coils

[Tinkerer]

Hi Reg,

I do not have all the answers for you right now. The gold bearing region starts about 300 miles down the highway that passes my property. (No gold on my own property) Hi have traveled through the region a few times but never looked for gold myself.
I know that the gold is being mined in every which way, under ground, surface, and in the rivers. In some places it is very fine dust that presents its own problems, but for us the interesting areas are where the gold is nugget size.
In some places there are thousands of diggers with picks and shovels that dig on claims that are not more than a few feet square each. In one place 120,000 diggers started digging on top of a hill and stopped 20 years later when in the place of the hill there is now a huge crater over 1000 feet in diameter and 1000 feet deep that has now a lake on the bottom that defies the pumps. At that place the largest nugget was something around 150 pounds. This happened about 20 years ago.
I believe in the year 1997 the output of the one location closest to me was around 700,000 ounces of gold, dug by hand by about 22,000 diggers.
So back to the size of nugget. For a start I would aim at about one gram. I have a nugget of 0.6 grams that I can use for testing. This brings me to another question: I borrowed some detectors to go try them out in the close by region. After that expedition I will be able to answer many more questions about soil, mineralization, size etc. Could you, or somebody from the forum give me some advice of how to run a small test series with these machines, so that we could use the information gained in a usefull way for designing the Pi?
The detectors are one Fisher Mscope, Impulse with a round 8 inch coil, on airtest it only just detects the 0.6 gram nugget. Then there is the Fisher Mscope Aquanaut 1280X with a DD coil of 10 inches. I also have an old slow PI machine 2000 pps with a 6 inch coil where I made a mod for ground balance to be tested.
With these machines I will make some test runs in the gold grounds and after that hopefully I will be able to answer a lot of questions.
Trash should only be a problem near places where lots of people live. Out in the wilderness you can spend days and weeks without ever seeing another human.
About the coils: I wound 4 coils. The first one was 6

 

[desertdon]

Those detectors aren't going to tell you much, because they aren't gold detectors, you may detect something if it is large enough. Reg explained that you need to be down to 10Usec Pulse Delay to detector smaller nuggets in the 5-10 grain range. The signals are very small and untrained people(natives) may have trouble telling signal from threshold. As he said the real problem is design DD coil to run at the lower number Usec Pulse Delay. A aquesdtion, if the gold is in general so small, why use a detector and not other inexpensive proven mechanical systems? apparently you do have water locately, you would probably find more gold continually as opposed to looking for small gold with a detector that can't do the job or people who don't understand. If you are just looking for small gold, all you are doing is mining the surface, because the detector will only detect a few inches. If you are looking for larger gold and it is possible there, then that is a little different. Don

 

[Reg]

Hi Tinkerer,

It sounds like the area has potential and hunting with a detector just might work out ok. Hope you get a scope soon. That will help in the evaluating different circuit changes.

Unfortunately, your selection of detectors is not the greatest for nugget hunting. I would like to see a copy of the schematic of the PI you built to see just what might be done to help it. My guess is there are a few things that could improve it and could be done easier than building a totally new PI. This would give a little better PI for initial testing.

Now, reviewing your post about your coils and the sizes/inductances, I would recommend you shoot for a coil in the 200 uh to 300 uh range. This will speed things up quite a bit when compared to your 1000 uh coil. The type of wire and the insulation of the wire make a difference also. Magnet wire makes a smaller diameter coil, but will have a much higher interwinding capacitance also. Simple stranded hookup type wire will work much better. Pick the right insulation, i.e. Teflon and things get better yet. Keep the coil smaller, maybe 8" to 10" max. A smaller coil has less capacitance also. If the capacitance is down, then a larger damping resistor can be used. This will speed things up also.

Now, if the area is very rocky and the gold is small, then getting the coil closer to the ground will be quite important. In this case, a smaller coil yet might be a better choice. Your idea of a 5" by 12" or so, rectangular coil might be better yet. You might even consider something like a 4" by 12" coil. This would take advantage of the increased depth because of the 3 to 1 or 4 to 1 ratio technique.

Surprisingly, coil shielding helps also when trying to speed things up. I would recommend using a technique similar to that used by Eric where the windings are wrapped with a spirawrap and then a shielding tape is applied. Care should be used in selecting the shielding material.

Next, I would limit the current in the coil by adding a series resistor between the FET and the coil, maybe a 24 ohm or so. This will also speed things up and add some stability to the circuitry so it isn't so likely to be unstable because of power supply problems. It is far better to have a very stable PI that generates less power in the coil than try to maximize coil current and having the pulse current and the audio interact. Remember, it takes a whole lot of extra current to gain just a little depth.

Now, if you look at the schematic of the Hammerhead (latest version) you will see that there is no cap feedback or any compensation of any sort on the preamp. This is about as good as it gets to get the delay down. Even a feedback cap of a few pf could slow things down a lot.

As for the saturation problem with the NE 5534, this is common but reduce the current into the preamp and again, you will reduce the problem. So, increase the size of the input resistor. Go too large and noise becomes a factor as well as a loss of gain. Go too small and it takes much longer to come out of saturation. Also, keep the damping resistance as large as possible. This will help also. In fact, don't be afraid of not having a perfect decay curve. As long as it is stable at the time of sampling, all should be ok.

As for clamping or limiting the preamp, I have not found anything that works well, other than varying the input resistor. I am sure there is a technique that might help, but I haven't found it yet.

Finally, one could modify the PI to allow the use of a DD type coil. This also allows things to be sped up since the high voltage is no longer applied to the preamp. It is much easier to get a DD coil to work at 10 usec than a mono coil. Also, the DD does a very decent job of minimizing the ground signals.

Anyway, the above info is a good place to start.

Reg

 

[Tinkerer]

Hi Reg,
I will look for a legible schema of my old design and post it. In the mean time back to the coils.
Could you explain some more on the 3to1 or 4 to one technique?

About the current limiting: Talking about a +/- 10us delay, and a 3 to 4 time constant TX, that would mean something like: (using standard ohm values and disregarding mosfet resistance and winding resistance to simplify the picture)

Coil # 1, 520uh for a time constant of say 5,2us = 520/100 Ohm
Coil # 2, 280 uh would be 280/ 47 Ohm or 56 Ohm
Coil # 3, 165uh would be 165/ 33 Ohm

Now, having the actual amps in the coil quantified, I can put the turns of the coils into the equation to find the actual field strength produced by the TX.

Timing could then be TX 30us, delay 10us, sampling 10us, interval 40us, second sampling 10us.
The problem I have with that, is that with the high resistance of the coil, ever more TX current gets bypassed trough the damping resistor.

Question: could the damping resistor be switched off line during TX, leaving maybe a 2k resistor on line? Then bringing it back on line for the flyback, maybe 2us after switching TX off. During the 2 us the flyback would damp trough the 2 K, (choice of value as high as the maximum flyback voltage allows) thus discharging a lot of energy fast.
One more question: The capacitance of the mosfet adds something like 18pf, how does a bipolar transistor compare to that?

By the way, for the tinkering I just use a bread board, but I have 2 or 3 old pcb

 

[Reg]

Hi Tinkerer,

Boy, you lost me on this time constant thing. I am not sure just what you mean.

Now, with that said, when I think of a time constant, I have a tendency to refer to a typical gold target time constant which is quite short.

As such, if a detector is to be used for detecting gold nuggets, the pulse on time can be reasonably short, say 50 usec or so and work extremely well for the vast majority of the nuggets that will be found. A 100 usec would be more than long enough for most gold hunting even when large nuggets are present. Personally, I haven't seen much difference between a 50 usec pulse and a 100 usec pulse, so I just use the 50 usec. This reduces the power requirements and still works very well. (To be honest, my Goldquest clone has a pulse on time of about 21 usec or so and it works very well).

Now, if I were hunting in an area where the gold found was typically in the pound range, then I would consider a longer pulse. However, most gold found today here in the US is much smaller than 1/4 oz. As such, a very short time constant works extremely well.

As for building a typical mono coil or even windings of a DD coil, I use a very simple program found on the net to determine the number of turns to meet the inductance requirements. That site is:

http://my.athenet.net/~multiplx/cgi-bin/airind.main.cgi

Now, I use the inch scale, so if I wanted a 300 uh coil 8 inches in diameter, I would simply plug in 8 in the diameter, .25 in the coil length. (The coil length is the physical size of the winding thickness once wound. I use a number between .22 and .25 and get very decent accuracy when using the .1% accuracy calculations on the answer page.

So, by plugging in the 8 for the coil diameter and .25 (.25 inch) for the length, I simply guess an initial number of turns. Lets try 25. With 25 turns, the inductance calculates out to be about 347 uh a coil, with 24 turns calculates out to be 319uh, and 23 turns is abut 294 uh or so.

I would probably use 24 turns. As for wire size, that depends upon the max coil current I plan on using. Since I want to keep the current down, I would not worry too much about using a larger size wire. So, I might use something between 26 awg and 28 awg.

So, to build a coil 8 inch in diameter, I would simply wind one using insulated stranded wire with 24 turns.

The fact that you asked about switching in a damping resistor a couple of usec after the pulse is turned off indicates you have done quite a bit of work in this area. Yes, I am sure it can be done, but with care, it won't be necessary. One can obtain 10 usec delay without doing that. Simply keep the pulse current lower, reduce the gain of the preamp a little and use a DD coil. This type of coil would minimize the ground signals which probably would be severe at the 10 usec with a mono coil. With the above mentioned combination, the detector should work just fine.

Now, to answer your question about FET's and transistors. Generally, the FET will work faster. Select one with a reasonably high breakdown voltage, say, minimum of 250V up to 500V or so. As for the FET capacitance, that varies as the voltage across it varies. So, I am not sure just what it is for an average. If you pick the traditional 740, then it will work ok. I generally use a different FET number for a little more speed.

As for the 3 to 1 or 4 to 1 ratio, I am hoping Eric will jump in here and explain the advantage. The ratio refers to the length to width ratio of the coil. This technique of coil building to gain a little depth is something that Eric mentioned quite some time back. Since I have not actually tried it, really can't comment on it much. It is something I am hoping to try shortly, though.

Hopefully, we can modify one of your old pcb's and have it work ok for testing.

Reg

 

[desertdon]

Carl on Geotech has built retangular 3-1 coils I believe, maybe he can offer some assistance. Don

 

[Tinkerer]

Hi Wirechief,
thanks for your offer of help. I think you are already busy on the next step, the receiving of the signal, and are making good headway. Your background in radio just might come in very handy.
Here are some thoughts of mine: could we make the coil resonant? The question now is: to what frequency? When I look a t the output of the preamp, I see a square wave composed of the clamped TX signal and the clamped RX signal, both saturating the opamp. But the part that interests me is the slope of the decay that follows.
My very limited knowledge tells me that square waves are composed of a multitude of waves of different length, but also the slope seems to be made up of different wave lengths. Now, out of all these frequencies, which are the ones that can tell us something about the metal object that we are looking for? Could it be possible that different metals actually produce different wavelength eddy currents? Which ones are useless noise to us? If we could identify them we could find a way to eliminate them.
In the mean time I go back to the coils, still a lot of work left on them.
Tinkerer

 

[Wirechief]

Hi Tinkerer and so very nice to meet you amigo. First I want to let you know that this metal detector technology is a whole new ballgame for me. And so right now I'm way behind you Tinkerer. But you are right about the square wave in that it has many different harmonic frequencies in it.So f=1/t and also odd harmonics 3f, 5f, 7f, as amplitudes decrease. At such low frequencies the bandwidth gets very narrow. But from what I understand at this point is in dealing with gold nuggets they are all different in size, density, and shapes of course and so it would seem each nugget would have it's own signature within the return energy or eddy currents. So I don't think a single resonant frequency will cover all the bases. I'm still not sure what the frequencies of the pulses in RF on a PI instrument are. I also understand that PI instruments are not very good with discriminating. But I tell you Tinkerer I have not had a subject such as this detector stuff get me to using the brain like I have been for a long time. So I am going to keep this on the brain for awhile. I will CU later Tinkerer and give me a yell when ever you feel like it. de John the Wirechief.

 

[Tinkerer]

Hi Reg,

Sorry if my language is not very clear. What I meant by time constant of the coil is its transient response. Below a jpg of the page in my textbook as well as 2 jpg of the schema of 3 PCB

 

[Reg]

Hi Tinkerer,

Thanks for the schematic. The design isn't bad except it does use a PNP for pulsing. This probably would slow things down a little. My guess is the transistor could be changed to a decent FET and speed things up a little.

As for the time constant thing, I suspected that is what you meant, but wasn't sure. The time constant involved with the coil is more complex than simply the coil itself. It involves the coil windings (inductance and capacitance), the FET characteristics, the damping resistor, the input circuitry of the opamp, and the shielding for starters. This is what makes it so hard to calculate.

BB Sailor has posted some interesting info over on the Technology forum (Carls forum) about the time constant and how to measure some of the factors involved.

As for the info on the DD and what has to be done, I posted some info on the same forum about the changes necessary. The link is under the Hammerhead project. Actually, the changes are quite easy to do. The key is to separate the preamp input from the transmit circuitry. On your schematic, this would mean cutting the connection to R 5 where it connects to the coil so that end of R 5 is open. Then the receive coil would connect to that end of R 5. If confusing, look at the schematic I posted on the GeoTech forum. Here is a link to the thread that contains the info on how to allow the use of a DD coil on the Hammerhead PI.

http://thunting.com/geotech/forums/showthread.php4?t=10729

Hopefully, the link will work fine. One thing I need to add on the posting is what is necessary to go back to the mono coil. This is a simple fix that is normally done in the coil connector. When looking at the pic, one can see that 4 pins are required (actually, only three are really needed) to allow coils to be switched. A DD coil would be wired similar to the schematic. A mono coil would have a jumper between pins A and B and the receive coil wouldn't exist.

So, allowing a DD coil to be used ins't that difficult. If the coils are never to be switched, then all can be hard wired. If the coils are to be switched, then a simple plug arrangement is necessary.

Reg