A
Anonymous
Guest
Hi,
Been thinking about ground balance for the GQ. I
Been thinking about ground balance for the GQ. I
Find's Treasure Forums
Welcome to Find's Treasure Forums, Guests!
You are viewing this forums as a guest which limits you to read only status.
Only registered members may post stories, questions, classifieds, reply to other posts, contact other members using built in messaging and use many other features found on these forums.
Why not register and join us today? It's free! (We don't share your email addresses with anyone.) We keep email addresses of our users to protect them and others from bad people posting things they shouldn't.
Click here to register!
Need Support Help?
Cannot log in?, click here to have new password emailed to you
A
Anonymous
Guest
Eric has buit a PI with ground balance control called the GoldScan, has built it for years and is sold in Australia. Why it has never come here or isn't incorporated in the Gold Quest I do not know. His detectors are very stable, but can't work all grounds, even though the SAT control helps in that regard. Reg has one and used it out West in his testing.
A
Anonymous
Guest
Hi,
Some vital bits I missed out. The long pulse Rx sample is 30us wide while the short pulse
sample is 10us. Only take one sample of each, (not enough channels on GQ for more, although could add more) have the -A channel integrating with respect to +A channel at one third the rate.
This will stretch the time constant of the short sample to match the long sample (More Candy?).
If ferrite only has been detected the signals will cancel since the ground signal from the 30us pulse at one third the value of the signal from the 90us pulse, will after strecting be equivalent.
(gain control for tweaking balance)
If a conductive metal target responds within the matrix, the target signal component of both channels will be equivalent irrespective of pulse length, thus after stretching the signal processed via the -A channel will be greater, since it is changing more slowly with respect to time. It will be detected....?
It's late at night, and I'm probably way off in the weeds. Does it make any sense, can I still
preserve any measure of credence ? Maybe I should keep my mouth shut <img src="/metal/html/wink.gif" border=0 width=15 height=15 alt="
">
Some vital bits I missed out. The long pulse Rx sample is 30us wide while the short pulse
sample is 10us. Only take one sample of each, (not enough channels on GQ for more, although could add more) have the -A channel integrating with respect to +A channel at one third the rate.
This will stretch the time constant of the short sample to match the long sample (More Candy?).
If ferrite only has been detected the signals will cancel since the ground signal from the 30us pulse at one third the value of the signal from the 90us pulse, will after strecting be equivalent.
(gain control for tweaking balance)
If a conductive metal target responds within the matrix, the target signal component of both channels will be equivalent irrespective of pulse length, thus after stretching the signal processed via the -A channel will be greater, since it is changing more slowly with respect to time. It will be detected....?
It's late at night, and I'm probably way off in the weeds. Does it make any sense, can I still
preserve any measure of credence ? Maybe I should keep my mouth shut <img src="/metal/html/wink.gif" border=0 width=15 height=15 alt="
">
A
Anonymous
Guest
Reading your post, I just thought of something. If a PI detector could transmit multiple frequencies, say 10, 15, 20, 25 us and so on, an onboard computer could notice the difference in the response of a target and eliminate those that respond better to the higher frequencies such as iron. A automatic reverser disc or sorts.
Just and idea.
Just and idea.
A
Anonymous
Guest
Hi Kev,
Boy, you have presented a lot of work in an effort to obtain some form of ground balance (GB). Obviously, one could try it and see just how it works, but I would start out with a very basic design first.
From there, one can see just what the advantages or disadvantages might be of using different techniques you mentioned.
The most basic approach would be to take a later sample and simply amplify it and then subtract it from the primary sample. By making the amplification level adjustable, one can inject more or less signal for GB. At or near complete ground balance, the ground signal is minimial, but there is a significant reduction in targets having a similar time constant as the ground. Certain other targets will generate a negative response when very near the point of ground balance also. So, target signals become more complex.
However, if one just approaches the point of the best ground balance or some level less, all signals simplify, meaning there still is a fairly significant reduction in the ground response with much less reduction or alteration of certain other desireable signals. At least, this is what I have found to be true.
From this point, one can determine just what might work satisfactorily for the ground conditions they encounter. Once these basics are known, then one might want to expand to a multiple pulse width just to see what can be gained, if anything.
I have found the simple subtract system works extremely well for all ground conditions I encounter. Now, when the GB is combined with a DD coil, the GQ is both quiet, noise and ground signal wise, and is very sensitive.
So, I have not seen any real need to try to expand the basic GB design, since it functions quite well whether I use a mono coil or the DD.
Reg
Boy, you have presented a lot of work in an effort to obtain some form of ground balance (GB). Obviously, one could try it and see just how it works, but I would start out with a very basic design first.
From there, one can see just what the advantages or disadvantages might be of using different techniques you mentioned.
The most basic approach would be to take a later sample and simply amplify it and then subtract it from the primary sample. By making the amplification level adjustable, one can inject more or less signal for GB. At or near complete ground balance, the ground signal is minimial, but there is a significant reduction in targets having a similar time constant as the ground. Certain other targets will generate a negative response when very near the point of ground balance also. So, target signals become more complex.
However, if one just approaches the point of the best ground balance or some level less, all signals simplify, meaning there still is a fairly significant reduction in the ground response with much less reduction or alteration of certain other desireable signals. At least, this is what I have found to be true.
From this point, one can determine just what might work satisfactorily for the ground conditions they encounter. Once these basics are known, then one might want to expand to a multiple pulse width just to see what can be gained, if anything.
I have found the simple subtract system works extremely well for all ground conditions I encounter. Now, when the GB is combined with a DD coil, the GQ is both quiet, noise and ground signal wise, and is very sensitive.
So, I have not seen any real need to try to expand the basic GB design, since it functions quite well whether I use a mono coil or the DD.
Reg
A
Anonymous
Guest
I believe that in order to preserve a stable searching state that you will need to use only one or two frequencies as is currently done.
Here is the twist.
You could have a button operated mode change switch to activate once a target is located that functionally ramps up, in a stairstep-like pattern, a series of pulse delays at 10, 15, 20, 25, 30, 35, 40, 50, 100, 200, 300us (or some interval that produces significant results). Here is where bench and field research is required. These pulse intervals are timed to change one per second. I chose one second as the most convenient time to make a deliberate single coil scan over a target location in this stairstep mode. This interval could be made adjustable to accomodate user preferences. There should be a brief audible blip, at the end of each sweep (to help center the target between blips), every second to synchronize the sweeps over the target. Each sweep is analyzed on a bar graph that shows the results on a simple bar graph intensity meter with a column of bars of each of the specified time delays.
With this design you would learn the response signals for particular targets. This should produce a very repeatable finger print-like response for particular known targets.
All of this can be designed with what we currently know about PI detectors. The only area that needs research is the coice of the particular stairstep intervals to produce that unique target response.
Here are some questions I would first seek to answer with this prerequsite rsearch.
1. What is the minimum number of intervals necessary to produce unambigous ientification?
2. What is the interval betweeen stairsteps that effectively contributes to the discrimination process.
3. How many levels for each stairsteped interval are practical or necessary to produce a good visual discrimination to targets that traditional MDs respond as being close... such as tabs and nicels.
This is a functional description of a design that someone who knows PIC design should be able to program very easily.
First we need some dialog from those who read this forum and know the answers to the above questions or suggest better questions.
Maybe we could have a group design that could be eventually built. This could become Hammerhead II or III (Hammerhead is the current PI project on the Geotech web page).
I hope this stimulates a good dialog . Eric and Reg please jumps in!
bbsailor
Here is the twist.
You could have a button operated mode change switch to activate once a target is located that functionally ramps up, in a stairstep-like pattern, a series of pulse delays at 10, 15, 20, 25, 30, 35, 40, 50, 100, 200, 300us (or some interval that produces significant results). Here is where bench and field research is required. These pulse intervals are timed to change one per second. I chose one second as the most convenient time to make a deliberate single coil scan over a target location in this stairstep mode. This interval could be made adjustable to accomodate user preferences. There should be a brief audible blip, at the end of each sweep (to help center the target between blips), every second to synchronize the sweeps over the target. Each sweep is analyzed on a bar graph that shows the results on a simple bar graph intensity meter with a column of bars of each of the specified time delays.
With this design you would learn the response signals for particular targets. This should produce a very repeatable finger print-like response for particular known targets.
All of this can be designed with what we currently know about PI detectors. The only area that needs research is the coice of the particular stairstep intervals to produce that unique target response.
Here are some questions I would first seek to answer with this prerequsite rsearch.
1. What is the minimum number of intervals necessary to produce unambigous ientification?
2. What is the interval betweeen stairsteps that effectively contributes to the discrimination process.
3. How many levels for each stairsteped interval are practical or necessary to produce a good visual discrimination to targets that traditional MDs respond as being close... such as tabs and nicels.
This is a functional description of a design that someone who knows PIC design should be able to program very easily.
First we need some dialog from those who read this forum and know the answers to the above questions or suggest better questions.
Maybe we could have a group design that could be eventually built. This could become Hammerhead II or III (Hammerhead is the current PI project on the Geotech web page).
I hope this stimulates a good dialog . Eric and Reg please jumps in!
bbsailor
A
Anonymous
Guest
Just changing the delay may not be enough to obtain good discrimination. You will need to change the TX pulse length to stimulate responses in different metals. The above stairstep concept should be expanded to provide the ability to also step through various TX pulse lengths to help identify various targets. The same questions apply except for substituting TX length for pluse delay.
You would need two push buttons; one for stairstepping through TX pulse length and the other for stairstepping through TX delay. Another additional capability is that once you have a variable response either between various TX pulse widths or TX/Receive delays, you could select the level that you want to further analyze in the other mode (meaning the TX pulse length or TX/Receive delay). This should provide a wealth of identification information. All of this is possible with a little creative PIC programming. If I were in the mood to pursue a post graduate degree in engineering, I might want to consider this as a possible research project (Hint for those graduate students lurking out there).
This post script is just the result of a little iterative reflection gained through considering the dialogs on this PI forum.
bbsailor
You would need two push buttons; one for stairstepping through TX pulse length and the other for stairstepping through TX delay. Another additional capability is that once you have a variable response either between various TX pulse widths or TX/Receive delays, you could select the level that you want to further analyze in the other mode (meaning the TX pulse length or TX/Receive delay). This should provide a wealth of identification information. All of this is possible with a little creative PIC programming. If I were in the mood to pursue a post graduate degree in engineering, I might want to consider this as a possible research project (Hint for those graduate students lurking out there).
This post script is just the result of a little iterative reflection gained through considering the dialogs on this PI forum.
bbsailor
A
Anonymous
Guest
Hi Mayumi,
It's the basis of of what I'm saying that different pulse widths (frequencies) have a corresponding effect on the target response. But for now I'm only wanting to remove a measure of ground response not so much a disc. I'm sure others are working on it though, and may use this approach.
Timing maybe the major constraint on any design, since to get a target to respond it takes time to build up eddy currents within the target. The more frequencies used the longer the time to gather information, the longer the time the slower the response of the detector. However as bbsailor has pointed out below once a target is located then one could switch to a disc mode where time is no longer a constraint, and a slow sweep is not a poblem.
Like searching for nuggets in VLF and flipping over to IB to disc a target.
Cheers
Kev.
It's the basis of of what I'm saying that different pulse widths (frequencies) have a corresponding effect on the target response. But for now I'm only wanting to remove a measure of ground response not so much a disc. I'm sure others are working on it though, and may use this approach.
Timing maybe the major constraint on any design, since to get a target to respond it takes time to build up eddy currents within the target. The more frequencies used the longer the time to gather information, the longer the time the slower the response of the detector. However as bbsailor has pointed out below once a target is located then one could switch to a disc mode where time is no longer a constraint, and a slow sweep is not a poblem.
Like searching for nuggets in VLF and flipping over to IB to disc a target.
Cheers
Kev.
A
Anonymous
Guest
Hi bbsailor,
Wow that's a lot of research, I'm thinking your going to need more processing power than a PIC can provide, 400MHz Motorola, or a DSP.
My time is limited unfortunately, hence my desire to tweak the GQ for hotter ground. A forum project would be great.
Cheers
Kev.
Wow that's a lot of research, I'm thinking your going to need more processing power than a PIC can provide, 400MHz Motorola, or a DSP.
My time is limited unfortunately, hence my desire to tweak the GQ for hotter ground. A forum project would be great.
Cheers
Kev.
A
Anonymous
Guest
Hi Reg,
Good to hear from you again. I realise that I've made a couple of silly mistakes, funny how you don't notice at the time, only after it's posted. I must brush up on my integrator theory, I'm from a digital background.
Reg, does your GB system amplify the noise and low frequency rejection sample taken just prior to the main pulse? Do you filter this also?
Thanks
Kev.
Good to hear from you again. I realise that I've made a couple of silly mistakes, funny how you don't notice at the time, only after it's posted. I must brush up on my integrator theory, I'm from a digital background.
Reg, does your GB system amplify the noise and low frequency rejection sample taken just prior to the main pulse? Do you filter this also?
Thanks
Kev.
A
Anonymous
Guest
Hi Kev,
Basically, a differential integrator is used for the main sample and a duplicate diff amp circuit is built for the ground signal subtract sample. The main difference is, if the main sample is taken at 10 usec, then the subtract sample is taken at 20 usec. Each of the integrators also use a much later sample as a means of eliminating the earth field effect and low freq noise. This later sample is fed into the + input of the diff amps. As an example, the main diff integrator may have a 10 usec main sample on the - input and maybe an 80 usec sample on the + input. The second diff integrator would have its main sample at 20 usec and the later sample at 90 usec.
Now, the output of the diff amp used for the ground subtract signal is then fed into a simple amplifier with adjustable gain. The output of this amp is then subtracted from the main diff amp signal before the next stage. There is minimal filtering on this stage to assure or minimize any potential lag in signals do to a difference in filtering.
The differential amp stages themselves have filtering which reduces the noise considerably. However, if the ground subtract signal is amplified a lot to provide the gain necessary for perfect ground balance, then yes, there will be more noise introduced.
In my case, I don't use full GB so the noise isn't that much of a factor. In fact, it is extremely low most of the time. By doing it this way, I don't have much of an increase in noise, but still have a very significant reduction in ground response.
Reg
Basically, a differential integrator is used for the main sample and a duplicate diff amp circuit is built for the ground signal subtract sample. The main difference is, if the main sample is taken at 10 usec, then the subtract sample is taken at 20 usec. Each of the integrators also use a much later sample as a means of eliminating the earth field effect and low freq noise. This later sample is fed into the + input of the diff amps. As an example, the main diff integrator may have a 10 usec main sample on the - input and maybe an 80 usec sample on the + input. The second diff integrator would have its main sample at 20 usec and the later sample at 90 usec.
Now, the output of the diff amp used for the ground subtract signal is then fed into a simple amplifier with adjustable gain. The output of this amp is then subtracted from the main diff amp signal before the next stage. There is minimal filtering on this stage to assure or minimize any potential lag in signals do to a difference in filtering.
The differential amp stages themselves have filtering which reduces the noise considerably. However, if the ground subtract signal is amplified a lot to provide the gain necessary for perfect ground balance, then yes, there will be more noise introduced.
In my case, I don't use full GB so the noise isn't that much of a factor. In fact, it is extremely low most of the time. By doing it this way, I don't have much of an increase in noise, but still have a very significant reduction in ground response.
Reg
A
Anonymous
Guest
Hi Reg,
Thanks for that info. I'll copy that, and experiment with a time control to alter integration rates of the ground signal. May take some fiddling around though, Erics got it finely balanced by the look of it. I'll let you know how I get on.
Regards
Kev.
Thanks for that info. I'll copy that, and experiment with a time control to alter integration rates of the ground signal. May take some fiddling around though, Erics got it finely balanced by the look of it. I'll let you know how I get on.
Regards
Kev.
A
Anonymous
Guest
Kev. I read your posts a few times and I'm not sure what you said. <img src="/metal/html/smile.gif" border=0 width=15 height=15 alt="">
If you look at Candy's idea in it's simplest form then you will see that he asserts that any mineral that obeys his concept will have a ferrite relaxation decay (FRD) dependant on the pulse length and there will be a fixed ratio for these, regardless of their decay times.
If you still planned to go ahead with your 2-pulse length idea then you could learn a lot if you built a circuit to attempt the following. Subtract a late sample in the long pulse from an early sample in the short pulse to get a form of "conventional" ground balance.
This could mean making a sacrificial long pulse, 4 times longer than the short pulse, sample at 10us in the short one and adjust the position of the long pulse sample to obtain a ground null. Your aim would be to do so with unity gain, equal length samples and subtract the EFE as well, simplifying the circuit.
In other words, if any FRD poked it's head into the early sample in the short pulse then it should be doing exactly the same in the later long pulse sample if you positioned it correctly and the fact that you are using unity gain would mean that the EFE was also cancelled.
This idea also allows you to get some measurements to see what you are in for if you decide to attempt something more complex and in-line with what Candy is doing, he universally cancels FRDs that have wildly varying decay times and with fixed circuitry and math. The GB control, whether manual or auto, is only there for the bits that don't quite fit the fixed math. (If NZ doesn't have our bad ground then this is hard to appreciate and you might find that carrying a heavy, battery chewing monster around a bit silly and of little benefit.
Some have suggested using a certain red brick to test their GB design but throw a stack of our rocks down with it and you will soon appreciate the term "universal canceling").
Now if you were to take the above proposal a bit further using the same 4X pulse length system and still sample at 10us in the short pulse then you may conclude there should be only be one fixed sample point in the long pulse that will give a fixed ratio for any FRDs appearing in either pulse and it would be at 40us in this case if you had attained the ideal. This would also result in having a fixed gain that (ideally) would not change for subtracting all FRDs. Think about this and I think you can take the idea of universal cancellation further from there.
Now for the bit that should turn you off the idea. You must minimize series resistance in the coil circuit to get the relationship Candy relies on and this causes a lot of problems on it's own. Added resistance acts like a ballast resistor and the effect from the FRDs won't appear much different in both pulse lengths and it upsets other relationships also. Switching at the fet gate must be done correctly or you will get similar problems that will add unwanted effects and the list goes on. If you use some of the switching methods proposed in some designs to turn the fet on then you will see a noticeable difference in the waveform at the gate for different pulse lengths.
The downside is that you have to draw large current unless you can figure out a way around this obstacle first. Your present front end and battery pack will need a major redesign.
The quickest way to upset the fixed math in a ML pi is to wind a coil with noticeably higher DC resistance even if it shows a better "depth" measurement in air. It may still appear to auto GB on a 2200 or GP but will be a lot noisier in hot, varying ground or even unusable with the manual balance series depending on how much you have increased the coils resistance. A test would be to try it on a manual model with known 'hot" minerals that the detector normally totally ignored.
I think that if you want to upset the GB concept Reg proposes of subtracting one late sample from one early one in any individual pulse length that currently uses high series resistance in the coil circuit, then remove the series resistor and the minerals that decay too early to bother it now should poke their head into the early sample and some that appear crunched up rather close in "time" should then have varying decay times. Lower the actual coil resistance and the effect should be even more exaggerated. Reg's proposal seems to rely somewhat on having high series resistance to keep a lot of the hotter minerals bundled close together and some to early to interfere at all.
My first proposal would have these FRDs spread out over a longer time so this then would have noticeable disadvantages also unless your ground minerals are well blended. It is only a form of conventional GB.
There is a lot to consider and I reckon in the end you will have to rely on precise measurements and fudging pulse properties to get anything to do with universal GB to work, regardless of the attack.
Good luck.
Rob.
">If you look at Candy's idea in it's simplest form then you will see that he asserts that any mineral that obeys his concept will have a ferrite relaxation decay (FRD) dependant on the pulse length and there will be a fixed ratio for these, regardless of their decay times.
If you still planned to go ahead with your 2-pulse length idea then you could learn a lot if you built a circuit to attempt the following. Subtract a late sample in the long pulse from an early sample in the short pulse to get a form of "conventional" ground balance.
This could mean making a sacrificial long pulse, 4 times longer than the short pulse, sample at 10us in the short one and adjust the position of the long pulse sample to obtain a ground null. Your aim would be to do so with unity gain, equal length samples and subtract the EFE as well, simplifying the circuit.
In other words, if any FRD poked it's head into the early sample in the short pulse then it should be doing exactly the same in the later long pulse sample if you positioned it correctly and the fact that you are using unity gain would mean that the EFE was also cancelled.
This idea also allows you to get some measurements to see what you are in for if you decide to attempt something more complex and in-line with what Candy is doing, he universally cancels FRDs that have wildly varying decay times and with fixed circuitry and math. The GB control, whether manual or auto, is only there for the bits that don't quite fit the fixed math. (If NZ doesn't have our bad ground then this is hard to appreciate and you might find that carrying a heavy, battery chewing monster around a bit silly and of little benefit.
Some have suggested using a certain red brick to test their GB design but throw a stack of our rocks down with it and you will soon appreciate the term "universal canceling").
Now if you were to take the above proposal a bit further using the same 4X pulse length system and still sample at 10us in the short pulse then you may conclude there should be only be one fixed sample point in the long pulse that will give a fixed ratio for any FRDs appearing in either pulse and it would be at 40us in this case if you had attained the ideal. This would also result in having a fixed gain that (ideally) would not change for subtracting all FRDs. Think about this and I think you can take the idea of universal cancellation further from there.
Now for the bit that should turn you off the idea. You must minimize series resistance in the coil circuit to get the relationship Candy relies on and this causes a lot of problems on it's own. Added resistance acts like a ballast resistor and the effect from the FRDs won't appear much different in both pulse lengths and it upsets other relationships also. Switching at the fet gate must be done correctly or you will get similar problems that will add unwanted effects and the list goes on. If you use some of the switching methods proposed in some designs to turn the fet on then you will see a noticeable difference in the waveform at the gate for different pulse lengths.
The downside is that you have to draw large current unless you can figure out a way around this obstacle first. Your present front end and battery pack will need a major redesign.
The quickest way to upset the fixed math in a ML pi is to wind a coil with noticeably higher DC resistance even if it shows a better "depth" measurement in air. It may still appear to auto GB on a 2200 or GP but will be a lot noisier in hot, varying ground or even unusable with the manual balance series depending on how much you have increased the coils resistance. A test would be to try it on a manual model with known 'hot" minerals that the detector normally totally ignored.
I think that if you want to upset the GB concept Reg proposes of subtracting one late sample from one early one in any individual pulse length that currently uses high series resistance in the coil circuit, then remove the series resistor and the minerals that decay too early to bother it now should poke their head into the early sample and some that appear crunched up rather close in "time" should then have varying decay times. Lower the actual coil resistance and the effect should be even more exaggerated. Reg's proposal seems to rely somewhat on having high series resistance to keep a lot of the hotter minerals bundled close together and some to early to interfere at all.
My first proposal would have these FRDs spread out over a longer time so this then would have noticeable disadvantages also unless your ground minerals are well blended. It is only a form of conventional GB.
There is a lot to consider and I reckon in the end you will have to rely on precise measurements and fudging pulse properties to get anything to do with universal GB to work, regardless of the attack.
Good luck.
Rob.
A
Anonymous
Guest
Hi Robby and Kev,
Your post is interesting, as I have been doing some tests on Australian ironstone over the last few days with a high power transmitter arrangement I am trying out. Ground balancing is not a prime requirement for what I am trying to do, but I always like to do additional tests as new circuitry is evolved. The equipment is still in its early stages, but I have a transmitter that can pulse up to 10A peak into a low resistance winding. Other features are that the pulse width can be varied with the pulse rate tied in so that, for a resistive load, the mean current stays the same. However, with a low resistance coil, the peak current is inductance limited, so a constant current circuit has been added to compensate for this. In other words, as the pulse is shortened and the peak current becomes less, due to the switch off occurring earlier on the coil time constant, the TX voltage increases so that the switch off current is the same as for a long pulse, which equals, say, five times the coil time constant. The coil used for these tests was, in fact, a Minelab SD 11in mono coil.
Looking at the receiver waveform on a scope, with a box of ironstone on the coil, you can see the FRD very clearly, and at early times it is saturating the receiver. Changing the TX pulse width from 150uS to 500uS does not make one bit of difference to the length or amplitude of the ironstone signal. Over the years I have tried similar tests with different circuit arrangements and have never observed any pulse width dependant decay signal on non-conductive iron bearing material. This is at odds with what Candy and others report.
The same tests conducted with non-ferrous metal objects, do show a change in both the decay shape and the time constant, up to the point where the width of the transmitter pulse equals that of the objects decay time. From then on it doesn
Your post is interesting, as I have been doing some tests on Australian ironstone over the last few days with a high power transmitter arrangement I am trying out. Ground balancing is not a prime requirement for what I am trying to do, but I always like to do additional tests as new circuitry is evolved. The equipment is still in its early stages, but I have a transmitter that can pulse up to 10A peak into a low resistance winding. Other features are that the pulse width can be varied with the pulse rate tied in so that, for a resistive load, the mean current stays the same. However, with a low resistance coil, the peak current is inductance limited, so a constant current circuit has been added to compensate for this. In other words, as the pulse is shortened and the peak current becomes less, due to the switch off occurring earlier on the coil time constant, the TX voltage increases so that the switch off current is the same as for a long pulse, which equals, say, five times the coil time constant. The coil used for these tests was, in fact, a Minelab SD 11in mono coil.
Looking at the receiver waveform on a scope, with a box of ironstone on the coil, you can see the FRD very clearly, and at early times it is saturating the receiver. Changing the TX pulse width from 150uS to 500uS does not make one bit of difference to the length or amplitude of the ironstone signal. Over the years I have tried similar tests with different circuit arrangements and have never observed any pulse width dependant decay signal on non-conductive iron bearing material. This is at odds with what Candy and others report.
The same tests conducted with non-ferrous metal objects, do show a change in both the decay shape and the time constant, up to the point where the width of the transmitter pulse equals that of the objects decay time. From then on it doesn
A
Anonymous
Guest
Hi Guys,
Thanks very much for your input, it's going to take a bit to grasp and absorb all this.
Our power supply guru here at ATR says I need to use two seperate voltages to take advantage of any history effects to do a mineral metal disc. Full voltage short pulse, lower voltage long pulse, similar I think to what your saying Eric.
Time for some meditation on these ideas you've provided.
Regards
Kev.
Thanks very much for your input, it's going to take a bit to grasp and absorb all this.
Our power supply guru here at ATR says I need to use two seperate voltages to take advantage of any history effects to do a mineral metal disc. Full voltage short pulse, lower voltage long pulse, similar I think to what your saying Eric.
Time for some meditation on these ideas you've provided.
Regards
Kev.
A
Anonymous
Guest
Just tried the Goldquest electronics, but again with a constant current arrangement for the TX supply, so that the pulse amplitude is always the same, whatever the pulse rate or width. This is different to the first tests in that the coil circuit has a high resistance (about 35 ohms total). Again, the box of ironstone on the coil. Only below 50uS Tx width do I see any shortening of the FRD. Between 50 and 100uS TX width there is little change in the FRD.
Eric.
Eric.
A
Anonymous
Guest
For the disc objective, why not just make N voltage measures at those intervals during the signal decay and study the resulting decaying curve by a programmed PIC? Different targets have different curve shapes.
For the ground elimination, save in a vector the measurements taken while the coil was not over any target but well over the current ground.
The actual measures to be used for disc are then the differences between the absolute values captured over a target and the corresponding no-target values.
The signal to a target detection is the sum of all those differences.
Simple and effective.
Willy
For the ground elimination, save in a vector the measurements taken while the coil was not over any target but well over the current ground.
The actual measures to be used for disc are then the differences between the absolute values captured over a target and the corresponding no-target values.
The signal to a target detection is the sum of all those differences.
Simple and effective.
Willy
A
Anonymous
Guest
Willy
There are many implementation methods. I was just trying to stimulate some good discussions about seeing if there is a way to use the known characteristics of PI signals to extract a little more information from an unknown target while being mindful of the constraints imposed by other factors such as maintaining ground balance.
Before one would design or build such a device, there are a few questions that need to be answered to further the design parameters and better understand the constraints.
Is there a significant difference in response between desired targets and non-desired targets, both shallow and deep?
Many forum discussions have described the benefits of having a short sample delay of about 10 uS (from the beginning of transmitter switch-off) or better to get a better response from gold. How many intervals should be sampled and at what spacing? How do you easily show the output to the PI operator? How long would the response of a deep target be available while the coil is in motion to take one or more samples?
Here is what I suspect but need to so some tests or get some PI forum input to confirm.
There seems to be a series of pulse delays starting at 10uS that provides good detection for gold or other similar conductor metals. If this is so, then using a binary delay string series starting at 10uS (10 + 0, 10 + 1, 10 + 2, 10 + 4, 10 + 8, 10 + 16, 10 + 32, 10 + 64) would place the faster pulse delays closer together (10, 11, 12, 14, 18, 26, 42, 74) where they might provide more discrimatitive information. Should these pulse delays go out to several hundred uS? The other issue is, how could you synchronize the sample extraction and analysis for the time when a target sample is available? If you could hold the coil still over the target then your method could certainly provide enough data, however if the ground needs to be neutralized by keeping the coil moving, you may need to make repeated passes over the target to extract, synchronize and analyze the required information.
This as all pretty abstract right now so I need to keep it functional so that I can better understand the opportunities for PI discrimination, constraints and practical ways test some of these ideas to just see if it makes sense to even attempt to build it.
Collective minds are better than one!
bbsailor
There are many implementation methods. I was just trying to stimulate some good discussions about seeing if there is a way to use the known characteristics of PI signals to extract a little more information from an unknown target while being mindful of the constraints imposed by other factors such as maintaining ground balance.
Before one would design or build such a device, there are a few questions that need to be answered to further the design parameters and better understand the constraints.
Is there a significant difference in response between desired targets and non-desired targets, both shallow and deep?
Many forum discussions have described the benefits of having a short sample delay of about 10 uS (from the beginning of transmitter switch-off) or better to get a better response from gold. How many intervals should be sampled and at what spacing? How do you easily show the output to the PI operator? How long would the response of a deep target be available while the coil is in motion to take one or more samples?
Here is what I suspect but need to so some tests or get some PI forum input to confirm.
There seems to be a series of pulse delays starting at 10uS that provides good detection for gold or other similar conductor metals. If this is so, then using a binary delay string series starting at 10uS (10 + 0, 10 + 1, 10 + 2, 10 + 4, 10 + 8, 10 + 16, 10 + 32, 10 + 64) would place the faster pulse delays closer together (10, 11, 12, 14, 18, 26, 42, 74) where they might provide more discrimatitive information. Should these pulse delays go out to several hundred uS? The other issue is, how could you synchronize the sample extraction and analysis for the time when a target sample is available? If you could hold the coil still over the target then your method could certainly provide enough data, however if the ground needs to be neutralized by keeping the coil moving, you may need to make repeated passes over the target to extract, synchronize and analyze the required information.
This as all pretty abstract right now so I need to keep it functional so that I can better understand the opportunities for PI discrimination, constraints and practical ways test some of these ideas to just see if it makes sense to even attempt to build it.
Collective minds are better than one!
bbsailor
A
Anonymous
Guest
Hi bbsailor,
I know that this idea may give more than possible disc information. I've noticed that while adjusting the reject control on the GQ, which advances or retracts frequency and sample times in unison, the sensitivity and depth detection of very fine gold is enhanced. It doesn't always work as there is a lot of aliasing of signals, however when the phenomena occurs it is very marked.
It would require some experimentation to work out what's actually causing the increased respones, enhance this, and stymie the undesireable effects.
<IMG SRC="/metal/html/ml.gif" BORDER=0 width=30 height=15 ALT="m~"> have had a lot of success with multi-frequency detectors, I own both the Sovereign and Explorer, and after owning other types believe that broad band scanning detectors are the future. Since multiple frequency detectors operating in the frequency domain have an edge on single frequency types, then why not with time domain detectors also?
Cheers
Kev.
I know that this idea may give more than possible disc information. I've noticed that while adjusting the reject control on the GQ, which advances or retracts frequency and sample times in unison, the sensitivity and depth detection of very fine gold is enhanced. It doesn't always work as there is a lot of aliasing of signals, however when the phenomena occurs it is very marked.
It would require some experimentation to work out what's actually causing the increased respones, enhance this, and stymie the undesireable effects.
<IMG SRC="/metal/html/ml.gif" BORDER=0 width=30 height=15 ALT="m~"> have had a lot of success with multi-frequency detectors, I own both the Sovereign and Explorer, and after owning other types believe that broad band scanning detectors are the future. Since multiple frequency detectors operating in the frequency domain have an edge on single frequency types, then why not with time domain detectors also?
Cheers
Kev.
A
Anonymous
Guest
Thanks Eric for sharing this info.
Maybe the GQ isn't easily adaptable, as Robby
alludes to above. Still I would have expected that
it would have behaved a little like, what Candy
describes, at the very least in your latest
experiment.
Maybe the GQ can use a softly softly approach rather
than brutal current bashing. If the FRD below 50us
is proportional to the pulse width, non-ferrites
shouldn't be?
If so, then a reciprocal correlation could be
effected much as I proposed above, adjusting the
integration constant such that the FRD is always
(near enough) cancelled, leaving non-ferrites within
the target volume detected. Albeit shallow penetration.
There will always be sought targets escaping through
the cracks, but if this is an option where the
ground response was masking all low conductives
anyway, then any finds are always better than none.
I'm looking forward to doing some bench testing
myself, but first I need to complete my DD.
My wife wont let me burn the midnight oil, if I want
to keep the peace that is <img src="/metal/html/tongue.gif" border=0 width=15 height=15 alt=":b">
Cheers
Kev.
Maybe the GQ isn't easily adaptable, as Robby
alludes to above. Still I would have expected that
it would have behaved a little like, what Candy
describes, at the very least in your latest
experiment.
Maybe the GQ can use a softly softly approach rather
than brutal current bashing. If the FRD below 50us
is proportional to the pulse width, non-ferrites
shouldn't be?
If so, then a reciprocal correlation could be
effected much as I proposed above, adjusting the
integration constant such that the FRD is always
(near enough) cancelled, leaving non-ferrites within
the target volume detected. Albeit shallow penetration.
There will always be sought targets escaping through
the cracks, but if this is an option where the
ground response was masking all low conductives
anyway, then any finds are always better than none.
I'm looking forward to doing some bench testing
myself, but first I need to complete my DD.
My wife wont let me burn the midnight oil, if I want
to keep the peace that is <img src="/metal/html/tongue.gif" border=0 width=15 height=15 alt=":b">
Cheers
Kev.