complete project

[pelanj]

From my point of view, there are a few variants how to make the audio.

1. Standard threshold based sound. Iron would make a low sound and amplitude would be dependend on the brake signal. The non-iron targets would make a high pitched sound with an option of VCO.

2. Silent mode - iron makes low sound, non-iron would make a high sound. Both would have separate volume (like a C.Scope CS6MXi), so that the iron volume can be set to a low volume just to inform about iron in the ground.

We need to make sure that good targets do not show up in the iron range under any condition. My development kit is on the way - it should arrive within next week.

[Tinkerer]

It is a motion machine, however the sweep speed can be really low. There is no ground balance on this model, the only knob sets the sensitivity. I usually set it to a low threshold sound and with a moderate sweep speed try to listen for fast changes in audio, the slow random changes are the ground. The response is not what I would call fast - with a fast sweep it does not react fast, but with medium and low speed it responds very precisely. This is not a problem, it just takes time to adjust to it, I am used to Tesoro lightning speed response

The audio is VCO, with the presence of target both loudness and pitch increase. I like that, however due to instabitily, the changes are perceptible even without a target, which does not allow hunting for "whispers" in the threshold. Except from this audio, there is a clicking sound, but it indicates battery voltage and not target.

I have never used any other PI, but I like the way the clicking works - it is quite easy to hear a change if the basic pace is around 1 Hz. Then,if the signal is really strong, a VCO like signal works the best for me. I have recently got a Tesoro Sand Shark - and I kind of like its audio - no clicking, but very recognisable changes in threshold in VCO mode. I also kind of like a feature of the Minelabs - if it is iron, there is a silence in the threshold tone.

One important question to this new design rises - what would be the masking/unmasking possibilities with discrimination? I mean - would a coin be detected next to a similar sized piece of iron or would it be "hidden" by it? Or would it be possible to let the iron "spit" and have the good target a smooth response for the good ones similar to VLF (esp. Tesoro detectors, which I am used to).

I agree with you, that a good audio can help a lot to make a good detector. So let's start with that part of it.

The TEM_TINKERER_IB-PI gives a positive response for non-magnetic targets and a negative response for magnetic targets. Therefore we can process the responses separately and give each it's on audio output. I have not developed that yet but with your help we may get it done.
This detector is not very sensitive to iron, but very deep and sensitive to non magnetic targets.

We also want a fast response for a motion machine. A non motion machine can use a slow response, but each type of machine needs to be swept at a sweep speed according to it's filters for best results.

A fast response means that the audio must not lag more than 10 to 30ms, depending on coil diameter, behind the target. A slow clicking sound can not manage that.

I am using the PIC16F690 at present. I have it setup for 4 different timing schedules and 8 different TX Taus. 4 different RX Samples. I suggest we start with that and when we find we need more processing power, we change to a different PIC.

Tinkerer

[pelanj]

It is a motion machine, however the sweep speed can be really low. There is no ground balance on this model, the only knob sets the sensitivity. I usually set it to a low threshold sound and with a moderate sweep speed try to listen for fast changes in audio, the slow random changes are the ground. The response is not what I would call fast - with a fast sweep it does not react fast, but with medium and low speed it responds very precisely. This is not a problem, it just takes time to adjust to it, I am used to Tesoro lightning speed response

The audio is VCO, with the presence of target both loudness and pitch increase. I like that, however due to instabitily, the changes are perceptible even without a target, which does not allow hunting for "whispers" in the threshold. Except from this audio, there is a clicking sound, but it indicates battery voltage and not target.

I have never used any other PI, but I like the way the clicking works - it is quite easy to hear a change if the basic pace is around 1 Hz. Then,if the signal is really strong, a VCO like signal works the best for me. I have recently got a Tesoro Sand Shark - and I kind of like its audio - no clicking, but very recognisable changes in threshold in VCO mode. I also kind of like a feature of the Minelabs - if it is iron, there is a silence in the threshold tone.

One important question to this new design rises - what would be the masking/unmasking possibilities with discrimination? I mean - would a coin be detected next to a similar sized piece of iron or would it be "hidden" by it? Or would it be possible to let the iron "spit" and have the good target a smooth response for the good ones similar to VLF (esp. Tesoro detectors, which I am used to).

[Tinkerer]

I have never really worked with PIC, but some ST7 and ATMEL. All the development tools stayed at my previous job, but I will order a PICkit 2 Starter Kit so that I get my own programmer. It should not be too different from ST7/Atmel programming.

The Ebinger is quite OK - it is an EB420 and is for sale on e-bay time to time. It is quite sensitive and I dug 1 DKK out of sand from the same depth as with a Minelab Quattro with the 15x12 SEF DD coil - cca 30 cm/12" It might have been good luck as since then I have had only average 20-25 cm deep finds. The worst part to the EB420 is that it is sensitive to ground sometimes - and at the beach where I hunt the ground response varies quite a lot. Maybe it is small pieces of iron/rust, I do not know. On the other hand, it is really sensitive to small pieces of metal, sometimes I had trouble to see what I have found - tiny wires from fishermen (some kind of steel) very thin and some 5-10 mm long.

I would really like to have a very stable detector as it is tiring after while with the EB.

I would like to identify the problems you have with the EB, so that we can avoid them with our design.
Is the EB a motion machine or is it a static machine?

Does it have ground balance? How do you tune it for the salty sand beach?

Does it have a fast response?

How is the audio? VCO? single frequency with increase of volume for target?

How do you set the audio? Very low volume? Low frequency? What frequency?

Tinkerer

[pelanj]

I have never really worked with PIC, but some ST7 and ATMEL. All the development tools stayed at my previous job, but I will order a PICkit 2 Starter Kit so that I get my own programmer. It should not be too different from ST7/Atmel programming.

The Ebinger is quite OK - it is an EB420 and is for sale on e-bay time to time. It is quite sensitive and I dug 1 DKK out of sand from the same depth as with a Minelab Quattro with the 15x12 SEF DD coil - cca 30 cm/12" It might have been good luck as since then I have had only average 20-25 cm deep finds. The worst part to the EB420 is that it is sensitive to ground sometimes - and at the beach where I hunt the ground response varies quite a lot. Maybe it is small pieces of iron/rust, I do not know. On the other hand, it is really sensitive to small pieces of metal, sometimes I had trouble to see what I have found - tiny wires from fishermen (some kind of steel) very thin and some 5-10 mm long.

I would really like to have a very stable detector as it is tiring after while with the EB.

[Tinkerer]

This project is of a high interest for me and if my knowledge permits, I will try to contribute my part. My field of focus is searching for (small) silver coins in ploughed and rolled fields and pastures. The coin diameters are starting at 10 mm. There is a lot of iron in the fields, so iron discrimination and either fast recovery or non-motion principle for unmasking the non-ferrous target is necessary. My secondary interest is hunting for coins and jewellery in the sand at the Baltic sea beaches. So far I have used VLF and FBS detectors and a military PI by Ebinger on the beaches. I dig targets (2.5 cm coin) from about 12-13" maximum with the detectors I have. I would really like to go deeper in the sand. In the dirt, I would like to get precise iron discrimination for coin sized objects to at least 12", to go deeper than conventional VLF detectors. If there was an option to use large coil for locating large deep objects, it would be great as well.

What I could help with? Unfortunately not much, but I have done some single chip and PLC programming and have a possibility to use MATLAB/Simulink for code generation (unfortunately not Embedded, but standard C/C++).

Hi pelanj,

your knowledge of programming is just what we need.
If you can help writing some code for the PIC16F690 that I am using, it will be of great help.

How did the Ebinger perform on the beach?

All the best

Tinkerer

[pelanj]

This project is of a high interest for me and if my knowledge permits, I will try to contribute my part. My field of focus is searching for (small) silver coins in ploughed and rolled fields and pastures. The coin diameters are starting at 10 mm. There is a lot of iron in the fields, so iron discrimination and either fast recovery or non-motion principle for unmasking the non-ferrous target is necessary. My secondary interest is hunting for coins and jewellery in the sand at the Baltic sea beaches. So far I have used VLF and FBS detectors and a military PI by Ebinger on the beaches. I dig targets (2.5 cm coin) from about 12-13" maximum with the detectors I have. I would really like to go deeper in the sand. In the dirt, I would like to get precise iron discrimination for coin sized objects to at least 12", to go deeper than conventional VLF detectors. If there was an option to use large coil for locating large deep objects, it would be great as well.

What I could help with? Unfortunately not much, but I have done some single chip and PLC programming and have a possibility to use MATLAB/Simulink for code generation (unfortunately not Embedded, but standard C/C++).

[Midas]

Thanks Midas, for the feedback.

Yes, I tried the beat frequency, it is very interesting, but takes quite a bit of a learning curve for the operator to make sense out of the variations.

The 2 frequency ranges we want for the TINKERERS_TEM, are not simultaneous. We want the pitch to go high for gold and go low for iron. With the CD4046, this is very easy. It has a VCO with a center frequency. A positive going signal, like for gold, increases the pitch, a negative signal, like for iron, decreases the pitch.

But, we also want the audio to increase in volume, for a target.

And then we have still another problem. The variation in pitch, of the CD4046, is about one decade. That means half a decade up or down from the center frequency. This seems quite OK, but the signal needs to swing all the way positive or negative for that.
However, our target signals are usually minimal. We want to be able to hear very minute signals as variations in the pitch. Something like 10mV.

Using a Log amplifier as the input to the CD4046 can solve that. All this is analog style.

I am sure it can also be done in digital style, but how?

Ah, and we want a fast response. Very important. We want to hear the target response with not more than about 10ms lag, for small coils, or 50ms lag for large coils. Nearly real time.

A crappy audio can ruin a good detector design.

An excellent audio can enhance the sensitivity and discrimination capability tremendously.

Tinkerer

OK I get it now. I'm also sure what you want can be easily done using a micro controller. Resolving 10mv over 5v only requires a 10bit ADC which is on board most micros these days. Then you just need a little bit of maths, logic, and skills in a suitable programming language and your away. Of course once you have all this flexibility at your disposal you could do a whole lot more with the audio than generate a simple varying frequency. I think it would be cool if you could paint an 'audio picture' of the shape of the response. ie short pulse that rise and fall in frequency in a representation of the actual signal response. Then people could possibly learn to recognize the sound of particular objects. However to do all that you would probably need to digitize all of the samples before they are summed together.

Midas

[Tinkerer]

You should be able to operate the pwm frequency at something over 20khz so you can't hear it and most speakers don't respond faster enough to transmit it anyway. The effect of playing two frequencies at once though is interesting. You end up with a 'beat frequency' that is determined by the difference between the two frequencies. At small differences it can be discerned but with large differences it merges into a strange tone. I guess you have to experiment to work our with what combinations of frequencies best conveys the two streams of information. Speaking of which what exactly is it that your planning on using to set the two frequencies, different sample points?

This site is good:

http://www.animations.physics.unsw.edu.au/jw/beats.htm

It has some examples... unfortunately most of which are rather unpleasant to listen to.
Midas

Thanks Midas, for the feedback.

Yes, I tried the beat frequency, it is very interesting, but takes quite a bit of a learning curve for the operator to make sense out of the variations.

The 2 frequency ranges we want for the TINKERERS_TEM, are not simultaneous. We want the pitch to go high for gold and go low for iron. With the CD4046, this is very easy. It has a VCO with a center frequency. A positive going signal, like for gold, increases the pitch, a negative signal, like for iron, decreases the pitch.

But, we also want the audio to increase in volume, for a target.

And then we have still another problem. The variation in pitch, of the CD4046, is about one decade. That means half a decade up or down from the center frequency. This seems quite OK, but the signal needs to swing all the way positive or negative for that.
However, our target signals are usually minimal. We want to be able to hear very minute signals as variations in the pitch. Something like 10mV.

Using a Log amplifier as the input to the CD4046 can solve that. All this is analog style.

I am sure it can also be done in digital style, but how?

Ah, and we want a fast response. Very important. We want to hear the target response with not more than about 10ms lag, for small coils, or 50ms lag for large coils. Nearly real time.

A crappy audio can ruin a good detector design.

An excellent audio can enhance the sensitivity and discrimination capability tremendously.

Tinkerer

[Midas]

Processor, is the key word. I am sure it can be done by software in the MCU. I just don't know how to do it myself.

The timing is MCU controlled. It really would make sense to use the MCU also for the audio generation.

2 separate frequencies and the volume increase by PWM, that is duty cycle. But how to do the change in frequency at the same time?

Can anyone help?

Tinkerer

You should be able to operate the pwm frequency at something over 20khz so you can't hear it and most speakers don't respond faster enough to transmit it anyway. The effect of playing two frequencies at once though is interesting. You end up with a 'beat frequency' that is determined by the difference between the two frequencies. At small differences it can be discerned but with large differences it merges into a strange tone. I guess you have to experiment to work our with what combinations of frequencies best conveys the two streams of information. Speaking of which what exactly is it that your planning on using to set the two frequencies, different sample points?

This site is good:

http://www.animations.physics.unsw.edu.au/jw/beats.htm

It has some examples... unfortunately most of which are rather unpleasant to listen to.
Midas

[Tinkerer]

Hi Tinkerer,
This is all sounding very good, and being public, I'd be able to build units for other people such as Rob, whose been looking for such a device for years. I will dig up that thread of Aziz's and study it.

I didn't answer your question about rubbish yesterday. There is a strange situation where you can dig 9 bits of rubbish, mainly rusted iron, to 1 bit of gold, these are areas where the Old Boys have been working the gold, and then patches off to the side that the Old Boys have missed will be 9 pieces of gold to one piece of rubbish, usually more modern such as spent ammunition, i.e. non-ferrous. So the ability to notch out iron would be a boon especially when the target is in water, and so much harder to retrieve.

The iron rubbish can be very small so coil i measurements would be useless as a disc solution I believe, especially with ground variations having similar effects.
In fact much of the gold is small, I attach an image of such, the coin is about 23mm across. 2 of the pieces are 0.07 grams, or 1 grain (480 per ozt). All these were detected with the Minelab GPX-4500 between 2 and 12 inches with a 14 x 9 inch elliptical mono coil.

The ultimate would be a machine having software control of the pulse and sample timings so that either large or small gold could be targeted, in fact my dream has been for a machine that automatically cycles through a couple of sequences tailored for both large and small gold. I've tried this on a std PI before and it sadly generated a lot of noise, noise that I believe SETA was developed by Minelab to overcome. They now have the ability to have a full spectrum of timings cycling, but have perhaps held off releasing it.

How sensitive do you feel the TEM can be made, is it perhaps only effective with coin sized targets and above? Is an earlier sample required to detect small targets as in conventional designs?

I know you're using Paul's UniPI, and I've been playing with a PIC24H DSP which is similar, and he's provided some of the source, so I'm sure we can collaborate sufficiently to get some code working.

The fly in the ointment for me at present is the disorder of my house and equipment, and in a couple of months I'll need to move everything out of my house into a shipping container while some walls and ceilings are rebuild.

Kind regards
Kev.

Hi Kev,

We are getting closer to your dream.
I have now developed the circuit to run at 4 different pulse rates.

6400cycle /sec, for smallish coils, like 40cm diameter. About 6A peak coil current.

4500cycles/sec, for coils up to 70cm diameter. About 9.5A coil peak coil current.

3600cycles/sec, for coils up to 1.2m diameter. About 11.5A peak coil current.

3200 cycles/sec, for coils up to 1.5m diameter. About 13A peak coil current.

All coils approximately 300uH.

The pulse rates can be selected with switches.

You can still use your 12V batteries, but will need 2 in series. In compensation, the PSU is very simple, linear, LDO's and low loss.

However, the power consumption for the 13A peak coil current will be about 1.6A or 40W. For the small coils it will be about 17W.

All the best

Tinkerer

[Tinkerer]

Firstly no it not a PI its IB, Secondly it can be set up to audio change with target if required through its processor.

Maybe ive jumped the gun on that one

Regards

Processor, is the key word. I am sure it can be done by software in the MCU. I just don't know how to do it myself.

The timing is MCU controlled. It really would make sense to use the MCU also for the audio generation.

2 separate frequencies and the volume increase by PWM, that is duty cycle. But how to do the change in frequency at the same time?

Can anyone help?

Tinkerer

[satdaveuk]

Thanks for the feedback.

Is the DFX a PI detector?

Does the pitch of the audio change with a target?

Tinkerer

Firstly no it not a PI its IB, Secondly it can be set up to audio change with target if required through its processor.

Maybe ive jumped the gun on that one

Regards

[Tinkerer]

Hi Tinker
The DFX Whites circuit would be perfect for the increase in sound , problem is i dont have it..... unless you want me to dismantal mine to have a peek?? and do a bit of reverse
You realize though thats my babe the bees, goes places alot of even the most exspensive dont, been driven by me the DFX man of the UK, they dont come any better.
.
On a serious note the way the DFX if set up correctly creaps up in volume as your getting near the selected buried item is amazing even if with crap all around it.
Its hard to explain soon as it locks in just takes you there deep, who needs a pinpointer when using the DFX, its in a world of its own, any experinced users know what planet im on.

Regards

Thanks for the feedback.

Is the DFX a PI detector?

Does the pitch of the audio change with a target?

Tinkerer

[satdaveuk]

Hi Tinker
The DFX Whites circuit would be perfect for the increase in sound , problem is i dont have it..... unless you want me to dismantal mine to have a peek?? and do a bit of reverse
You realize though thats my babe the bees, goes places alot of even the most exspensive dont, been driven by me the DFX man of the UK, they dont come any better.
.
On a serious note the way the DFX if set up correctly creaps up in volume as your getting near the selected buried item is amazing even if with crap all around it.
Its hard to explain soon as it locks in just takes you there deep, who needs a pinpointer when using the DFX, its in a world of its own, any experinced users know what planet im on.

Regards