Big Coil Problems on a PI

[Reg]

Hi All,

I noticed on a different forum that there has been a few problems with big coils working on certain Minelabs. Sometimes the problem may last for just a while. In the worst circumstances, it could happen the majority of the time. Now, one other thing, the problem doesn't necessarily show up on all machines, meaning if the coil is moved from a 3000 to a 3500, it just might work consistently on the newer machine and not on the older one. It is even possible that one 3000 might work just fine and a different one might display the intermittent problem.

Ok, what is wrong with the older one that would cause a coil to work just some of the time? Here the a real kicker and an answer owners might not like. The answer is, maybe there is absolutely nothing wrong at all. The real answer lies in the difficulty of building a large coil, the detector components, the actual design, and just how PI's work.

Now, I can only guess what is happening on the ML's since little information is available on the actual technical design of them However, based upon some very reliable information passed on to me, I do know there are design changes between different models that "could" result in one model working a little differently than another. What I don't know for sure is just where the design change came about. My best guess is some of the major changes that could effect this condition came about at the 3500 or maybe just before they came out with it, meaning some previous models may incorporate changes not found in earlier models.

Only ML can answer this one for sure. However, some of the information found on an Australian forum would add credence to why the 3500 might be more stable with big coils. This info indicates that testing done recently indicated that the SD's go deeper than the GP's. It would also explain why the 3500 seems to be quieter and do better on the smaller gold.

Furthermore, it could explain why certain models do seem to have a greater problem. As ML modified their detectors, I am sure they used different techniques to enhance the newer model. Pushing the limits is one of them. So, it is quite possible that this condition of large coils not working all the time is mainly found or found more often on certain models.

With that said, I will continue with the explanation but base it upon the GS design which I do know more about. Now, before anyone jumps up and says there are dramatic design differences, between ML's and other PI's, I will caution them that fundamentally, they all work the same. This is clearly evident when looking at various waveforms and the actual circuitry. The big differences lie in current and voltage levels and techniques used in processing signals. There is no "magic" involved.

Ok, the rest of this info will help explain to GS 5 or even GQ owners as to why they may have to turn the delay back, which is the solution that will allow the coil to work. Fortunately, the same info will explain why PI's with no delay adjustment will cease to work.

For starters, coil building is an art and such art becomes more difficult as the coil gets larger. Yeah, one can build a coil quite easily, but build one that works right is a different story. Now, one of the big reasons lies in the fact that winding large coils causes windings to be closer together for longer distances. Fewer windings are involved and those windings are much closer together meaning the first and last winding may be quite close on a large coil and farther apart on a smaller one. This distance thing becomes important because two windings close together form a "capacitor" effect. The more capacitance involved, the more difficult it is to get a coil to work right. What ultimately happens is the larger coil takes longer to "settle down" and work right. Changing insulation types and how it affects a coil become much more obvious on bigger coils but doesn't completely solve the problem. The bottom line is very large coils are most likely right on the ragged edge of working, simply because of their size.

On a PI, the coil has to be "stable" and the signal has to be decayed or leveled out at the time of the sample or bad things happen. If the signal hasn't leveled out, then the sample can be taken at a time that too much signal is being processed, resulting in the saturation of the processing circuitry. If this happens and the signal has bumped into its limits than any further signal caused by a "target" won't cause any further change. So, nothing is heard. Just what is heard at this limit depends upon the opamp used, the gain involved, etc.

On Eric Foster's machines, this limit isn't a hard limit, meaning that once a coil has reached the condition that the sample is taken too soon, the signal becomes extremely muted. If it is way off, it become almost no signal at all and one has to listen very carefully. Move the delay back a little and the signal becomes more obvious but still muted. It will sound like one is listening long distance. Carefully turn the delay up a little more and suddenly, boing and the coil is extremely sensitive. At this point, the sample is taken when the signal has leveled off and the amplifier affected is no longer saturated. Now, some serious amplification can occur and a target signal is very loud. Now, one could change the particular opamps involved and this "soft muted" signal could suddenly become a sudden shut off of the signal when saturation was reached. This "soft landing" of the signal muting helps the operator know when it happens.

Ok, this partly answers what can be happening, but other factors are also involved. The heavy current in the coil causes heat and this heat causes things to change. Some of the big change is in the FET used. As the FET heats up, its characteristics change and although such changes are small, they can alter just when the problems occur. The FET is the switch that turns the current on and off in the coil.

So, a coil just might work fine for a while and then suddenly, it doesn't work or doesn't work well and everything is normal. Other components can be contributing also and the combination could be such that the problem only shows up at certain temperatures. Even moving the control box from the sun to the shade may make a difference if everything is right at the very edge. If things are really at the critical level, even slight differences in FET's can cause the problem to rear its ugly head. The bad thing is the FET involved is really still within its proper limits, meaning there is nothing wrong with it, but because of some minor variations, one certain FET could work and another from a different batch might not work the same.

Now, on one of Eri's machine, one simply turns the delay back a little more and suddenly the coil works fine again. This is something that can't be done on the ML's, so the ML's suddenly don't work.

I hope this helps explain what probably is happening and why tech's at ML can't find anything wrong with the detector and coil manufacturers can't find anything wrong with their product either, since both are well within spec. Once again, both products could be just fine but the coil may not work well all the time.

Reg

[Eric Foster]

Hi Reg and All,

I agree with Reg’s explanation, although it is a puzzle as to why Minelab PI’s are so critical on coil parameters. One of the basic features of the PI principle is that, because the coils are untuned and resistively damped, a PI detector should be able to cope with a fairly wide variation in these parameters. Some years ago I bought a SD2200 for test purposes, and actually ended up manufacturing aftermarket coils for it. These ranged from 3in, 8in and 11in monos, to 8 x 24in and 10 x 30in rectangular coils for skid or trolley mounting. The reason for making the 8in and 11in monos, was that some customers wanted fully waterproof coils for searching beaches down to, and into the water. My open centre coils were fully potted, and being an open ring, had little water resistance when submerged, and were not bouyant. The larger rectangular coils were as a result of considerable correspondence with Chris Hake, who trolley mounted them and trundled across the salt lakes in WA looking for nuggets.

The thing I found was, that even small departures from 300uH inductance and 0.4 ohms resistance resulted in a coil that was either noisy or simply didn’t work. Coil and coil/shield capacitance was also a factor, but did not seem quite so critical as the first two. The way I managed to make a good coil was as follows. I looked at the response of a Minelab 11in mono when connected to one of my Deepstar PI detectors. The transmitter was adjusted to pulse a similar amplitude of current into the coil and the initial damping resistor was the same value as the SD2200 appeared to have. The scope was connected to the first receiver amplifier output so that the low level ringing and/or overshoot and recovery time could be observed. Any size coil that I made had its response tailored by the addition or subtraction of resistive damping, and also by some novel inductive damping to cancel out any capacitive overshoot. Once I got the responses matched, the coil always worked on the SD.

As Reg says, as a PI design strives for ever shorter sample delays, the coils will become more critical, particularly in terms of inductance and capacitance. Damping needs to be more finely adjusted too, so that almost all vestiges of ringing and overshoot have vanished. In my SD2200 tests, it did not appear that signal sampling was taking place at a “short†delay. The small object sensitivity and range, fairly well matched that of the Deepstar, which samples at 15uS. It did appear to me at the time, as though something additional had been built in to the SD circuitry, so that the detector would not perform satisfactorily, unless you used a ML coil. However, that may not be the case, and it might be just a side effect of the circuit topology used.

The addition of a variable delay control overcomes a lot of problems. It may only mean shifting back a couple of uS to get the detector working again, as Reg explains. Usually, big coils are used for hunting for larger and deeper nuggets anyway, and these nuggets will have reasonably long decay curves, so moving back a few microseconds is not going to make much difference to the received signal.

Regarding the Goldscan 5, it was quite a challenge to make a detector that would operate not only on its own coils, but also those of ML, Coiltek and Nuggetfinder. This is largely because of the big difference in winding resistance, and differences in the way the shielding is done, which influences the circuit capacitance. However, it has been achieved, and from tests that have already been done, many of these alternative coils will still work properly even at the shortest delay on the GS5. While in Australia recently, a prospector produced a coil about 30in diameter and the GS5 accepted that OK.

Eric.

[Capricornstreasure]

Thank you Eric, Req,

Parts of your explaination dovetail with my thoughts on why larger, or even

poorly built, coils could cause the problems I've been reading on. No

doubt your variable time delay is a good fix to assist in matching coils

to a given machien. ML not having that adjustment it should be subject

to funky coil problems.

The Fet problem <suspect> is an output from, just a guess,

the machien and has more to do with loading <or overloading> but thats

just my opinion and would cause on off on again problems. I dont

know that this type of problem would be intermitant~ FETS tend to

work or not. Please feel free to pipe in....

As ML machiens dont "know" what coil they have on them a bit could be

said about programing~ the machien does a form of self calibration in

the GP series. I suspect the 3500 has some form of self adjustment due to the

increased automation to compensate for the coil used.

As to older SD's I think they are just more "open" as

far as tolerances go, and that also affects sensitivity, and probly why

they dont seem to suffer as much with coils~ but thats based on lack of problems

from what amounts to third party knowledge. Only ML knows for sure

if they even track such things for product development. And it goes for

'punching deeper'.... that is a sensitivity issue as much as it is a power one.

Agian, opinion. I would love to see your notes.

It might be helpfull in your explaination on overshoot and ringing if

you could get up a couple of O'scope pictures and a little more basic

explaination of why it occurs. An idea.

Thanks again, talking soon,

pz

[Reg]

Hi PZ,

FET's normally do just work or not work, but they can change characteristics a little when heated. This heating change is what I am referring to. I first noticed it when trying to build a coil that operated well bellow 10 usec. I would get a coil to work and then, the coil would quit working for no logical reason. Turn the detector off for a little while and back on and the coil would work again.

At first I wasn't sure just what was happening. Eric briefed me in on the details as to what was happening and this made it much easier to actually look for and see just what was happening. Because it really doesn't take much signal change to cause the problem, one really has to know what to look for. On my PI's the gain of the amp is 500 so it doesn't take much change at all.

So to be honest, I have actually experienced a temporary coil "failure" because of the temperature drift.

Now, there a couple of techniques that would reduce the problem including reducing either the gain of the preamp or reducing the current in the coil a little. Both would help or could help signficantly.

An automatic compensation circuit would have to update on a regular basis to be of value. Could it be done? Yep, that isn't the hard part.

Overshoot and ringing occur as the result of a coil not being damped properly. A poorly damped coil will or could oscillate for quite a while, while one that is reasonably close may only overshoot once or twice.

The key to his oscillation and whether it hurts or not is how soon it quits. If the coil is stable when the sample is taken, then there is little or no harm. When I first tried a ML compatible coil on one of Eric's machines, I noticed a slight oscillation but that occurred well before the sample time, so the coil worked fine. I will see if I can come up with a couple of pics to display what I am talking about. Unfortunately, I have changed computers and many of the pics haven't been transferred.

In the meantime, I have added a pic that shows a decent coil decay signal. You will see there are two signals, a large spike type signal and a smaller one more in the center. The smaller one is the signal as seen at the output of the preamp. You will see that this signal drops down and flat lines for a brief time then rises quickly to level off again. The vertical dashed line is a scope setting that indicates the time from when the pulse is shut off to when the sample levels off and can be sampled. In this case, the time is 4.6 usec. The signal really shouldn't be sampled until it comes back up and levels off again.

I have also added a pick of a ML coil that was tried on the same detector. As you can see, the signals vary signficantly. The ML coil displays some oscillation and instability. However, it does level off soon enough that it can be used ok. Now, on the ML coil, the signal doesn't level off until about 10 usec. or so. In fact, the dashed line indicates the time of 7.97 usec or so and the signal is still rising. As such, one should wait until that signal levels off which is just before 10 usec. Actually, it levels off closer to 9 usec. The center red vertical line is 10 usec. So a person could probably sample at 9 usec and be ok.

Hope this pics help.

Reg

[Reg]

Hi PZ,

Here is a little detector trivia. When was a dual voltage pulse used as a method of ground balance and when was a reverse pulse also incorporated in a detector?

If you said ML you are wrong. Would you believe the year was 1977 and patented in 1978 by George Payne? Yep, the pulses were current pulses, which are a little different than a PI but the general technique is the same, which is to pulse a strong current into the coil and sample at the right time, compare the two and bingo, you have a ground canceling pulse detector.

Here is the link to the patent info:

http://www.thunting.com/geotech/pages/metd...s/US4110679.pdf

Now, this really isn't a true PI, which works a little differently, especially on what happens when the current is turned off. In this case, the first PI built to ground balance was built by Eric Foster, some time in the early 1980's.

If you use a detector today for nugget hunting, chances are much of the technology incorporated is the result of the work done by one or both of these two great design engineers.

Eric Foster is known world wide for his advancements of the PI, while George Payne is better known for his introduction and advancements of the VLF in the consumer market.

Reg

[Capricornstreasure]

Thank you Req (Thumbs Up)

I'll ruminate on that a while for sure The pics are pretty good

and informative for the samples taken. I'd like to get ahold of a

malfunctioning machien....

Names Pat.

Talking again soon,

pz

[Reg]

Hi Pat,

Glad you like the pic's. Now, here is something else to chew on? Any guesses as to what might be causing the slow rise (or droop) of the signal on the ML coil? This is the one with the oscillation and the numbers 7.97 us on it.

Also, would this act like a defective coil if it were to be tried on a detector that normally operates at a delay of 7 usec?

Now, I will say this, when you look at the ML coil pic, the long "droop" in the signal rise looks somewhat like what happens when a FET changes characteristics with heat. In other words, the signal would start out looking more like the first pic where the signal rises quickly and levels off early and later, as the heat affects the FET, the signal begins to droop something like in the second pic where the final leveling time is later. In this case, however, the droop is not caused by heat but by something else.

One more little item of interest, this droop in the second pic is also somewhat of a simulation of an invisible nugget. In the case of a "invisible nugget", those like the ones John Blennert has found, would cause the signal in the first pic to droop much like what is happening in the second pic. (One should disregard the earlier oscillation on the second pic while thinking about this one).

So, if the signal in the first pic were to droop like the signal in the second pic because of a target, such droop that ends or levels off well before 10 usec would be indicative of the signal from an invisible nugget or something else that has a very short time constant. As such, if a signal sample were to be taken at 10 usec, nothing would be detected since the signal (the droop) is gone by then. Now, if the sample were to be taken at the 7.97 time, the droop is still present and this would result in a nice strong signal.

Reg

[Capricornstreasure]

Some questions for your one, and a confession of some ignorance on my part.

Also comparitive analysis of the siganals I see....

On both pics there are two traces. Please allow me the indulgence of labeling

them A and B. A is the one with the Delta T of ~4 micro seconds and B is the

~7 micro second displayed.

Both A and B show a large pulse of approximatly 2.5 volts~ what is this pulse?

My guess is its the output pulse <are you using a 10x probe? Did you calibrate

before measuring?>.

The second question is whether the coils are the same size and were they placed

in the same location/orientation when the measurement was taken? <well two

questions in one there > In short what was your set up.

Display A does not appear to be complete for the lower ampltude signal. Compaired

to display B the first part of the waveform looks truncated otherwise the signals

are identical except in the last part of the rise. This can slew your delta t measure.

The rise time to the level state for B is about 2.8us but appears to be interupted

about 1us into its upturn. This suggests to me some minor feed back but

does not look like its caused by the coils ringing. Thats why I ask about the

coils sizes and orientations.... it looks induced.

I say that because the output pulse <larger waveform> shows a ring at the base

of its pulse and that seems to be a response to something other than the coil.

Ringing does not normally follow a pattern of this type on the front end in my

experience with one shot oscillators.

A display shows a sharp clean rise of about .8us to steady state and even the output pulse <the

larger waveform> shows sharp features. Nice but I do want to know if the

test set up is identical for both coils.

Before I go any further can you orient me on these items of interest?

My ignorance comes from not having done these measurements myself so

I dont know what a good targets waveform would do to the steady state ie

would it show as a dip and rise or an oscillation? I can try it with my scope

but its a small digital modle and not fit for pictures~ field stuff.

As to one of your questions, would a time delayed machien of 7us be affected by

the signal time of greater than 7us.... Certianly. Depending on the sample

window if the overshoot is excessive enough and the reciever is sensitive to the

signal at that level it would probly result in a warble or constant tone as

seen by the reciever. Its looking for a variance on the steady state that a

target will make when its signal cuts the coil.

Thoughts or am I on a bad track?.

Talking soon,

pz

[Reg]

Hi Pat,

Please excuse my poor marking of the traces. I haven't updated to a decent drawing program since changing computers, so trying to follow or draw a traces is crude at best.

Now, I have added three more pics. On these pics I have attempted to indicate key issues.

On the first pic, the yellow trace is the pulse flyback signal. This is the signal that occurs when the pulse is turned off quickly. Yes, I was using a 10 to 1 probe for this signal.

The second trace on the first pic, the one in red is the output of the preamp. This signal is a very decent response curve. My add on drawing doesn't do it justice since in doesn't follow the actual trace. Now, the original preamp trace indicated the signal levels off at about the light colored vertical line, while my hand drawn trace levels off closer to the middle vertical red line.

The slight difference between the orignal trace and my red trace where the red trace droops a little before leveling off again would or could indicate a very weak gold signal from a small nugget and could indicate the type of signal coming from what is commonly called "invisible" gold.

Now, on the second trace I have inserted, I have focused on the output of the preamp signal only. On this pic, I have exaggerated what some target responses might do to the signal.

I have also highlighted the instability of the coil signal. The yellow down up down initial response is oscillation in the coil that does dampen out quickly. The receive signal does start to rise normally after the oscillation, but then ramps slowly (indicated by the lighter blue line. This slow ramp could indicated a target that has a very short time constant. In this case, my guess is the signal is responding to what is being used for shielding or maybe something (some form of a target) added to control the rise of the signal. The dark blue line is where the signal has finally leveled off. Normally, sampling should occur along the blue line.

Ok, if the sample were to be taken before the vertical dashed line, such sample would display the signal from whatever is causing the slow ramp. If this signal were strong enough and constant, then the result would be the coil would no longer function.

Ok, I have added a third pic which is an expanded version of the first. In this pic, I have simply made a few crude drawings to the pic to indicate what happens to the signal when a couple of different targets were to enter the coil's field. One change is the addition of a grey signal that ramps slower than the original red line but faster than the white trace.

Remember, the traces I have added are very simplified versions of what happens. The signals would actually rise exponentially and not like a simple straight ramp type signal I have drawn. My drawing is extremely simplified to show how the signals droop and the voltage levels change at the potential time of sampling.

Now, the grey trace would indicate a small gold signal much like that from invisible gold sitting on the coil, while slow rising white trace is much more like a larger gold response. In this case, my gray trace has leveled off at 8 usec (2 usec per vertical line), so any sample taken after the 8 usec would not cause any response. The white trace levels off at about 20 usec (end of the screen) so any sample after that time would also result in no signal.

Now, I have added two dashed lines, a white one and a blue one. The white dashed line occurs at about 7 usec while the blue dash line happens at about 16 usec. Ok, now if the sample were to be taken at the blue dashed line, the signal indicated by the grey trace would be gone before the sample occurred so no signal would be heard. The white trace indicates a signal that lasts long enough that a sample at the time of the blue line would cause a response.

I hope this makes sense. Once again, sorry for the lousy hand drawings.

Reg

[Capricornstreasure]

Req,

No worries on the drawings, its my turn to hash them up a bit now.... only

it will take me a bit longer. I'm still thumbs with pics~ and video of late blah

Now that I am oriented on what I'm seeing I'll get back to ya. I still wish

to know if you are using identical coils~ there is a time variance I'm not

understanding that does not appear to be induced by the O'scope <to big>.

I think your using two sizes of coil. Not that it matters much for the direction

of our discussion. Once I lined up the traces an explaintion presented itself

for one of my questions to ya~ but I'll address that when I get the pics up.

Looking at the file title of the pictures your using your Hammer Head for the

demonstration?

For simplicity I will be sticking to the sample window starting at a fixed 7us~ I'm

basing my part of the discussion on a PI which has no delay adjustment.

However your explaination in picture three does point out how that variable

would be benificial~ it would aid in operator calibration <minimum sensitivity> for

various coil sizes.

From a troubleshooting perspective:

I disagree that the coil would not work if the decay signal is still present~

its not the coils fault but the machiens start time or sensitivity level.

The coil will work even if its still ringing~ it will just take a bigger signal than

the coil is putting out to override the false signal~ the result is garbage but

it does not stop working. My opinion anyhow. It would be percieved by

myself as a malfunction in the machien and most likely result in a blank signal

or warble in the headset. (Why not?) An example is found in the use of a detector near EMI sources~ lots of

operators in noisy areas by ignore the pops and whistles and concentrate

on the big repeatable signals. By that statement I'm relating the pops and whistles

to the interference caused by a ringing coil.

To big a Flyback, as you call it, could fry the Preamp, I guess, as well

and reveal weak components that is also a problem in the machien if the coil is in spec.

Insufficient dampining is another problem but I dont know that it is

built into the coil or machien~ my money is on the machien as its the most

controllable of the two variables ie: Coil or Detector.

As a side note, the delay variance can be mathmaticaly accounted

for if the machien is able to sence the size of the coil used. The machien would adjust for a

theoretical optimum for a given size of coil percieved. No reason why a machien

cant do it automaticaly~ I suspect the 3500 has done something like this.

I'm stopping here as I'm drifting away from our main line of discussion~

the O'scope signals and what they represent. Like I said I'll get some

pics up directly to point out some things I see.

Talking soon,

pz

[Reg]

Hi Pat and all interested people,

Here is a very simple test.

Pat, if you have a ML detector, please try it. Also, anyone else who has a ML please try it also and let me know what happens, especially if the result is different from what Pat finds to be true.

Now, if a ML can adjust for a coil change, then this test will clearly prove it.

First turn on the machine with the coil away from all metal.

Now, pass a nickel or similar coin across the surface and note what happens. Obviously, the ML should be able to detect the coin with a nice strong signal.

Next, tape the nickel or similar coin to the center of the coil so it can't move and note what happens while keeping the machine on.

With the nickel taped to the center, pass a second nickel across the coil and let me know what you hear now.

If the machine no longer works with the nickel taped to the coil, then turn the detector off, wait a second, and turn it back on again. Once again, with the nickel still taped to the coil, pass a second nickel or other object across the coil and see if the coil works now. If the detector can compensate for the coil, then it should work this second time even though it didn't work when the coin was first taped on.

Let me know what happens.

Reg

[Capricornstreasure]

Req,

I run a 3000 that alone invalidates the test. Neat idea but here is what I understand of the process. See if you

agree. Some explainations are for other readers not familure with electronics

so bear with me please.

I'll do it however I expect it will only be balanced out .... the taped

nickle.

By automatic changing of the time delay in a 3500 I was refering to

coil size and optimal time to resume sampling and the possibility to catch weaker signals. Larger coils should

take more time to calm down and thus affect this delay and consequentialy sensitivity to smaller targets.

Coil line length can be measured easily and

adjusted for by compairing velocities of fixed signals at turn on to datum

stored in a program. TDR's use simular principals. <Time domain reflectormeters>

For the 3000 and below the sample time is probly fixed to the optimum for

an 11 DD stock coil line length. That any other coil works is because its made

to be quiet before this fixed time.(a presumption on my part)

I am omitting the facts of increased signal line (how big the coil is physicaly in length) distance to the reciever and

powerloss effects due to this increased line lenght as well as the signal to reciever distance ratio,

ie the signal will be weaker due to the distance traveled from time of emmision of the target to the time of signal interception

by the coil as it passes over. <this is an inverse problem and squared to boot>.

The pulse that gets the target ringing will be constant as long as the target

is under the coils radiation pattern but the time it takes for the coil to physically sweep over

the target close enough for the target to induce sufficient EMF on the coil for a response varies

with distance and whether its inside the pattern.

Sensitivity would be difined by when the sample

began, the target being inside the pattern of radiation and cutting the coil with its emissions

and the sensitivity of the reciever to the induced EMF whenever that induction occurs and at what strength.

(The inducted signal will still have to overcome the line loss)

This affects only very small targets in relation to coil size. Its a matter of

resolution to coil size in the problem I propose.

By taping a nickle to the coil the math affected by the machien should be

a ground signal sample minus ground signal <with or without nickle> and

all other signals are good. Because the nickle is not moving relative to the

coil it would only be percieved as part of the ground and the detector will raise its level

of response accordingly, just like hot ground~ you only lose depth.

The response from the second nickle <the one on the ground> should be undiminished.

At depth however it would make a big difference.

This is my hypothesis. I'll let you know the results when I conduct the test

tomarrow.

For what I proposed on the 3500 you have to change the coil size.

We'd really need two machiens, a 3000 and 3500 should be compairable,

and 3 coils, mono's would be best, at 11, 16 and 20 inches with 4 targets, two of

equal size of minimum detection range for the 20 inch coil and two for

the minimum response for the 11 inch.

Strength would be measured on a fixed amplifier in open air by response

strength ie what do you hear. I'm not discounting better amplifiers in the

3500 over a 3000 so maybe two 3500 would be better for the analysis

of whether the time delay is affected or not. The same coils must be

used on all macheins tested however to yeild verifiable results.

Air tests might help with that set up on a dual trace O'scope and pick offs

at the points you have defined Req, the coil and Preamp and I'd love one

at the output stage as well or a sound level meter would work on a fixed

output from the machien.

Thoughts.... anyone~ your invited Req and I cant have all the fun.

This does stray rather far afield from the points of intrest. Pics are still in work.

pz

post note : That PI machiens do not respond <typicaly> to hot ground is a

result of electrical canceling of signals that VLF machiens can not

accomplish easily without a drastic reduction in sensitivity. It is a fundimental

falicy, in my opinion, to say that PI's are immune to the affects of hot ground

when taken in relation to Depth sensitivity. I say this to keep it clear about how small

signals at depth can be masked by hot ground and this spiel of mine is in relation

to mild ground and shallow targets. I do not even try to take in all the variables

in my thoughts on the subject of PI coil depth verse sensitivity verses target size

or orientation. To much information.

[Reg]

Hi Pat,

Maybe we can get Rob to try this test on his 3500 and see what happens on that version also. Personally, I am really interested in the results.

Now, one doesn't have to actually move the coil to get the same results. One can simply place the coil in an open area, tape the nickel to it and then use a second nickel for the further testing and simply pass the second nickel past the coil that has the first nickel taped to the center, rather than try to move the coil with the coin mounted on the coil. The results should be the same.

Reg

[nuggethunting]

Hello Reg,

I just tried it with my GP3500 and the both coins turned to gold. Just kidding, I wil give it a whirl tomorrow and see what happens if no ones beats me to it.

I just wanted to let you guys know this is a very interesting thread you guys have going. Great to have Reg and Pat both contributing great information.

Talk with you soon,

Rob Allison

[Capricornstreasure]

Got the Midas seach coil do ya Rob?

Not in any race for the results. I'm working with theory Req with equipment.

I dont have access to a lab as I would like and I'm packing my life

up for my walkabout beinging this month so I'm real limited on what

I can do for controled experiments.

Should be interesting results but the time delay for the 3500 hypothesis

I proposed may not be readily apparent without electronic aids.

It would be nice if someone from ML would chirp up but I dont hold

any hope in that.

Moving thru the morning...

pz

[Reg]

Hi Rob,

The coin turned to gold huh? With your natural ability to find nuggets I can almost believe it.

If you get a chance, test both a mono and a DD, preferebly a 11" or smaller on both types of coils.

No need to do much more than attach the coil and make a quick test.

Thanks Rob,

Reg

[Capricornstreasure]

Ok Req,

Ran the test and got some interesting results.

I ran it using two nickles <had three but I'll get to that> and two 22cal

bullets from the target collection. I used masking tape and brought some

aluminium foil tape in case I needed to up the masking. It turned out not

to be necessary.

I ran it using my biggest DD in N for the XP, GP was coil DD and sensitivity

both tested in Normal and Sensitive. Tuning was performed twice and

the coil was only balanced once for the first run and balanced each time

a bit was taped on after the first run.

The 22 and nickle sounded off loud from over 6 inches up without the addition

of metal onto the coil. After the first nickle went on there was a noticable

increase in threshold and it became smooth~ no warbles or pops. The

22 cal bullet immediatly droped in intensity only sounding loud at half the distance,

3 inches, and the nickle target lost about 1 inche. I taped a second 22 cal

bullet to the coil and the results did not change much. This changed when

I added the second nickle..... I lost the 22 cal target and had to use sensitive mode while

SCRAPING the target bullet to get a response. The nickle target was lost at about 3 inches more

and the response was weak above that. In sensitive as noted the bullet

all but disappeared and the nickle only gained a few inches of hieght.

I removed the two nickles and left the 22 cal bullet attached and resumed

playing with the other bullet target.... It came right back but was slightly diminished.

During the experiment the threshold increased in intensity and smoothed out

even a bit with only the 22 cal bullet attached.

Even after retuning and balancing I could not get the 22 cal target to come

back in normal mode for the GP with two nickles attached to the coil.

When the nickles and bullet were removed

Man did that thing act funny..... warbled and spit until I retuned and re balanced.

Pretty cool.

What do I feel this test proved to me..... that its a Bad Idea to stick a nickle

on your coil. But really it showed me that the machien was not able to

cope with the added metal ringing near the coil and it droped its sensitivity

in a most profound way. I suspect if an O'scope where attached it would

prove that the increased decay time created by the nickle simulates a bad coil and would indeed

cause a steady threshold with no variance and hence make one think they

had a dead coil. Kind of like operating a mono in cancle. I also feel it proves

that the 3000 does not adjust its time delay in any good way. It does result

in loss of depth as hypothesized due to lack of sensitivity by the reciever being

over driven by the constant noise of the nickle(s).

What you get Rob?

Thoughts Req.....

[Reg]

Hi Pat,

Thanks for the info. Now, for starters, I messed up when I mentioned placing a nickel on a mono coil. The distance between the nickel and the windings is such that the nickel is most likely not great enough of a conductor to cause the saturation I expected. Most likely, it would take maybe 3 or more to get the same results as one nickel carefully placed on a DD coil.

Now, as you found out, a nickel does work quite well on the DD coil, meaning if the nickel is placed within the main detection zone, some serious changes result in how the coil works. Basically, the nickel does change the decay curve of the coil sufficiently that some later amplifier is partially saturated, which is what caused the "smoothing" of the signals. My guess is the depth capability would be the same or very close, but the signal is seriously muted, initially giving the indication of a sensitivity reduction. This is the result I expected.

Actually, you mentioned aluminum tape. This can be used also and it would be less likely to move.

Now, as you found out, this smoothing of the signal does make it much easier to hear the signals. Hey, this is a unique way to minimize noise on really bad days. Unfortunately, all signals are obviously weaker, audio wise. What should be done is to see just how much of a real depth loss occurs. Also, if the nickel is moved a little sideways to the point that the signal is still quite smooth but noticeably louder than previous positions, then further tests can be done. At this point, one could once again, test for maximum detection depth to see just what does occur. I think you will find that not much true depth loss occurs. There will be a noticeably audio change initially giving the impression of a depth loss, but in reality, the loss is not what one intially thinks.

Now on a different note, with careful positioning of a test object such as a nickel on a coil, one could get a better idea of just how the ML's are designed by simply comparing what happens to the signal from that coil and comparing that change to what happens on a different PI where all factors are known. Itis a rather unique way to "look inside" without really doing so.

What will be interesting is if the 3500 displays similar results. Once again, the key is using a DD coil and positioning the nickel on the center of the coil within the main detection zone of the DD coil. This will tell if the 3500 has a built in adjustment for coil variations. If the nickel is placed and it causes a signficant reduction in sensitivity and then the 3500 is turned off and back on, any signficant difference would indicate they do compensate for the coil.

Now, I am not sure just what will happen on the ML's but if the nickel is moved such that it is positioned either left or right of center so it is more over the center of one of the DD windings, then the results could change dramatically.

One reason for this is the signal from the nickel is opposite when over one of the windings, meaning if the nickel generates a strong postive response when over the center overlap zone, then when over either winding should result in a weaker negative response.

Any target over either winding and not over the overlap area will produce an opposite signal. This opposite signal is what causes a DD coil to seriously reduce the ground response on a straight PI. Actually, it does it on any PI, but most noticeable on a PI with no ground balance or ground balance turned on.

Ok, just for fun, one might play with this little trick using different objects mounted on a DD coil at diferent locations to see what happens. Given the right combination, one can "enhance" a coil such that it will appear more sensitive. At least, it can be done on some PI's. My guess is it can be done on the ML's also. My guess about ML's also being affected is based upon some info I was given by a guy who had done quite a bit of experimenting with ML's and various coils.

One can have fun with this experiment if they simply change the test object. One rather extreme case would be to mount a piece of ferrite or even a ceramic magnet and see if they can produce a 'hotspot' on the coil.

So, this little test opens a lot of different avenues for those willing to experiment.

Reg

[Reg]

Hi Pat,

My discussion on depth loss as the result of an object mounted on the coil sort of contradicts your depth loss indications. What I have found with my PI's is basically what I mentioned, the audio is seriously affected leaving one with the perception of a significant depth loss but the true depth loss is not that great, providing one listens very carefully and/or uses some form of an enhancer to enhance the muted responses. I guess I should have mentioned that in my previous post.

Now, if you are really experiencing the losses you mentioned on the ML, then something else is happening within the ML that doesn't happen in other PI's. Obviously, the signal processing is different. Most likely the muting or smoothing of the signals also mutes harmonics so filtering would be affected. This would sort of reinforce some thoughts I have on how ML processes their signals.

Reg

[Capricornstreasure]

Req,

My art work is just as bad as yours~ we using the same editor?

Take a look let me know what you think.

A bit about me, I was trained by the US Navy as an Aviation Electronics Technition

and have further training as a Metrologist, Calibrations Technition. 'Trons

aren't a minor hobby and I know how to set up experiments and test/make devices.

I currently work in my field as an Avionics Technition.... I fix and build things

and have a pretty good Idea of what I'm looking at. I bring this up so you

understand where I'm coming from.

There are two issues in your experiment and I want you to understand from my perspective.

One,

True Depth is defined by myself as the extent that the output signal

is able to penetrate the soil. This has little bearing on Detection Depth which is

related to reciever sensitivity and target signal strength. The target signal must

overcome the ground signal in order to be interpreted by the reciever. It is

not a matter of time delay but signal strength and the recievers ability to

cope with all signals presented to it. The analogie is that you cant hear a person

wisper in a room full of yelling.

Time delay only plays its function in that the ground signal and coil decay pulse

must deminish sufficiently to allow the reciever to correctly interpret the remaining

signals apparent to it. DD coils aid in this by canceling the ground signal at the

coil because the ground decay signal is presented to both coils at the same time

after the output pulse and thier phases are reversed to each other. This provides

a natural canceling effect and leaves only the coils decay time to deal with.

Remaining signals only present themselves on one side of the coil

and in the overlap zone are of suffieciently different time (induced by the

velocity of the target sweeping under the coil) and power as not to

cancel themselves out.

Time delay can not cope with induced signal decay

caused by large metal targets on or near the coil. Increased delay would only

interupt the sample time of the machein to the degree that it would make it

useless to very small targets with small signals before the next cycle.

Addition of metal to the coil results in the presence of a large decaying signal

whose time extends past the time delay and swamps the reciever and by virtue

of the induced decay signal.

This increase in decay signal strength washes out any signal that is not equal to or

greater than the decaying signal. Increased threshold volume is the result~

because the target is not moving it can not induce the power fluctuations that are

a result of the target moving thru the coils detection zone and thus does not

give the pitch changes that a moving target would provide <no velocity>.

The decaying signal will raise the input power available to the reciever from the

coil. That power level is directly related to the recievers sensitivity to smaller

targets levels of power as related by the analogie provided. The result is

loss of Detection Depth and during the experiment was demonstrated quite

adequatly for my purposes.

A conclusion to my first point :

Introduction of metal to a search coil can reduce interference by other sources

as a result of decreased sensitivity induced to the reciever. But this also

results in loss of Detection Depth. True Depth is not affected, true, but it would

take a damn big signal to punch thru. It is easier to adjust the threshold

of the machien~ you still retain a greater degree of sensitivity for the loss

in Detection Depth than you do by applying metal to the coil.

I may experiment with this further as a field aid.

My second point is in relation to my speculation of the GP3500 to adjust

its time delay to compensate for larger coils.

This experiment has no bearing on this point because it is realted to coil

size and dampening time. In a fixed system, time delay is fixed, some

sensitivity must be sacraficed to allow multiple coil sizes to work on the

same machien in order to account for increased dampining times.

As an extra "Tweek" if the machien knew what size of

signal line it was dealing with I could theoreticaly adjust its time delay

to give the best possible results for ANY given coil. Its just a very little

amount of variance but it could be suffiencient to increase the machiens

sensitivity to very small targets without modifications to the amplifing circuit

and signal processing. I can only see this being tested by trying different

coil sizes against thier Smallest detection sizes. Its not a mater of

micro seconds..... its a much smaller time frame I am guessing but it is

an Edge. Heck it might be totally inconsequential if the time frame is small

enought but it was just a thought.

No knocking the experiment but by reading your reply I have to wonder if you

think I performed the experiment and knew what I was doing. I changed

from you suggested Mono to DD because I know that is where the true

results lay for this type of experiment. I did not agree with your premise

of time delay being affected by addition of metal to the coil. My target placement

and procedure where ruff but accurate enough for what was being examined.

Besides a Mono would falsify the experiment, in my opinion, and is too difficult

to account for in terms of truth in results. Placement of the metal on the disk for a mono

would be an edge placement and the variable becomes Distance from the

reciever. In a DD the overlap zone is much closer to the reciever and the

canceling affect of the coil helps to prove that signal timing is not an issue

to the reciever. Also I used a GP3000 and my proposed thought concerned

a 3500.

For the Coil issue under discussion we are talking about somthing that is

entirly related to the coil and the machien and outside interference has no

bearing exept where noted in the traces on the pictures posted in this

reply.

I am enjoying this conversation though and pulling a fair amount of info

from it. This is not an arguement for me, but a debate. I look forward to

your replies.

Past time for Dinner now, talking soon.

pz

[Reg]

Hi Pat,

Jeez, you wore me out just trying to read all that you wrote.

To begin with, I am glad you tried the experiment and provided the information. Oviously, you ran the tests. You have to remember, I am not exactly sure just what happens on the ML since I don't have one to try. All I can go by is what happens on the PI's I have to experiment with. Will or could they differ? Yep, they sure could. So, I wasn't challenging you or questioning what you said, but rather trying to figure out just where the differences lie.

Ok, now to try to answer your questions, for starters, your assessment in the second pic is not quite right. It is a good accessment but needs a little adjustment.

Remember, the lower trace is the basic signal as seen across the coil. This signal is what is fed via a resistor to the input of the preamp. Also, remember, the upper trace is the output signal from the preamp.

Now, with that in mind, the area you have bracketed in white on the lower trace is part of the actual ringing of the coil as is the next hump you have outlined with the red horseshoe. In other words, the ringing is about 1 cycle long. The frequency of this oscillation is about 500khz. This ringing ends about where the right side of the horseshoe meets the trace.

The smaller display above, where the signal goes down, then up then back down, is simply the preamp output trying to follow the oscillating input signal. The area of the upper signal you have circled, the ramp up could be the result of the ringing/damping mechanism or it could simply be the result of the shielding material used. Some manufacturers use a coating of shielding paint that contains nickel. A sufficient amount could result in the decay you see. Finally, there is also the possibility that this slow ramp was planned, meaning the manufacture introduced something that would cause such a ramp in signal, maybe to assure proper damping. Personally, right now, I am not completely sure of the cause. Strange things happen in coils and the strange responses hard to explain. In many cases, the answer is not that easy to determine.

Your evaluation of delay is close but not quite. Just when you can sample the receive signal is a matter of several factors, one being the gain of the preamp. If the gain is very high then any sampling of the decay curve will most likely cause a severe case of "nickelitis", meaning the signal will be muted because of the preamp overload, much like what happens if you place several nickels on the center of the coil. To assure that doesn't happen, most PI's sample after the decay has settled down and flat lined. I am fairly certain that is what ML does also, based upon your testing.

Now, if all sizes of gold are the desired objects to be detected, then it is absolutely necessary to sample as soon as possible. Small gold signals decay extremely fast and their decay time coincides with the coil decay. So, if the decay of the coil is somewhere between 7 and 10 usec and the gold decay time is 5 usec, then there is no gold signal present after the 7 to 10 usec has passed so any sample past the 5 usec time will indicate nothing. If it takes 7 to 10 usec for the coil to flat line, then the earliest possible sample time would be set by the 7 to 10 usec plus any associated delays that occur within the preamp. Normally, the preamp will add a few more usec's to it.

Right now, based upon what I have been told, the best guess is the SD's sample at about 15 usec. This would be in line with the size of small gold I have seen them detect on a consistent scale. The GP's are obvously sampling sooner because they are more sensitive to smaller gold. Now, based upon what I have observed, my best guess is they are sampling at about 12 usec or so. Why do I say this? Well, my low powered unit did sample at 10 usec and I could barely detect the "invisible nuggets" that were "invisible" to the GP's. So, the GP had to be sampling later.

Ok, getting back on track, it is imperative that the coil settle down and "flat line" as quickly as possible. The slightest amount of osciallation, or "droop" as the result of something such as the nickel could easily cause a coil to not work or not work correctly.

BTW, the droop, or slow rise to flat line that you have circled is also sort of indicative of what happens when the FET heats up. In other words, heat can cause the droop type signal to occur where there was no droop to begin with.

DD coils: DD coils can sort of ground cancel by themselves, simply because of how their detection zones react. A DD coil has 3 main detection zones, if we ignore the outer edge and what is out past the edges of the coil housing. These three detection zones are: the narrow center overlap strip running front to back. This is the zone where the two inner windings overlap. The other two zones are the areas above below the two windings themselves outside the overlap area. This would be the larger areas on either side of the center ovelap area.

Now, any target entering the winding area will cause an opposite signal to what would happen if the same target were to enter the overlap zone. So, if a nickel causes a positive response in the center zone, then it would cause a negative response in the winding areas. This also holds true for ground signals. As such, since the ground is generally under the entire coil, some of the signals are opposite those under the overlap area and as such, do try to cancel some of the ground signal caused by what is under the overlap area.

BTW, just for fun, you might try mounting nickels on the other detection zones and see what it does to the coil and how it works. Also, simply passing a nickel slowly across a coil, one should be able to hear the signal change in all three zones. If it doesn't, then ML is controlling the output signal by synthetically creating the audio response. This can be done by rectifying an modding the signal using other methods. Right now, I doubt they do that.

If ML does indicate the negative dips in the audio as the nickel passes over the two coils, then one should realize that this signal will try to cancel any target signal. This becomes important on deep targets as well as the sweep speed used. In other words, it becomes important to sweep slowly to assure such signals have time to settle down.

I'm pooped. Hope I have touched on the things you mentioned.

Cheers,

Reg

[Capricornstreasure]

Req,

I think you just "Coined" a new term, Nickelitis. *ROFL* (oh thats just to punny)

I see where your coming from and I am considering it~ will for the next day

or so. As to the new experiment I'm also considering how to test

your thoughts. Differents strokes for different machiens indeed.

Appologies for the eyestrain~ I felt it on this end to but I could not

condense it without lossing alot of meaning~ that and its my hope others

can follow it.

As to the traces, I'm digesting that to and the Fet info as well. It is repeatable,

the ML traces, right? Just have to ask for my own touchy feely of the situtation.

That is a HUGE ringing, I would expect something at a much higher frequency

but I can speculate that what you say maybe true about the coatings and

shielding or there is some sort of capacitive effect that is not being accounted

for..... Like I said, I'm digesting it..... Hopefully I wont have to blow another

Meg of Robs space in reply.

Well moving on for the next bit. Talking soon~ Anyone else want to play in

the pond come on in..... the water is fine...... Nickelitis..... BWAHAHAHAHAHAHAAA

I need that.

pz

[Reg]

Hi Pat,

Hopefully, Rob will jump in and let us know what happens with the 3500 when he runs the nickel test.

Now, one quick thing, the frequency of oscillation of a coil is typically in the 500khz range. The ideal coil would have a higher frequency, but that is extremely hard to do. Generally, special winding techniques or are used to reduce the capacitance involved.

There is quite a bit of technical info on coils, PI's, etc on the Geotech forum for those interested. It will take a while to wade through it all though. Here is the link to the forum:

http://thunting.com/geotech/forums/

From here, one can follow different threads about PI's, VLF's, look at schematics, learn more about coils and of course, more about PI's.

Have fun reading all that is there.

Cheers,

Reg

[ELDORADO.VE]

Hi Reg,

Attached are two pics of the scope. The divisions are 1us. The coil is 310 uH, 11" diameter and 12 Ohm, including resistor. Tx is 65us.

The difference between the 2 pics is only different clamping diodes, all other factors equal.

This makes it obvious that by changing to different types of clamping diodes, some faster sampling rates can be obtained.

I wonder if you have tried some Schottky diodes?

ELDORADO.VE

[Reg]

Hi Eldorado,

I am curious as to the detector you are using and why damping diodes are being used at all? Most PI detectors do not use any limiting diode for damping and rely on the FET to provide that form of peak limiting. Then the damping is done with just a resistor. By doing this, one can shorten the decay curve recovery time which will allow for faster sampling.

Now, is this damping diode connected directly across the coil, or somewhere else in the circuitry? The next question is just where within the circuitry is this signal pic taken? Also, what numbers were the two diodes in question? By chance, you are not referring to the back to back diodes connected at the input of the opamp, are you? If so, these diodes are simply limiting diodes.

The polarity of the signal on the pic looks like it might be a pic of the signal at the coil itself, but one can invert the signal at the preamp and get a similar response. If it is the coil response, you might want to just use the FET for limiting and see how fast the coil recovers. My guess is you will gain at least a usec or two.

Now, if all that was changed was the diode in question, it appears that the key difference was the diode's capacitance. This can vary signficantly between diodes.

BTW, if another coil is brought close enough to the operating coil, a similar response can be generated. So, one has to be careful that the distortion is not caused by an external source.

Your coil is obviously made with small wire to have a 12 ohm resistance and 310 uh. You mentioned a resistor is included in the resistance. Just where is the resistor in the circuit?

Finally, to answer your question, no I have not used any Schottky diodes in the damping circuit. The reason is I don't use any form of diode for damping, other than what is in the FET case itself. Since most FET's have built in damping diodes, there generally isn't any need for addtional ones.

Reg