn't the cannon ball made of iron too? I would suggest to take another photo just with a very rusty nail (because we have much more (mis)chance to find rusty iron than steel lugs). And if it happens that you have some hot rocks in your collection, maybe you should take a shot of those ones too. By the way, I discovered that my hot rock attracts magnets, in fact I can hold the rock suspended by a neodymium magnet.

Regards,
Nicolae

Yes, at first sight it is puzzling.
However, remember that the cannon ball and Alu plate are 30cm from the coil and the other targets are 3mm from the coil.

What I tried to do with the cannon ball, is to simulate the worst Australian ground. As a simulation, I think it worked.
How close it imitates the bad ground, I don't know.

If we keep trying we may succeed. If we do nothing we will never.

Tinkerer

[hobbes_lives]

Here's a link to the Hays coil housings...

http://www.hayselectronics.com/parts.htm

Take a look at B and N, and perhaps W. I don't have technical drawings of the housings, but perhaps exact dimensions can be obtained from Mr. Hays once we've settled on a size. He can be contacted at the bottom of his page.

Say: TX 100uS, max 2A coil current, 12V, 600 PPS?

These values all sound like a reasonable starting point. I realize there's some discrepancy about the optimum pulse width for a given target, but I think 100uS is a good starting point. For a coin-shooter, the pulse width should perhaps be longer, but let's start with 100uS and work from there.

What about sampling timing? Should we set some goals for that from the onset? Perhaps just some basic timings that we can work from for prototyping?

I assume we are planning on using a microcontroller. I have some experience with the dsPIC30f and PIC18f devices.

What about the RX coil and preamp? Are we going with a differential design? I believe that would be optimal.

Also I suggest VCO audio, as human hearing is more sensitive to frequency shifts than to volume changes.

We should throw together a block diagram.

[nick_f]

My quick reply wasn't ample, huh?
A primary goal should be to achieve some form of ferrous/non-ferrous discrimination. Your IB-PI experiments should obviously be the stepping stone for this goal.

How can we analyze the waveforms to obtain the discrimination? Using Sample and Hold circuits followed by a A/D converter? How many bits would be required? How many samples during the Rx response?
In this case, do we need to use a microcontroller?
Or maybe we could use something simpler, like a comparator (or a few comparators and a multiplexer).

Regards,
Nicolae

[Tinkerer]

Tinkerer, I was looking at the graphs and I am puzzled why there is such a big difference between the cannon ball and the steel lugs and steel nails. Isn't the cannon ball made of iron too? I would suggest to take another photo just with a very rusty nail (because we have much more (mis)chance to find rusty iron than steel lugs). And if it happens that you have some hot rocks in your collection, maybe you should take a shot of those ones too. By the way, I discovered that my hot rock attracts magnets, in fact I can hold the rock suspended by a neodymium magnet.

Regards,
Nicolae

Hi Nicolae,

I did a quick repeat on the cannon ball photo. The experiment shows several features.
The TX voltage is not exactly the same. screwing around with the pot of the voltage regulator. It shows the reason for using a regulated TX voltage.
The damping is not exactly the same, dito with the pot.

Anyway the sampling window looks close enough to get the same idea.

Picture # 1 NO TARGET
#2 CANNON BALL
#3 CANNON BALL at 30cm and the silver ring at 3mm. this is to show, that with the simulated hot ground present, the 0.55gram silver ring still gives a similar response.

Tinkerer

[Tinkerer]

Here's a link to the Hays coil housings...

http://www.hayselectronics.com/parts.htm

Take a look at B and N, and perhaps W. I don't have technical drawings of the housings, but perhaps exact dimensions can be obtained from Mr. Hays once we've settled on a size. He can be contacted at the bottom of his page.

These values all sound like a reasonable starting point. I realize there's some discrepancy about the optimum pulse width for a given target, but I think 100uS is a good starting point. For a coin-shooter, the pulse width should perhaps be longer, but let's start with 100uS and work from there.

What about sampling timing? Should we set some goals for that from the onset? Perhaps just some basic timings that we can work from for prototyping?

I assume we are planning on using a microcontroller. I have some experience with the dsPIC30f and PIC18f devices.

What about the RX coil and preamp? Are we going with a differential design? I believe that would be optimal.

Also I suggest VCO audio, as human hearing is more sensitive to frequency shifts than to volume changes.

We should throw together a block diagram.

I have to work with the parts at hand, to get parts takes me months. this is why I suggest we use a PIC16F690 for the first version.
For V2 I would like to use a PIC18F4550 and make the firmware USB programmable, so that we can really experiment with various ways to extract a maximum of information from the signal.
Once we know where we are going, a dsPIC30f will probably open new horizons.
Agreed on the variable pitch audio. However, we need at least 2 tones. One for all targets, one for the presence of FE. It could also be one pitch and modulation thereof for FE. I would like to try to generate this with PWM.

Initial sample timing schedule: 5uS after switch ON, 15uS after switch OFF, 5 uS before TX pulse switch ON. Duration of the samples 5uS.

There is nothing that prevents us using a differential preamp directly from the RX coil. On the breadboard I have 2 limiting diodes only for safety sake for when I goof up.

Tinkerer

[Tinkerer]

How can we analyze the waveforms to obtain the discrimination? Using Sample and Hold circuits followed by a A/D converter? How many bits would be required? How many samples during the Rx response?
In this case, do we need to use a microcontroller?
Or maybe we could use something simpler, like a comparator (or a few comparators and a multiplexer).

Regards,
Nicolae

The PIC16F690, has a 10 bit A/D. That is good enough for the beginning.
We need to build the front end to see exactly where to put the sample points. If they are far enough from each other, there is enough time to sample directly.

Tinkerer

[hobbes_lives]

The PIC16F690, has a 10 bit A/D. That is good enough for the beginning.
We need to build the front end to see exactly where to put the sample points. If they are far enough from each other, there is enough time to sample directly.

It's not the sampling itself that's an issue, it's what's done with the samples after the fact. I suggest we go with a dsPIC30f device right off the bat. The instruction set is designed for DSP, there are way more accumulators to work with, the instruction size is 16-bit, I can go on and on, really. It'd suck to write code for the PIC16, and then have to rewrite it for the dsPIC30 when we realize that our project has outgrown the PIC16.

[Tinkerer]