Ground balance on SD 2100 (Reg)

[Anonymous]

Hi Bob,
Black sand acts differently on a VLF than on a PI.
You are right about how easily black sand can cause problems on a VLF. I have demonstrated how easily it is to miss a gold nugget hidden by black sand multiple times to club members of the GPOC in Colorado Springs.
Actually, it is easy to get the XT 17000 to miss a gold nugget in black sand if the machine has first balanced over typical ground. It takes about 3 passes or so to balance to the sand. Until then, it can easily miss a hidden nugget also. I know, I owned a couple of them and still own a couple of XT 18000's.
Now, black sand, i.e. magnetite particles, the ones what produce a fur ball on a magnet do not generate a signal on most PI's. (I am sure this will cause people to disagree, but the signal from black sand is gone in just a few usecs, well before the sample is taken).
What normally causes the signal is the "void" caused by the black sand. Interlace a deep stringers of black sand in maghemite clay and what you have would be more like lumps or ridges of the clay as far as the PI is concerned.
Some time back, I had three 5 gallon buckets of black sand I poured into a large plastic barrel I had cut in two (lengthwise). For a short time I was able to test the effects on nuggets buried in this stuff and test just how little signal the black sand (BS) produces on a PI. I could bounce the coil up and down, and sideways over the BS and get little or no response. Now, the sand does have an impact on the depth capabilites but not like one would expect.
Unfortunately, my grandson decided the plastic barrel would make a great base for a bike ramp and poured the black sand out. It took him a while because it was heavy, but he managed. I never was able to gather up the sand again. Ah, life's little frustrations. Now, if you think that is a bummer, you might be interested in what he did to my two ATV's, but that is another story. BTW, ATV's don't float. Well, one sort of did but my Polaris 500 didn't.
Reg

 

[Anonymous]

Hi Russ,
I mentioned this below, but normally black sand will not produce a signal on a PI. Generally, the response from black sand is gone before the sample is taken.
Now, I was able to get a piece of magnetite to generate a response on a cheap PI kit that had no earth field effect compensation. However, it was suspected that the PI responded to the low frequency signal generated by the weak magnetic field concentration caused by earth's field. Most factory made PI's compensate for this and cancel this effect.
The explanation about the clay or charcoal causing a signal is valid because both generate a signal that lasts a long time so it is detected when a sample is taken.
I think the key words in what you wrote pertaining to the clay, etc, were "can produce a sudden signal variation". Add any form of clay and black sand stringers and you will produce a form of "sudden variation" which can generate a response.
The Alkali is more likely to cause a problem on a VLF, than on a PI.
Now, the question about the 100K pot vrs the 50K pot. I think the idea here is to give you more of an adjustment. If there is a ratio of resistors used in the ground balance circuitry, then increasing one could increase the range. However, as I mentioned, I am not sure just what ML is doing so I can't say for sure.
Salt Coils, also known as noise cancelling coils are built such that they have a "figure 8" type receive coil. One can think of this a long (front to back) receive coil that has been pinched in the middle, length wise, and then twisted one half turn. So, one half of the coil faces up (front half maybe) and the other half faces down (back half).
Half of the coil creates a positive signal and the other half creates a negative response. If they are over flat ground that contains a lot of mineralization, then half of the signal is plus and the other half minus, so they cancel. The coil will work great over flat ground and effectively cancel the ground signal.
Target signal response will be a little strange on these coils. If the front half produces a positive first (high low) signal, then the back half will produce a negative first response (low high). The middle sort of produces a much weaker response, and follows the stronger of the two coil signals.
They are called "Salt coils" because they work very well on the salt flats of Australia where the ground may be composed of clay's that could produce a very strong signal. However, since the ground is very flat and most likely quite uniform in composition, the coil design cancels the effects of the ground.
Since half the coil faces up and the other half faces down, the two halves cancel any noise that they absorb, so in noisy environments, they do work extremely well.
I have not tried the Coiltek brand but have built a few "salt coils" just to see how they work. They are interesting to say the least.
Reg

 

[Anonymous]

Thank's for that Info. Reg! I won't ask about the ATVs.... <img src="/metal/html/lol.gif" border=0 width=15 height=15 alt=":lol">
Take Care Reg!
PH Bob <img src="/metal/html/wink.gif" border=0 width=15 height=15 alt="">

 

[Anonymous]

Hi Reg,
We have hunted the wash from every direction, I know that sweeping from side to side rather than up or down the wash will sometimes help with mineraliation and it didn't help. We have not tried the idea to raise the coil 6" above the surface, I will try it next time I have a chance. I have a good friend that raises his coil in hot ground with his gold bug 2 and does very well. Thanks Reg

 

[Anonymous]

Hi again tj,
I forget to mention it's a dry desert in the southwestern part of Arizona. The wases only run during rain storms and can be prone to flooding during our monsoon season.

 

[Anonymous]

Russ, your instruction manual is right. 2100s were prone to saturation on ch2 and this is well known in some areas here in Oz.
You need to go back to a factory battery and coil to test it and make sure you don't use a coil extension lead.
If your problem is magnetic materials and not graphite or conductive soils then supplying a higher than normal voltage or anything that suggests more current will add to your problem and some won't balance ch1 at all on some ground with some after market coils.
Last decade, I met a guy that used to run his stock 2100 for up to an hour before using it on ground containing large amounts of magnetite and since then met a few others that did the same. I thought this was silly but tests made me change my mind. They did this to wipe off the higher surface voltage on a fully charged battery as it widened the signal just enough to cause momentary saturation as the coil passed over stringers of concentrates under the soil making it noticeably noisier.
Conventional pi sample after a high gain pre-amp at a point where the spike drops to what the designer, the gains used and the supply rails designate to be zero and this is always well back from the spike itself.
This gives a small signal for a select large sample of magnetite because it decays early and a very large signal for a select sample of ironstone soils. Minelab sample at the spike itself to obtain a relationship that can't be obtained otherwise and the convention is then reversed and the same magnetite sample is then the most likely mineral to widen the clamped spike and cause saturation whereas the same ironstone sample, even though it has a huge decay time, widens the actual spike by a much smaller amount. So add the effect of magnetite to a higher battery voltage, a coil that results in a fatter signal and you have a better chance of saturation. Try using ML coils and a stock battery and see if this shuts up the wash first. You won't lose depth in this ground regardless of what you may think. You might then try a NF coil, as theirs are very close ML specs.

 

[Anonymous]

Hello Robby,
Well I feel like I'll be going backward, all the nuggets found on ch 1 with big monos were deep 18" to 22". I don't think the ML 11" DD will find anything that deep, at least I could prospect the rest of the area if it works. What you are saying could very well be my problem and I'll try the stock 6v battery and stock 11" DD coil. Thanks for your responce!
Regards, Russ

 

[Anonymous]

Hello again Russ , the terms 50k and 100k refer to the resistance of the GB adjustment controls ( 50k = 50 000 ohms and 100k = 100 000 ohms ), if you change these two components ( you must change both )you will reduce the sensitivity of your machine by only a small amount but you will regain the use of ch2.
All in all you cant harm your detector by changing these pots .
Something to consider is that when searching for large and deeply buried objects with an SD is that the ch2 on your machine is not used as it is a higher frequency channel ( sensitive ) designed to detect smaller targets closer to the coil and it runs in tandem with ch1 which is a lower frequency channel which is simultaneously probing the ground for the larger , deeper targets so in reallity you are not losing depth by switching ch1 off only sensitivity to smaller targets closer to the coils underside .
If the ground noise is a constant howl without much variation in an area that recieves the runoff from a watercourse and is the lowest point in the area then the cause is almost invariably salt saturation and by using a salt coil you will overcome the problem although you will have to get use to the different response of the coil to a target ( see Reg's post )also note that a salt coil is also more sensitive to magnetite in the form of iron oxides eg: hotrocks etc and may cause more grief if the ground is loaded with this type of mineralization.
All said it wont hurt to experiment a bit in order to stabilise ch2 to regain some sensitivity to the smaller gold will it ?
Regards TJ.

 

[Anonymous]

Hi Robby,
Interesting stuff you mentioned. At about what point is ML sampling on the decay curve? The reason I ask is I would expect the SD to detect much smaller gold if the sample was soon enough that it detected the magnetite. When I tried to determine the time constant of a magnetite hotrock, it appeared the signal was gone or almost gone at about 5 usec or so. To be able to have magnetite cause a problem, I would expect the SD to readily detect 1 grain nuggets. That isn't the case with my dad's SD.
Since his SD 2100 doesn't do that well on very small nuggets in the grain range, this would lead one to believe that the sampling delay time is quite long, maybe 15 usec or so. Other SD 2100's that I have seen acted about the same.
To add credence to this idea is the fact that the "invisible" gold nuggets sent to me by John Blennert can't be detected until the delay is down to less than 10 usec and most ML's including their new 3k can't detect them. To get a decent signal on these nuggets the delay has to be about 7 usec to 8 usec. Now, this short of a delay still doesn't enhance the magnetite signal with any significance. At least, it doesn't on mine.
Reg

 

[Anonymous]

Russ, it was only meant to be a starting point and if you could beg, borrow or steal a larger coil from another brand then take it from there. TJ's idea with the pots would also be worth noting.
Reg,
This is my take on it. I maybe should just post pictures and no text, as I’m not the best at getting a point across.
As I see it, magnetite has no chance of causing saturation in your detector.
It has an effect that follows the decaying spike and dissipates in a time related to the spike width as we normally see it (with normal scope scaling) and this is usually a few times its width but it should happen a lot quicker in your circuit if you aren’t using a circuit that has the current rising rapidly in the coil at switch off and have a spike that isn’t clamped or doesn’t get to avalanche. Added energy has some place to hide and you won’t see as much effect from magnetite at the spike.
Sampling after slight overshoot can result in a further loss of the effect. These things would influence the outcome at the spike’s slope and curve quite a lot.
You are looking at time in regard to coil switch off but you are also pushing the spike back away from your sample as much as possible.
In Candy's tx method the clamped spike’s width reflects the rather large difference in energy dissipated in each pulse and magnetite’s (and everything else’s) effect is crammed into the spike’s width.
Looking at HH_SHARK_decay1.jpg that you posted over at Geotech then your 4.63 sampling point is roughly the spike’s width back from the spike itself and it may well be more sensitive to magnetite than in another picture that you posted of your signal using a ML coil. At 8usec and noting the overshoot, you are at least > 3 times the spikes width back from its curve to zero as seen at low gain or using a scopes usual scaling and the magnetic effect from magnetite and other minerals containing the effect should be lower.
If you could make the spike even narrower by using less current by resistive limiting, changing the coil properties or other means, then magnetite would have an even lesser effect even if you still sampled early at 4.63usec. You would be then placing your sample further away from the magnetic effect but you are also left with some minerals long decay time. There is a catch though as you lose energy in some cases to gain an effect.
Regarding ML sampling very early………..
If you were to use MLs method of transmission, which uses a low resistance coil circuit with the long and short pulse spikes deliberately clamped and then look at the slope and curve itself, you would see a variety of effects that conflict with the normal view depending on the scaling used on the scope.
Sampling early on the slope still makes it sensitive to smaller nuggets but can subtract from slightly larger ones and subtracts even further from longer TC nuggets and it needs pictures to make this effect clearer. It no longer obeys the rule as seen back after the curve to zero.
Most iron objects such as a nut, a bolt or nail on end or a hammerhead would give an increase regardless of where the sample is taken. Sampling at different points down the spike’s slope and at the curve gives a rather conflicting bag of results if only one pulse length is used but sampling after the curve gives results as in normal pi.
Like I said, pictures are needed to get this across better.

 

[Anonymous]

Hi Robby,
ML's waveforms look different than mine for sure. Quite a while back, I took pics of the ML output using a test coil of wire, a loading resistor and placing this test coil a finite distance from the ML coil. In this case, the test coil was about 5 inches from the ML coil. It is a great way to "see" what is going on in a non invasive way.
I have attached a pic of what I saw on the ML. This pic displays the decay curves of the two different pulse widths superimposed on each other.
What isn't shown here is I also took a similar sample of the output of a Goldscan and found it to have a stronger pulse.
Now, I agree that if you take a sample while on the decay curve you can get strange results. I have done this and found you can increase the sensitivity, but also make the detector more sensitive to earth field effects.
What I don't know about the ML, is what type of IC they are using as a preamp and the gain of the stage. The pic attached displays the decay curve takes well over 3 usec, maybe 4 usec or so, to fully decay on the short pulse and over 6 usec on the long one. Now, unless ML is using little or no gain in their preamp, any sampling should detect a saturated amp any time before it nearly fully decays. In this pic, little can be done before about 4 usec or so. Now, if they were to sample at 3 usec, then 1 grain nuggets should be readily detected. However, I am sure there is some lag time through the preamp, so I suspect that any sampling on the decay curve would have to be at some time greater than 5 to 6 usec or so on the short pulse. Even at that time, there should be a decent detection of very small nuggets. However, the SD does not display that type of sensitivity. At least, my dad's doesn't.
Reg

 

[Anonymous]

Hi Reg, the effects are different than after an amp!
I would have liked to prepare better pics but here are some old ones for now.
Spike (A) is an actual picture of the decay from a 235usec pulse with the two waveforms overlaid. The gray line is the coils signal in air and the green line is the effect from a long TC object. You can see that a large nugget can give a small response if you were to pick the right section to sample.
There is an overall amplitude increase here but it is too small to see. The green line should also be seen as being a bit to the left at the curve itself. The hash and line width tends to hide some of the effects.
I have cheated in (B), as small TC objects effects are harder to see and have rather over exaggerated the response as seen in the horizontal part of the signal after the curve. This isn’t all that unfair as it depends on the scope’s scaling and scaling is what it is all about. The green line meets the gray line further up for a small nugget and further down for a larger one.
A selected sample here of magnetite shows a large increase in the spikes width with a small increase after the curve. A select sample of ironstone type soil shows an equal widening of the spike and increase in amplitude after the curve but the widening effect is much smaller than that of the magnetite sample. If the effects of these samples are measured after a high gain pre-amp then the magnetite gives the smallest signal and the ironstone soil sample gives a very large signal. Obviously, scaling plays a large part in these effects.

 

[Anonymous]

And here's (B)

 

[Anonymous]

Hi Robby,
Your pics display info very similar to the info in Ott's patent on the Periscope PI. If you sample far enough up the decay curve, you can develop a form of discrimination and, yes, I would think the magnetite should provide a wider signal.
However, I am not sure that is what ML is doing. They could be, but it is highly unlikely if they are using a uprocessor for their timing. If they are using a crystal and simple dividers then, yes, it is possible they are sampling as you state.
However, if you do sample on the curve, then I still suspect the sensitivity to extremely small gold would be much better than I have seen on the SD. There is no reason that I can think of that would allow the SD to sample soon enough to "see" the magnetite change, but not see small gold.
Stubborn aren't I?
Reg

 

[Anonymous]

Hi Reg,
When I tested the SD2200 on various metal targets, I was getting the same sensitivity as with the earlier Deepstar, which sampled at 15uS. Also the ML coil barely recovered by 15uS when pulsed with that level of current. This is with the ML coil driven by the Deepstar transmitter adjusted for the same TX pulse current, and looking at the output of the first amplifier.
If you are sampling during the switch off spike, or before ringing is fully damped, the detector will respond to magnetite. This is because the coil inductance is increased by the magnetite's presence and the switch off time and ringing frequency cycles are lengthened. This is more than likely the reason that short delays appear to give more sensitivity to magnetite. You are just running into the zone where small changes in coil inductance are sensed. If a lower inductance, or lower capacitance, coil were used then the magnetite response would disappear again.
Eric.

 

[Anonymous]

Hi Reg,
I think we are on different tracks here. A pre-amp can't be used if you wish to sample at the spike or even up the curve, as it wouldn't fit common supply rails. It needs reducing before sampling and filtering and gains are applied after this. The curve as seen after a high gain pre-amp can be either that of a fully damped coil actually settling later in time or the result of some fiddling but this curve is usually well back from the spike's actual curve itself as is shown in your HH_SHARK_decay picture.
In Candy's patents, his text and drawings suggest waiting several times the true decay in order to avoid the magnetic effect and this puts him back quite a long way and I don't think he does it this way.
Sampling early on the spike's slope gives different effects but the term 'sampling early' looses some of its meaning here and doesn't mean more sensitivity but sampling after the curve gives a response much as you would expect. This is explained in US5506506 where he does intend sampling the spike but for other reasons. My pictures only show one set of results that can be obtained. You can actually reverse these effects somewhat.
Anyway, the only way I can get the results needed is to sample the slope... leading to the assumption Candy does also and he may even start his sample at coil switch off. If so then it would end at around 6 usecs in a 240 usec pulse's curve if it was supplied by 6v. The next sample might then start immediately. I might still be wrong but this is where I am at.
If he does sample here then supplying a higher voltage like 7.3v would result in what can be seen as a timing change of at least 1.6 usec (according to one of his idealized formulae) without moving the sample's actual position. Using a coil with different properties could do much the same, as everything is then either added to or subtracted from the spike's effective width.
Current limiting the tx signal, overshoot before sampling, not clamping the spike etc all lessen the magnetic effect and it is then small in relation to the coil's inductance change caused by magnetite. The idea of sampling the tx waveform and subtracting this to obtain ground balance can't be used with a tx waveform that flattens out well before switch off because the magnetic effect is too small to use. The magnetic effect is large in Candy's tx model.
Anyway, I think we can imagine the effects that can lead to saturation or the things people could do to make it even more likely to occur.

 

[Anonymous]

Hi Robby,
I am not disputing that sampling early would cause the magnetite to generate a strong signal. In fact, I suspect if sampling was done early enough then magnetite would create the problems it does on a VLF. That would be counter productive.
I also agree that more information can be obtained about the target if an early sample is taken. This could have a positive effect.
However, right now, I can also see the potential for more problems than solutions by sampling on the decay curve. I should also say that I haven't spent any time trying this though.
I am assuming that your ultimate objective is a better ground balance circuit, rather than just wanting to know what Candy is doing.
Personally, I would rather not know what Candy is doing, specifically. I find it allows me to thing of things that may be totally different alternatives, rather than try to verify or improve on what he what he is doing. I find it is too easy to get "trapped" into thinking along the same lines. Sometimes I feel it is just better to think outside the box.
I think it some ways that is what you are doing but state much of what you find to verify what Candy is doing. Please keep up the good work. I enjoy your thoughts on the subject. In many cases they are way over my head. In any event, your ideas have always given me new ideas to try on future projects.
Reg

 

[Anonymous]

Hi Eric,
Thanks for the info. I am going to have to look more closely at just what is happening when I do sample while on the decay curve. I have tried my best to avoid it up to now, but I can see that I do need to know more about what actually happens, rather than make assumptions.
As for your info about the SD timing, it is along the lines of what I have found also. It is a very sensitive detector, but the lack of sensitivity to extremely small gold indicates the delay is in the 15 usec range.
Although this is a guess on my part, I suspect the later delay is the result of the pulse used. A stronger pulse could result in a longer delay but more sensitivity on some gold sizes. This would be in line with what some people are saying.
Later generation units could be reducing the pulse on current on their short pulse so they can shorten the delay a little. This would or could result in more sensitivity to smaller gold, but could slightly reduce the gain or sensitivity to larger and deeper gold. Some people have indicated something along this line. Again, this is totally a guess on my part, but it does make sense.
Reg