Question on MPP coil

[green]

[Remember one key thing about small targets. They need a discharge coil TC that is 5 times faster than the target TC to fully stimulate that target for maximum potential detection. A 2uS target needs a coil discharge TC of 0.4uS. A 300uH coil needs a damping value of 750 ohms to meet this criteria and that means that low a capacitance coil and TX circuit is required to achieve this.]

Playing with spice. Simulation of a fairly high resonance 300uH coil that critically damps with a 800 ohm resistor. (Time constant) L/R=.375usec, critical damped coil=.23usec, coil with diodes and 1k resistor=.61usec. Is the simulation predicting what really happens? Are the decay time differences enough to cause a significant change in detection distance with a 2usec target time constant? Ran the simulation with a very low time constant target to see the effect on coil decay. If I made the target time constant 2usec could the simulation give me an indication of detection distance difference?

Are the decay time differences enough to cause a significant change in detection distance with a 2usec target time constant? assuming the same delay time.

[green]

[Remember one key thing about small targets. They need a discharge coil TC that is 5 times faster than the target TC to fully stimulate that target for maximum potential detection. A 2uS target needs a coil discharge TC of 0.4uS. A 300uH coil needs a damping value of 750 ohms to meet this criteria and that means that low a capacitance coil and TX circuit is required to achieve this.]

Playing with spice. Simulation of a fairly high resonance 300uH coil that critically damps with a 800 ohm resistor. (Time constant) L/R=.375usec, critical damped coil=.23usec, coil with diodes and 1k resistor=.61usec. Is the simulation predicting what really happens? Are the decay time differences enough to cause a significant change in detection distance with a 2usec target time constant? Ran the simulation with a very low time constant target to see the effect on coil decay. If I made the target time constant 2usec could the simulation give me an indication of detection distance difference?

Are the decay time differences enough to cause a significant change in detection distance with a 2usec target time constant? assuming the same delay time.

Tried changing target time constant to 2usec with spice. Looks like target signal was higher with the diodes and 1k resistor, opposite of what I thought might happen. Might be a detectable difference in detection distance. The more I think I learn the less I understand.

[geoscash1]

The more I think I learn the less I understand.

Good gracious, at least I'm not the only one that feels that way...lol

[baum7154]

Hi Dan,

I'll answer those in the order asked.

1 - Are litz and ptfe coils inductance about the same - Yes, all within 10uh of each other, 304uh to 312uh. Resistance 2.1 to 2.6 ohm.

2 - Are the coils critically damped or using the same damping - No, each coil was critically damped using my adjustable damping tool so each coil used a different damping resistance.

As far as self resonance, I have no idea how to check or test for that or if I even have the proper tools to do so..

_____________________________________

Hi Geoscash1,

Regarding the Self Resonant Frequency of a given coil you can use a metal detector and its coil to stimulate the undamped test coil with its PI pulses from a distance of about 12". Then look at the test coil ringing with your o'scope isolated by a 1 to 4 pf cap in series between the probe and the coil. Measure the time of one cycle of ringing and divide that into 1 second, i.e. if you get 1us for one cycle then 1second/1us=1,000,000. The SRF is then 1 mhz. The accuracy of this process is dependent on your scope timing accuracy and the ability to measure 1 cycle of the ringing pretty well.

Best of luck,

Dan

[bbsailor]

Tried changing target time constant to 2usec with spice. Looks like target signal was higher with the diodes and 1k resistor, opposite of what I thought might happen. Might be a detectable difference in detection distance. The more I think I learn the less I understand.

Green and all other coil builders,

The best way to learn is by doing and trying various things until you start seeing some improvements. Unfortunately, with metal detectors and coils there are so many variables that you must choose your battles wisely.

Coil Size. Eric Foster published a graph that basically says that if you can just detect a target at the distance of the coil's radius, that is the limit in coil size for that targets size and TC and any larger coil will produce less detection distance. But if you can detect that target farther than the coil's radius, you can try a larger diameter coil to get more distance on only that target. His chart allows you to estimate the largest coil size to try.

Target Time Constant (TC). All targets need to be charged with a pulse approximately 3 times longer than their TC because 3X will get you up near 95% of maximum current and provide a chance for the TX current to stimulate the target more deeply than just on the target surface. If you look at a LR current charge curve you will see that at one TC the current raises to about 63%, at 2 TCs it raises to about 86%, and at 3 TCs it raises to about 95% and from there on to 5TCs it is relatively flat and does not add much to the target's detectability unless the target you are seeking is massive and could use an additional TC or two to fully saturate.

Delay Time. During the time it takes for the the TX coil to turn off, fully discharge itself (discharge slope governed by the discharge TC) and the flyback pulse with its oscillations caused by coil and TX circuit capacitance, the coil must wait for a quiet time when the receive (RX) time can be enabled. During this delay period, the eddy currents in the target are decaying based on the targets TC. Here is where the damping resistor value is designed to critically damp the coil for the earliest possible time when the RX signal can be enabled. But here are the realities of electronics that can sometimes interfere with our best intentions.

a. MOSFET saturation. When the flyback voltage of the coil exceeds the maximum MOSFET voltage the MOSFET looks like a short circuit for any voltage above that maximum voltage looking like a flyback spike cut off at the MOSFET voltage such as 400V to about 600V or what ever you chosen MOSFET voltage is. Remember, the longer the TX pulse the higher the flyback will be so look at the peak pulse flyback voltage at the longest TX time you might use. The optimum damping resistor value is to allow the earliest sampling time without catching the very end of the damped coil discharge oscillations or the residual eddy currents still decaying in the coil, coil shield, coil housing or coax wire near the coil.

b. Coil Capacitance. Coil capacitance works with the coil inductance value and the coil discharge energy to form a short term LC circuit that still oscillates after the initial flyback pulse decays to 0V and pops up again at a lower value and again at a lower value. This is called a damped oscillation and something that an optimum value of damping resistor should minimize so the stable 0V point is as early as possible and the RX time can begin. Now, any extreme spike voltages should be kept below about 0.6V by the two clamping diodes. The first op amp stage gain will send it in to saturation if it's gain is set too high or the supply voltage is too low. Here is where two stages of about 33 gain is better than one stage of 1000 gain because the op amp comes out of saturation faster and thus less potential delay time and more target eddy currents to detect.

c. Coil discharge TC. Eric Foster mentioned a research paper he read that said that a target stimulation TC that is five times faster than the target TC will stimulate it as good as if the turn off was instant. With that being a reasonable law of physics that tells us that if you want to fully stimulate a target there is a point where you should aim. This point is defined by the coil inductance divided by the damping resistor effective value. Why effective value and not real value you might ask? The input resistor to the first op amp stage usually has an input resistor of from 1K ohm to a max (that I have seen) of about 2.2K ohms. This value is effectively in parallel with the damping resistor because the clamping diodes conduct at any voltage above 0.6V. More capacitance in a coil, coax cable MOSFET or PI TX circuit causes a lower value of damping resistor and thus limits the smallest and lowest TC target that can be effectively stimulated by that coil and circuit configuration.

Choosing to pump more energy into a TX circuit will improve the depth but at what consequences? Theory tells us that to double the detecting distance you need 64 times more power. It is 2 to the third power for the TX signal going to stimulate the target and another 2 to the third power for the return signal in the target to reach the coil or 2 to the sixth power or 64.

Eric Foster did a very clever thing, he used a lower power TX signal but at a very high frequency and integrated many thousand samples to improve the signal to noise ratio in low TC weak targets at the lowest delay. He choose to use the principals of a "Lock-in Amplifier" to improve the signal to noise ratio and approach the problem on the RX side rather than on the TX side. This now means that the amount of time the target stays within the coil field determines how many samples are integrated. This translates into "slow down you search coil sweep speed" to gain the full advantage of this design!

As you can see, if you have been following along so far, is that variables often do not work independently, they are sometimes and often related. This is why you must choose your battles wisely; know how what you choose to work on is related to other factors and seek to understand the big picture so your work can be productive.

Have a productive 2017.

I'm done for the year 2016.

Joseph J. Rogowski

[baum7154]

Thanks Joe and Happy/healthy New Year!

Your post is a good compilation of the issues to consider for anyone building a detector especially for small/short TC targets (exactly like small gold). Design is almost always a trade off or a balance of desired performance outcomes.

With the 3DSS coil design my goal has been to minimize coil, shield, and feed line capacitance to optimize for small gold. Now I'm looking for the proper (175/46 ?) PTFE insulated Litz wire to see how the coil TC may be improved.

My detector has always been the Chance PI but its original design had limitations when seeking small gold. As you mention, a 2 stage amp is much better for avoiding saturation and I have modified my detector to achieve this. The original Chance PI firmware is not capable of sampling below 8-9us so testing of coils and targets at 5us was not possible. I modified Chance with overclocking to allow a 6.98us sample delay...still short of the desired 5us interval. Now Dantech has rewritten the firmware to allow sampling down to about 3-4us if a coil capable of operating at that delay is used. There are other changes to transmit pulse rate and width, and sample window width as well. While Dantech is not interested in small gold targets, I believe his modified firmware will offer improvements to what I have done to make it a better small gold detector. I need to put his new firmware in my Chance to evaluate performance on small gold. Hopefully it will enhance detection of 5 grain and smaller gold targets.

Thanks for putting your post together. It is good to have a copy around as a reference to insure we are addressing all of these design issues.


Regards,

Dan

[bbsailor]

Thanks Joe and Happy/healthy New Year!

Your post is a good compilation of the issues to consider for anyone building a detector especially for small/short TC targets (exactly like small gold). Design is almost always a trade off or a balance of desired performance outcomes.

With the 3DSS coil design my goal has been to minimize coil, shield, and feed line capacitance to optimize for small gold. Now I'm looking for the proper (175/46 ?) PTFE insulated Litz wire to see how the coil TC may be improved.

My detector has always been the Chance PI but its original design had limitations when seeking small gold. As you mention, a 2 stage amp is much better for avoiding saturation and I have modified my detector to achieve this. The original Chance PI firmware is not capable of sampling below 8-9us so testing of coils and targets at 5us was not possible. I modified Chance with overclocking to allow a 6.98us sample delay...still short of the desired 5us interval. Now Dantech has rewritten the firmware to allow sampling down to about 3-4us if a coil capable of operating at that delay is used. There are other changes to transmit pulse rate and width, and sample window width as well. While Dantech is not interested in small gold targets, I believe his modified firmware will offer improvements to what I have done to make it a better small gold detector. I need to put his new firmware in my Chance to evaluate performance on small gold. Hopefully it will enhance detection of 5 grain and smaller gold targets.

Thanks for putting your post together. It is good to have a copy around as a reference to insure we are addressing all of these design issues.


Regards,

Dan

Dan,

The smallest spiral wrap I can find is this: http://www.usplastic.com/catalog/item.aspx?itemid=34672 It is 1/16" ID, and 1/8" OD with a 1/32" wall. If you find some double wrapped Litz wire that is near 1/16" diameter you can then apply the spiral wrap over about 65 ft to 75 ft (or what ever your coil wire and lead length is). This will be a tedious task but when all else fails you will now have Litz wire in a low dielectric shield that will snugly allow you bundle a coil. Make your coil with one or two additional turn(s) than if the Litz wire did not have the spiral wrap on it because a spread out coil has less inductance. Obtain some .5" ID spiral wrap and apply it over the coil bundle. The spiral wrap will expand so make the spiral wrap length a few inches longer than the coil circumference. Add a second layer and then put the shield over the second spiral wrap layer to minimize coil to shield capacitance. If you use an aluminum shaft on your search coil shaft down to near the plastic attachment to the coil, use low dielectric foam inside the aluminum tube to secure the the twisted coil leads. Make sure that the aluminum tube is grounded to minimize noise.

Here is something to try in your area to see if there is any potential advantage. Find out how close you could get a PI metal detector small enclosure that houses only the TX and RX portion of the metal detector. The audio, power and control portions (larger sections) can be mounted up higher on the shaft for hand control of variable parameters. This allows the shortest lead wires and the least capacitance.

I sent Eric Foster some of the single strand AWG 30 Teflon insulated wire (.024" OD) and Scotch 24 shield when I was researching my fast coil article and he indicated that he could get a 300uH coil to sample down to 5 uS on one of his PI machines. One of the largest contributors of capacitance is the coax cable, so shorter is better or extending the coil leads in a creative way is better. Another capacitance sources is the MOSFET COSS so a series diode between the MOSFET and the coil helps lower this capacitance from the coil circuit's perspective.

There is always a tradeoff... add more power for more detection distance or reduce the delay to get a stronger signal earlier and integrate many thousand samples to extract the desired (narrow window) signal even in a pool of wide band noise. The proof is in physical testing to put theory against reality. It never hurts to try something new.

Here is a concept that I am bouncing around in my head and have not tried yet in metal detectors but have tried and published (Web Search: "Low Impedance Pickup Research") where I use a current transformer (CT) to pickup the metal vibrating strings from a guitar with only a single turn of very heavy wire with a magnet in the center. My thought is if the PI target RX signal could go through a current transformer either pre-wired CT with a mounted primary (one turn) or a toroid where you add the RX signal wire through the opening you could get some low noise passive voltage gain. This would make for a good simulation to study before trying it.

What are the time constants of the typical small gold nuggets in your area that you may seek?

Thanks

Joseph J. Rogowski

[baum7154]

Joe,

My goal is to detect nuggets in the 3-4 grain range and this equates to a 2 or 3 us TC. According to Green a .25 X .25" piece of Al can sidewall has a TC of 1.5us. My Chance PI sees this at about 2.75" using my 4" X 12.5" 3DSS racetrack coil. Along with this capability comes detection of a LOT of trash, lead and aluminum bits.

I completely abandoned coax as a feed due to high capacitance and now only use a continuation of coil wire twisted at a rate of 3 turns per inch. This amounts to about 1pf per inch of feed and there is no solder in the coil.

The Chance PI design always incorporated a fast series diode between the mosfet and the coil to minimize capacitance and it is good practice I think. Dantech has removed this diode from his version because he is not interested high coil speed for his coin/artifact searching.

Regarding PTFE insulated Litz being available, I think it is not, because the process for applying a PTFE jacket requires high heat in excess of the heat required to melt the Litz insulation. So nowI am looking to get the same spacing of a 600 volt PTFE jacket using a similar dielectric material in a 3DSS coil design. Thanks for looking into the spiral wrap availability!

I like the idea of the current transformer for low noise and it bears some looking into.

Regards,

Dan

[bbsailor]

Dan,

Will a round 4" coil detect farther than your 4" X 12.5" 3DSS coil?

The reason I ask is that the location of the gold nugget may be anywhere within your coil's active area. It seems that you may find switching between your coil and a probe might help you pinpoint the location of your target nugget. Here is a tip that will help you determine if a ferrite is a good probe candidate. Scan the ferrite at the lowest detection delay and look for a ferrite with minimal or no response. I have had success using the (ICH) ROD7.5/50 Soft 3C80 ferrite material available at:http://www.surplussales.com/Inductor...s/FerRods.html. I used this ferrite on my modified CS6-PI operating at about 13KPPS and about 7.5us delay. Soft ferrites release their magnetic charge very fast unlike harder ferrites that hold their magnetic charge longer and act as targets.

I used one of these ferrites with a .25" hole opening rubber grommet stretched out on each end of the rod. When installed you will have about 1.6" of rod length left to wind your coil. To get about 300uH I wound about two layers of AWG30 Teflon insulated wire .024" OD and placed the starting wind under the grommet to secure it. I ran a few layers of Teflon plumbers tape on the rod to minimize capacitance between the coil and the ferrite rod. I placed a thin piece of raw wire under the grommet to ground the ferrite to minimize noise.

Test the target detection distance with this single rod design. If this is not enough depth, then try two rods glued end to end to make a rod almost 4" long. As an alternative, you can try three rods bound together in a triangular format with the coil wrapped around this triangular ferrite format. Test each ferrite arrangement for minimal or no detection under your 3DSS coil at your minimum delay.

You can increase the sensitivity by adding more turns until the increased turns cause the delay to be too high for your desired target TC.

A simple DPDT switch will allow you to switch between the coil and the probe. The probe should damp at a slightly higher resistance than a coil.

I hope this helps you find those small nuggets better.

Joseph J. Rogowski

[6666]

Google gives me this
http://forum.ampage.org/forum.php?cmd=vt&tid=5447

[baum7154]

Dan,

Will a round 4" coil detect farther than your 4" X 12.5" 3DSS coil?

The reason I ask is that the location of the gold nugget may be anywhere within your coil's active area. It seems that you may find switching between your coil and a probe might help you pinpoint the location of your target nugget. Here is a tip that will help you determine if a ferrite is a good probe candidate. Scan the ferrite at the lowest detection delay and look for a ferrite with minimal or no response. I have had success using the (ICH) ROD7.5/50 Soft 3C80 ferrite material available at:http://www.surplussales.com/Inductor...s/FerRods.html. I used this ferrite on my modified CS6-PI operating at about 13KPPS and about 7.5us delay. Soft ferrites release their magnetic charge very fast unlike harder ferrites that hold their magnetic charge longer and act as targets.

I used one of these ferrites with a .25" hole opening rubber grommet stretched out on each end of the rod. When installed you will have about 1.6" of rod length left to wind your coil. To get about 300uH I wound about two layers of AWG30 Teflon insulated wire .024" OD and placed the starting wind under the grommet to secure it. I ran a few layers of Teflon plumbers tape on the rod to minimize capacitance between the coil and the ferrite rod. I placed a thin piece of raw wire under the grommet to ground the ferrite to minimize noise.

Test the target detection distance with this single rod design. If this is not enough depth, then try two rods glued end to end to make a rod almost 4" long. As an alternative, you can try three rods bound together in a triangular format with the coil wrapped around this triangular ferrite format. Test each ferrite arrangement for minimal or no detection under your 3DSS coil at your minimum delay.

You can increase the sensitivity by adding more turns until the increased turns cause the delay to be too high for your desired target TC.

A simple DPDT switch will allow you to switch between the coil and the probe. The probe should damp at a slightly higher resistance than a coil.

I hope this helps you find those small nuggets better.

Joseph J. Rogowski

----------------------------------------------------------

Hi Joe, I don't have a 4" round coil to test for the answer to your question. The field of my racetrack coil has a very sharp center blade so this pretty well defines it being on the long coil center.


The Ferrite Probe idea is well developed and would be a nice addition to the detector. In your experience what was the lowest delay the probe has operated at or avoided detection at?


Dan

[bbsailor]

----------------------------------------------------------

Hi Joe, I don't have a 4" round coil to test for the answer to your question. The field of my racetrack coil has a very sharp center blade so this pretty well defines it being on the long coil center.


The Ferrite Probe idea is well developed and would be a nice addition to the detector. In your experience what was the lowest delay the probe has operated at or avoided detection at?


Dan

Dan,

See post #36 on a thread named "Ground Balance Theory" dated 03-06-2013 where Eric Foster alias name "Ferric Toes" talks about detecting down to 5uS. You are already doing many things he recommends. However, at lower delays the ground minerals become more of an issue.

Also see this web link: http://www.sic.rma.ac.be/~pdruyts/pu...yCopyright.pdf

I have only used my recommended ferrite cores down to about 7.5uS. If you get your 3DSS coil down to 5uS then try sensing the recommended ferrites.

The real challenge will be to keep the noise level in the probe low. Shielding the probe is also necessary. Try using Scotch 24 with no shield metal touching on the circumference overlap around the probe. Your other challenge will be the probe leads to keep the noise low without adding too much more capacitance. Search on this forum and Eric Foster's forum about his use of soft ferrites in his probes.

I hope this helps. Keep us updated about your progress.

Thanks

Joseph J. Rogowski