MAGNETIC GROUND SIGNAL

[Anonymous]

I thought I would repeat the test that I did on the magnetic ground signal and tidy things up a bit to give a more conclusive result. My contention was, that the magnetic decay signal is independent of transmitter pulse width (within certain limits), which is at variance with some other results that have been acquired, both theoretical and experimental.
To ensure that only the pulse width varies and that the field from the coil is steady state for a period of time before the switch off, a special coil arrangement was used. The coil was wound on a ferrite core of 3/4in diameter and 4in long. The core used gave no magnetic relaxation signal of itself and was used so as to give a more concentrated field, so as to get maximum signal from the sample being tested. Coil inductance was 100uH and a series resistance of 10 ohms was used. This gives a coil rise time constant of 10uS and a total rise time to the point where the current is limited by the resistance, of 60uS. The control that varies the pulse width has a range of 150

 

[Anonymous]

Brilliant Eric,
Thanks for sharing your findings. This has produced the opposite results to those I'd gleaned from other sources, just as you pointed out. Could it be that a point is reached amongst the mixture of LR&C etc., where there is a reversal of decay responses, this would explain the conflicting findings. You have been utilising the asymptotic steady state current and finding a decay constancy with simulated ground effects, while Minelab are using the area of maximum current growth and reportedly finding a decay response constancy for dense conductives.
However, it essentially doesn't matter which ones do what, so long as there is a discernable difference that can be utilised, while maximising power efficiency. Will a steady state model be more efficient, than a terminated high current one.
It appears from this test arrangement (flat top TX pulses) that the decay response is proportional to TX pulse width up to the point where the pulse length equals the subject's tau*5, (2p coin tau of 50us ?) after which the response is independent of TX pulse width. Mineralisation generally having a very short time constant (?) quickly reaches the point of volume saturation and with increasing pulse width still gives the same decay response. As one extends the pulse width, each non-ferrous item on the conductivity index successively reaches its volume saturation point and from then on, the decay is constant irrespective of pulse width.
I'm not sure if magnetic earth has so short a tau, after all, these are such long pulses (relatively). At 10 tau the coil current would have reached 99.995% of max. There was no change in the decay between 15 tau and 60 tau, I'd LOVE to know how short the pulse could go before the decay curve changed for the ironstone sample.
If a specific magnetic ground sample has a very low reluctance and, or high relative inductance, then it's tau will be fairly long, and long pulses will be required to saturate the volume.
I'm sorry this is probably all very basic and likely been covered numerous times, but are these correct assumptions? I'd like to get the theory sorted before getting bogged in any empirical stuff.
These findings give me confidence that the GQ can be modified to handle more difficult ground.
Thanks again.
Kev.

 

[Anonymous]

Great work Eric. You may have come up with a method that will work for Australian ground by looking at the variations in metallic targets and not the ground signal. This will require a number of pulse widths to be transmitted so as to cover the wide range of targets. The pulses could step up in width from a minimum to a maximum. The shortest pulses may require using a seperate coil. As the ground signals stay the same using a high R coil they can be subtracted out. Metallic targets will still provide an output after the subtraction process as their widths are different. Let's post any ideas on this method before someone tries to patent it if they have not already done so!!!

 

[Anonymous]

Hi Dave,
The problem may be the short time constants of the smaller nuggets. A 1gm nugget has a measured time constant of 10uS, with a total decay time of around 60uS. To see any difference in decay shape and time, you would need to start with a very short pulse. Achieving this with a flat top would be difficult. Even more so with, say a 0.5gm nugget where the decay time would be halved.
On my Goldquest, the coil switch on TC is 10uS and with the frequency control at max. the TX pulse is 30uS. So even there we are not flat topped with the current at 94% of max. Would need to go to 100uS really.
Eric.

 

[Anonymous]

I can only just see the change in TC for a 3.5gm nugget when going from 30uS to 140uS TX pulse. Any nugget smaller than this shows a constant decay curve. Interestingly, in this range the ironstone is also changing, but not by much. There may be a problem here in that at 30uS the pulse has not reached a true flat top for the reasons in the last post.
Eric.

 

[Anonymous]

Eric, How about this? Zap about 400 Volts into the coil for about 3 or 4uS so as to get enough energy into it and sample just as soon as you can after the pulse ends. Use a 10uS pulse as the second pulse and again sample as soon as you can. I suggest using a fast sample time of about 2us. The samples should be taken with a sample and hold and not a sampling integrator. The 400 Volts can easily be generated with a simple switching boost supply circuit. Shut the boost circuit off during receive sample times so you don't pick up any noise. Now cycle through a series of other pulse widths to cover the targets with a longer TC. What is VERY interesting is that most iron and a lot of junk has a nice long TC. This means that you can look for small nuggets etc around nails and rusty tin cans when using the fastest transmit pulses. The method also cancels the earths magnetic field and most power hum as well as a lot of interference. Don't worry about RF interference as 4uS is one half cycle at 125KHz. Even a large coil will make an extremely poor transmitting antenna for such a low frequency. I suggest using a different tone for the fastest pulses. This would be a real asset when prospecting as a nugget for instance could sound of with a high beep even when there is a tin can next to it. Another idea is to use more than one transmit coil. This will allow us to optimize one of the coils for the short pulses and the other for the long pulses. There are many obvious variations one could make to the above. Your tests already show that this method MUST work for longer TC targets. Let's have a go at the short pulses. ANYONE HAVE ANY MORE IDEAS????? Please post them, Dave. * * *

 

[Anonymous]

I know I'm a pain but could you post or send me pictures of the actual coil waveforms in both pulse lengths used for this experiment.
Rob.

 

[Anonymous]

Eric.
Apparently you didn

 

[Anonymous]

Hi Robby,
The top and middle pictures on the left are the transmitter waveforms. They show the current through the coil by looking at the voltage developed across a 0.1 ohm resistor in series. As the magnetic field developed by the coil exactly follows the current waveform, this is the important one to show. The voltage waveform across the coil is interesting, but it is the current that is relevant to the field produced. At the end of the voltage waveform appears the high voltage back emf, or flyback, pulse which is simply caused by the coil/field trying to maintain the same level of current at the moment the transmitter switch goes open circuit. This spike does not occur in the magnetic field and therefore has no bearing on the eddy currents generated in a target, or in the case in question, affecting magnetic domains in soils and rocks. The voltage spike starts at the instant the coil field commences switch off and is proportional to the switch off rate. As long at the field collapses about ten times faster than the fastest object time constant, then it is as good as if it were removed instantaneously. As to what kicks the target, it is the sudden step in magnetic field, either from zero field to a plus field, or as usual in a PI, from a plus field to zero field. Eddy currents are also generated when the field switches on, but they are not usually read because the current/field grows more slowly, due to the coil inductance, and you would have to read the received signal while the transmitter was on. The transmitter pulse has to be long enough so that any eddy currents generated at switch on have died away, otherwise they will oppose those generated at switch off, and you will end up with less signal. In other words, what is happening in the pulse on time is that any switch on effects are given time to disappear. This is not the case in a ramp, or sawtooth coil waveform. The switch on current/field is still changing at the point of switch off and eddy currents or relaxation effects are still being actively generated at the point of switch off, and will act in opposition. With long and short ramps, as Candy uses, the flyback has more energy in the long pulse simply because the current/field has reached a higher amplitude on the coil TC curve.
Going back to the flyback voltage spike that was debated once before, suppose it were possible to do this experiment. We have a ferrite rod with a rotating shutter in front of one of the poles. A coil is wound on the rod and a continuous dc current passed so as to generate a magnetic field similar to a bar magnet. The rotation of the shutter and the width of the slot is such as to give a succession of 0.5mS pulses of magnetism on to a conductive target. I would say that the resulting eddy currents in the target would be the same as if you pulsed a current through the coil for 0.5mS. The first has no flyback pulse, the second has.
Eric.

 

[Anonymous]

Ok Eric you have proved I'm a good shot as I haven't missed my foot once.
I resized and rotated your pics of the rocks decay, made one a transparency and overlayed them in Photoshop. They match up well.
The ML concept has been compared with what happens in nature close to the earth surface and the effect of the weak, non-saturating earth field on ferrites is well known in geology. The decay period is relative to the time spent in the earth

 

[Anonymous]

Hi Robby,
I think the reason why the ironstone response is acting like a conductive target is purely related to the viscosity effect, and not saturation as I earlier stated. As the pulses are shortened there will come a point where the magnetic decay will become proportional to pulse width, but it may also be indistinguishable from a true metallic response.
If the back emf doesn't "kick" the target then terminating the pulse while it's still in its most rapid growth stage will, and I think this may add to the effect utilised by Minelab. As the pulse is undergoing it's most rapid growth there will also be
a back emf growing in the ferrite, the relaxation process will begin from a position where the ferrite must first overcome it's own back emf. With a steady state method this back emf is not be present, as the current and flux growth have peaked.
It is the rate of any change that will determine the response magnitude.
In the paper Dave pointed to. Looking at figure 4. It can be seen that, especially for the aluminium sphere, the response decay begins below the predicted ground model, then traverses above and then below it again, I think the answer is linked to this, but I'm struggling with the math at this point to work it out.
Cheers
Kev.

 

[Anonymous]

Hi Kev,
Metallic responses differ from ironstone responses in that they decay with a different law. Metallic responses are essentially a true exponential where the log of the voltage plotted against time is linear. An ironstone, or magnetic viscosity, decay is linear if you plot the log of the voltage against the log of time (i.e. a 1/t decay). This, I think, explains the deviation between the non-ferrous targets and the viscous soil line.
Eric.

 

[Anonymous]

Hi Robby,
I wonder if it was the marketing department that compared the magnetic decay with what happens in continental drift. Rock magnetism is there as the result of a different mechanism. As the hot magma rises by convection, it is initially too hot to acquire any magnetism i.e. the iron minerals are above the Curie point. As it spreads horizontally each way, like at the mid-Atlantic ridge, it cools down and is weakly magnetised, freezing in both the strength and direction of the Earth

 

[Anonymous]

Has everyone read the patent by Bosnar? This is a PI design used as a mine detector. Two of Candy's patents are cited as references. Click on the link to view the patent courtesy of Carl Moreland's Geotech site.

 

[Anonymous]

Hi Dave,
Thanks for that. Bosnars patent is unusual in that it was readable and made sense.
Eric.

 

[Anonymous]

Eric,
Candy does briefly mention magnetic saturation as being a problem if it were to occur. He doesn't say it is in any way possible in a Pi though.
You are right in that it is the marketing side that made the claim that saturation can occur in some soils and one would assume that circuit overload would be a more likely interpretation of this. If you were to look at the reasons Candy presents for the initial change in ferrites, to some form of re-alignment and the subsequent decay, then you might well assume that magnetic saturation would be very unlikely in a normal Pi.
Kev,
You seem to be right in that the small change you see in Eric's model is in fact the expected effect when using a rectangular pulse at those pulse lengths and isn't sufficient to get things moving to give noticable seperation. I think Paltoglou mentions something re the ratio becoming non-linear as you use pulse lengths under 1000usecs. You would have to read this and make your own interpretation.
The exitation of ferrites as explained by Candy needs another read as some particles would kick in early as mentioned in the paper Dave posted and these particles would have to share some of their field with their slower neighbors and others that are deaper in the volume. As more of the stationary particles began to turn then the sharing would have a lesser effect over time as the slower particles started to make more of a contribution. The initial in-fighting would probably slow down the process for quite a while.
He did mention this initial sharing of fields within ferrites in one article or patent and how it affects the exitation time but I can't find it.
The explanation as to why Candy adopted the ramp makes sence as the seperation is certainly very noticable.
I suspect my computer has some kind of bug as it keeps freezing, needing a re-boot. Nortons is handing out messages suggesting something is wrong re embedding but can't say what it is. Fdisk and format c: might be the only answer I'm afraid. It'l teach me to keep my image up to date.
Has anyone else got spam that makes Nortons comment "Your computer is halted because an un-authorised script is running" when you just click on the mail without opening anything or the site being advertised?
Rob.

 

[Anonymous]

Thanks Dave, this method uses the effect that I have been thinking about, I'm well behind the times.
I thought one could take advantage of the fact that the magnetic and metallic curves traverse at two points so that by taking only a few A to D samples at 3 points, close to TX shutoff, at a medium point relative to pulse length and at later times, the dy/dx of these points will show whether or not the decay fits the 1/t to 1/t exp2.5
The requirement for only a limited number of samples reduces the time constraints, allowing more accurate processing of the results.
I must read more of these datasheets. Thanks must also go to Carl for providing this great resource.
Cheers
Kev

 

[Anonymous]

Dave, Recently I've been thinking along the same lines re the capacitor discharge into the coil, like a CD ignition system, which I believe produces very fast rise times.This may be useful in a DD coil or two box. Cheers , Allan.

 

[Anonymous]

Allan, It's best to start with a dedicated switching boost supply chip with its own coil rather than chasing your tail to get the search coil to put energy back into the circuit. A very good friend of mine in Bulgaria used a boost supply in his design and switched it off during the receive sample times. I am getting ready to play some games here in a few days as I won a Llamda LLS6120 digitally programmable lab power supply on eBay the other day. It has a range of 0 -120 Volts @ 0 - 1.4 Amps. I am going to have a play around using short high voltage pulses so as to get enough energy into the coil and allow me to sample at very low delays after the very short TX pulse ends. I think you know where I am headed with such a set up, Dave. * * *

 

[Anonymous]

Hi Dave,
Here's the reservoir capacitor I got in for my high power transmitter. 6800uF at 385V working. Batteries go on a trolley and I am belt mounting the reservoir. Have to be careful not to sit on the terminals when bending down digging. If you stand in the middle of the coil you end up on another world. Calling it Pulstargate.
Eric.