complete project

[Tinkerer]

No help?

No ideas?

Well, so I must put on my own thinking cap. Watch out for the smoke!

Going back to the audio:

The VCO of the CD4046 gives about one decade of frequency variation. We could use it from about 100Hz to 1100Hz and making the 0V at about 300Hz. This leaves a larger pitch variation for the gold, positive going, and a smaller variation for the iron, negative going.

300Hz is also easy on the ear, for a continuous sound, indicating the proper function of the detector and also indicating slow, small variations in the ground response.

We will also LP filter the 50% duty cycle square wave, to make the sound less harsh.

The problem is now, how to make this pitch variation non linear.

We want a fast response to minute targets. Large or very close targets give such a strong response that it will saturate everything in reach.

A log amplifier seems to be the answer. This will give a relatively larger change in pitch for the small targets.

The CD4046 also has a mute pin. This is useful for the switch ON time where many detectors scream at maximum ear shattering volume.

OK so let's draw up a circuit and test it.

What about the increase in volume? This is another wish on the list. A target should produce a change in pitch and also an increase in volume. So this is the next problem to solve.

Any suggestions?

Tinkerer

[Tinkerer]

Tinkerer is there a high resolution version of your schematic available anywhere please ?

I think at the post below, is the last version. EAGLE FILES.
I do not recommend to build it. I posted it for discussing it's good and not so good parts.
We might re-use some parts in the TINKERERS_TEM_IB-PI.

For example the 2 tone audio (needs improvement) with the mute and dead band.

http://www.geotech1.com/forums/showpost.php?p=99524&postcount=6

Enjoy

Tinkerer

[6666]

Tinkerer is there a high resolution version of your schematic available anywhere please ?

[Tinkerer]

Hi
Its looking good and have gone such along way in short time
Has anybody made any lay files for this project yet ? or am i been cheeky/lazy
Serious level taking my hat off to you Moodz and everyone else on here who has worked so hard on a bespoke design such as this.

Regards

Thank you for your kind words, Dave.

Below is an older circuit and board. There are some good things in that circuit, but others could use some improvements. For example the Audio. The response is not dynamic enough.
The use of the CD4046 could be much improved.

Tinkerer

[satdaveuk]

Hi
Its looking good and have gone such along way in short time
Has anybody made any lay files for this project yet ? or am i been cheeky/lazy
Serious level taking my hat off to you Moodz and everyone else on here who has worked so hard on a bespoke design such as this.

Regards

[Tinkerer]

Hi Tinkerer,
Nothing like an image or a schematic to get the theory across. Thank you.

I've been thinking about the audio and believe that the volume and frequency should increase relative to target strength. I don't think the clicking bone type is responsive enough, nor is the simple volume increase, but the two combined creates a strong impression.

I was looking at the raw Rx signal before filtering and was wondering what would happen if this was synchronously demodulated with a precision high speed PLL chip? The result maybe a very interesting decay curve which when sampled at 2 or 3 points could provide some extra information about the target. At least it should provide a consistent almost noise free signal.

I'm ashamed to say I've not built the 12V frontend you posted yet, and it's so simple too....next week I hope. I've got some PLL chips kicking around somewhere among my bits too. Tomorrow I'll have a hunt for some high voltage caps among my junk.

All the very best
Kev.

Hi Kev,

thanks for the feedback.

I agree with you that a good audio can make a big difference in the detecting.
We want a fast response detector. For a 40cm coil, motion detector, the response should be about 100ms. That is, we want to hear that nugget within 10cm of sweep. Considering a sweep speed standard of 1m/s.
For a non motion detector the sweep speed can be considered quite a bit lower. The response can then be slower too.
A slow response means we can integrate more samples for better S/N.

Ear phones or bone conductors can click. For a fast response the minimum click speed would need to be about 15Hz. But the clicking does not work well with loudspeakers.

The human ear has some quirks that need to be taken into account. We can easily hear a difference in frequency, but a relative large increase in loudness is needed to be distinguishable.

Then we need to consider the annoyance of a continuous sound, yet, if we hear no sound at all, we worry that the detector does not function.

What is a good minimum frequency? 600 to 800Hz? A square wave makes a very harsh sound. This can be softened with a LP filter. A sine wave needs more power to be heard.

A piezoelectric element is very power efficient, but makes a horrible sound.

How loud does the speaker need to be? If there is wind then the speaker needs to be quite loud. On the other hand, the modern earphones are capable of beautiful sound at low power.

One frequency?
2 frequencies? High-low?

I really know very little about making a good audio for a detector. Some help would be greatly appreciated.

You were looking at the RX signal.... could you give me a link to what you were looking at? I have posted so many scope pictures and simulations, I am lost in the maze.

I can show the raw, unfiltered signals of large targets on the scope, but for small target behavior it is better to look at a simulation.

How do you feel about a 40cm diameter coil? I think it would come close to your needs.

All the best

Tinkerer

[Davor]

was wondering what would happen if this was synchronously demodulated with a precision high speed PLL chip?

I'm afraid you'll have a very steep learning curve about the PLL phase noise implications. It should do miracles on IB phase detection though.

[Kev]

Hi Tinkerer,
Nothing like an image or a schematic to get the theory across. Thank you.

I've been thinking about the audio and believe that the volume and frequency should increase relative to target strength. I don't think the clicking bone type is responsive enough, nor is the simple volume increase, but the two combined creates a strong impression.

I was looking at the raw Rx signal before filtering and was wondering what would happen if this was synchronously demodulated with a precision high speed PLL chip? The result maybe a very interesting decay curve which when sampled at 2 or 3 points could provide some extra information about the target. At least it should provide a consistent almost noise free signal.

I'm ashamed to say I've not built the 12V frontend you posted yet, and it's so simple too....next week I hope. I've got some PLL chips kicking around somewhere among my bits too. Tomorrow I'll have a hunt for some high voltage caps among my junk.

All the very best
Kev.

[Davor]

It will be nice to see how TEM works with log-weighted PI Rx. So far I simulated it only against Surf PI frontend.

[Tinkerer]

Thanks for the explaination. Is there any technical papers or info on TEM?

Regards

http://www.geotech1.com/forums/showpost.php?p=111998&postcount=819

There are several posts there giving ample descriptions.

Attached is a pdf with a description of the TEM method used in Geophysics, with good graphs.

I have posted dozens of LTSpice simulations over the years, but they are scattered all over the forum.

All the best

Tinkerer

[Altra]

Thanks for the explaination. Is there any technical papers or info on TEM?

Regards

[Tinkerer]

Tinkerer,

Interesting project you have here. One thing that dawned on me. When considering your PRR, lets say 10khz or 100us. There isn't much time to take a ground sample. For tiny targets, not a problem. But larger targets will start to null. Or maybe I don't understand the TEM ground balancing method? Using conventional gb methods you may have to stay
below 5khz or do some kind of burst mode of short pulses and balance it against a long off period?

Thanks

Hi Altra,

Thanks for the feedback.

In some ways, the TINKERERS TEM method seems not much different from the traditional PI method. However, in fact there are many things very different, some more obvious than others.

For example: The cycle time. With a traditional PI, with 5000 PPS and 50us TX pulse, a 150us target will have decayed about 63% at the time of the beginning of the next cycle.

With the TINKERERS TEM, I use a forced decay (for want of a better name). It makes the target eddy current to decay about 5 times faster.
Therefore, with a 5 times faster cycle repetition rate I have about the same percentage of eddy current decay.

As far as the GB, I am still experimenting with different methods.

Tinkerer

[Altra]

Tinkerer,

Interesting project you have here. One thing that dawned on me. When considering your PRR, lets say 10khz or 100us. There isn't much time to take a ground sample. For tiny targets, not a problem. But larger targets will start to null. Or maybe I don't understand the TEM ground balancing method? Using conventional gb methods you may have to stay
below 5khz or do some kind of burst mode of short pulses and balance it against a long off period?

Thanks

[Tinkerer]

Depends on where you are at on the charging curve. Also, lowering the TX peak current can make the recovery faster, allowing you to sample sooner. So there may be a net benefit to pulse width reduction for detecting tiny shallow nuggets. Eric's GoldQuest has a PPR of 10kHz, not sure about the pulse width but it would have to be 50us or less.

- Carl

Hi Carl,

very few things have not been tried before. In this case, I believe that Eric Foster experimented with something similar, using SCR's. What may not have working too well, with SCR's, seems to work fine with modern Mosfets.

I am sampling at the peak target response, not the weak remnants of a signal after some delay. However, I totally agree, that there are many factors involved and that it is always a compromise between all of these factors that gives the best result.

Being weak in Mathematics, it seems that I can only find this compromise, or sweet spot by experimenting. Ahh, there are so many possible variations, it just takes a lot of tinkering.

Fixing some of the variables reduces the amount of possible variations. But, which variable should I fix?

Coil diameter? Inductance?

Power? Tx voltage? TX charge time?

PPR?

Sample width?

Pre-amp supply to get good headroom?

To name just a few. Any change in any one of these factors changes the sweet spot to a different place.

I use a near linear charge curve, by the way.

Tinkerer

[Carl-NC]

However, with the same coil size, reducing the TX pulse width by half, also reduces the magnetic field density by half. This is not good for small targets. The most important thing for generating a good response from very small targets, is high magnetic field density.

Depends on where you are at on the charging curve. Also, lowering the TX peak current can make the recovery faster, allowing you to sample sooner. So there may be a net benefit to pulse width reduction for detecting tiny shallow nuggets. Eric's GoldQuest has a PPR of 10kHz, not sure about the pulse width but it would have to be 50us or less.

- Carl