Another free feed

[Kev]

Hi,
When going over well hunted areas looking for nuggets, try to look for spots others pass by. I was working my way up a lead and not having much success, when two large rock masses (schist tors) rose up either side of the lead.

Now every man and his dog had detected up the trail between the rocks but I happened to notice a little ledge about knee height on the side of one of the tors, and it had been capturing a small amount of dark earth that had worked its way down from the top of the tor.

I had to consciously lift up the coil to scan it and 'boing' a signal, now one usually finds a heap of 22 shells around these tors, since its a great place to scan the area from and pick off bunnies. It sounded more solid than a shell casing and it sure was, as the pic shows, it was a nice little pennyweight piece.

So the moral of the story I guess is, imagine you are a nugget, eons ago slowly working your way down the slope, look for spots where you might come to rest, maybe on a rock, which after a lot longer is left high and dry as the earth and softer rock around it is worn away.

Cheers
Kev

PS I had to come back in and edit this post since this is on a PI Technology forum and this post is a bit off topic. So the technology bit is, I was using a compound pulse stream of 20us 60us 20us 60us etc., when I found this piece. In the past I have found 20us 40us 60us 20us 40us 60us etc., very good, much better than a totally synchronous train, but the point is, I believe the former train pattern is less affected by mineralisation than the latter, I don't at this stage know why though?

 

[wyndham]

Kev, Is the pulse stream you mentioned on the GS5 or is this a mod to another GS or a different detector all together. I'm not at all familliar with the controls of the GS? Just flying blind here, Wyndham
PS great little keeper, thought about putting him back and see if he grows up?

 

[Kev]

Hi Wyndham,
I've added a PIC microcontroller to my Goldquest to allow software control of the pulse width and delay timings. I have a selection of four pulse trains at any one time, selectable via 2 GPIO pins, and the delay is variable between 4 and 12us via 1 A to D input.

The coils I use need 50 to 60us to achieve steady state current, so the longer pulses and short delays make her a pretty neat machine.

I keep all the tiddlers. If I threw them back, I'll be long gone before they reached maturity..:sad:

Cheers
Kev.

 

[wyndham]

Kev, I just looked up the pic micro and saw that they have a basic pro language for it. Is this the way to go and, if you don't mind saying, which chip did you use8/16 bit. Did you make your own or a kit w/ a compiler. there's a lot of info on different sites, just trying to get a handle on your project.Wyndham

 

[Kev]

Hi Wyndham,
I used a 12f675 8 bit surface mount 8 pin package running at 20MHz. It's a powerful little chip, and with an instruction cycle length of 200 nano seconds gives adequately precise timings. It runs quiet, and I only turn the A to D on and read the reject pot value between the first and second samples.

I wrote my code in C. It's very easy to learn, and quite complex operations can be knocked together in a very short time. Previously I built one of Mark Stuart's micro controlled PI's ( see the Geotech site ) and altered the PIC code so it sampled a second time to subtract the earth field signal. The gates are all there, you only need to add one transistor and a couple of resistors and cut a trace and you have a good motion PI to experiment with. Alternatively you could modify Carl's Hammer Head to use a PIC and produce a very good experimentation platform.

The compiler I use is an old free version of C2C. See here for later versions
http://www.picant.com/c2c/c.html

My programming hardware is a simple, homemade David Tait design, and the software is IC Prog, software and circuit schematics downloadable from here
http://www.ic-prog.com/index1.htm

Wyndham if you're comfortable with BASIC go with it, as PI's are not that time critical if you're running the PIC above 10MHz.

Cheers Kev.

 

[wyndham]

Kev, I saw that pic info on Carl's site and found the pic interesting. I got one of Carl's HH boards but have not done anything yet. Maybe this summer I'll get my wife's son to help,he's studying electronic design and might be a good summer project for us.
BTW what keeps anyone from a completely micro based PI. I'm not good at the board level understanding, but roughly speaking, and I do mean roughly.
In a computer that has a sound board(or other DSP boards)(either added or on board) couldn't a signal be sent to audio out(tx)to a coil and returned(rx)(audio in) for signal filtering, where the signal is software processed for target ID.
I'm not suggesting that we lug around a computer, but bench testing could be done on coils or PI's.
I guess I enjoy software logic over hardware logic and I'm trying to get to a place I better understand the PI hardware. Wyndham

 

[Kev]

Hi Wyndham,
A PC based system is a good idea, especially if you are using a 3GHz processor, and write the signal processing code in a lower level language, or at the very least compile for speed and not size, which is definitely not a problem these days. There was discussion somewhere a while back about using a Palm PC, but you couldn't just add an A to D PCI card like you can with a desktop, you'd need to use a separate device with maybe IR communication. As you've already said, a sound card gives you the option of digitising an analogue input, so if you can write code to utilise the audio DSP you could soon be on the cutting edge.

Certainly as a test bench development platform it makes real sense, since many applications such as spectrum analysers, oscilloscopes and TDRs are freely available. Digital is the way to go now because we can look at miniscule time slices in macro.

Sounds like a great project to work on with your step son, he'll be able to help you with the hardware, and you can help him with the software. We're on the back foot with step families, and any investment of time like this pays off in the end.

Best wishes
Kev.

 

[Prospector Al]

Hi Wyndham and Kev,
You have both touched on things that I have been working on recently, and I thought I'd share my experiences.

Using a microcontroller as the main component in a metal detector seems like a very good idea, at first blush. Some of the high end devices have many ports and A/D converters and DACs on board.

Adding a power amp for driving the coil and a preamp for the low-level
signals from the receiver coil would make a detector with very few components, and with the added advantage of being reconfigurable at will, merely by changing the program.

A PC-type microprocessor consumes about 10 mA per MHz, so to conserve battery power, a microcontroller running at a lower frequency would be a better option. I've been using PIC18F6520 and I wrote the code in assembly language.

This device has a DAC on board and seven 8-bit ports--more than enough to connect an external A/D converter and to output all the gating pulses, sound and indicators and to fetch data from various switches for user input.

Yet, I abandoned this approach, for the following reason.

For optimal performance, the pulse repetition rate and the gate delay must be controlled very precisely. When the pulses are generated in code, the execution speed of a particular loop may vary. It's very tricky to ensure that interrupts that service various functions don't
interfere with the basic performance of the detector: Generating the eddy currents and sampling the data at a very precise time intervals.

I found it simpler to have a basic system of timing, built from discrete components, for generating the coil pulses and the signal
sampling pulses. The microcontroller is called into action by interrupts, when the data are ready to be accessed and for input from user controls. These actions are not time-critical and can therefore be processed at leisure, between the coil pulses.

In my opinion, a hybrid system is the best way to go, but after implementing the basic functions with discrete components, it becomes
debatable if adding a microcontroller is worth the trouble...

Of course, using a processor becomes a necessity if one wishes to display information on an LCD screen. But I wonder, is that a worth-
while addition? It may look impressive, but does it really add any significant benefit to a detector?

Prospector Al

 

[Kev]

Hi Al,
Thanks for sharing your thoughts. I agree with you about the timing issues. I don't use interrupts or even polling as these can put your timing out by one, two or more instruction cycle lengths. I now use decremental count sequences for timing functions, but I'm also using analogue circuits for signal processing so the demands on the PIC are small. If I were going totally digital, I'd use a network processor model, one PIC for pulse, signal Rx, ground channel Rx timing and another PIC solely for signal processing etc.

I think Wyndham was looking at the PC as a learning tool to utilise it's power to see what the PI guts are like. There's a few audio DSP chips sitting in the draw at work, and I've wondered about utilising them for post integrator processing, they'd really draw out wanted signals from all the aliases, and murmur. I just don't have the time at present, there's more pressing ideas to pursue.

Cheers
Kev.

 

[bbsailor]

Kev, Prosepctor Al and Wyndham,

Here is another approach using some inexpensive analog test equipment available on ebay. Alot of this analog equipment is being dumped in favor of digital equipment, but it still has value for PI tinkering.

Get a pulse generator that had a sync out, delay out, a delay control and variable pulse width control and a 10V output. These are common features on Systron-Donner Pulse Generators (PG). Drive your coil through a MOSFET with the PG output driving the MOSFET. Do not drive the coil directly else you will fry the output transistors in the PG. Build the first PI amplifier using an NE5534 or NE5534A. This way you can play with the gain that gives the lowest noise, best bandwidth and isolate the test equipment from the high voltage flyback spike. Carl from Geotech is offering a prototype citcuit board with just the coil drivers and RX front end that would be perfect to do this.

Obtain from ebay (usually less than $100.00) a device called a lock-in amplifier. This is the equivalent of the back end processing done in a PI machine. Sync the lock-in amplifier off the PG and adjust the delay to window on the lock-in amplifier to exactly where you want to sample the output of the NE5534. The basic trade trade off for extracting very weak signals in the presence of noise is the time length of the sampling window. Sampling windows up to a second can be examined to get a feel for how fast you can sweep the target past the coil (during a test set up). You can sample longer but that is not too realistic for the typical sweep speed of the coil in a PI machine.

You can use off the shelf analog equipment to get a good handle on the operation of a PI machine. Do a web search on the words "lock-in amplifier" or go to the following web site for a quick tutorial on lock-in amplifiers. http://www.signalrecovery.com/lockinex.htm

This is a much faster way to build an experimental PI machine as most of the circuitry is packaged in commercial instruments. The DC output of the lock-in amplifier can feed a Sweep Function generator with a voltage controlled frequency sweep input. This way you can get a rising audio tone when a target is detected, just like on a PI machine.

Studying the operation of a lock-in amplifier will help you understand the inner workings of a PI machine a little better.

I hope this stimulates you to check out the above link and independently search the web. It will be worth it!

bbsailor

 

[Prospector Al]

Hi bbsailor,

You're right about EBay--it's a good source for equipment that is fully functional, but on the verge of obsolescence. (Owing to the digital revolution.) I recently bought a 60 MHz Tektronix scope and
a spectrum analyser on EBay, for a pittance.

Regarding experimentation with a test bench: It certainly has its value, but the final evaluation has to take place in the field--the ground effect is difficult to simulate. I have trays with black sand
and various mixtures thereof, but I have found that the worst ground is not necessarily the one with the highest magnetic susceptibility.

The magnetic viscosity effect is quite unpredictable and the most difficult one to eliminate. The Customs folks are absolutely paranoid about importing soil samples from Australia...

I may have to find someone in OZ who's willing to test a detector for me--I don't want to make the 29-hour trip again, until I'm sure the detector is going to work well under the worst conditions imaginable...

As Kev suggests, the basic timing is best implemented with counters.
I'm toying with a variation on the theme: Preloading the counter from
a memory chip, instead of having it hard-wired. The "1-wire" serial
memory works quite well. This approach also makes it possible to adjust the gate delay for each coil size and load the value automatically on power-up.

Fun and Games...

Prospector Al

 

[Wirechief]

Hi Kev, Al, BB, and of course Wyndam! I have had thoughts of interfacing my PSP to a detector just for the fun of it. But as Al said in reality these add ons really don't do much for the performance of a detector. I prefer to just listen to what the audio is telling me. But gadgetry is fun to work with and I like gadgets. A couple of days ago I found a solid copper and turquoise Indian style bracelet at work. I was using the Tesoro Golden uMax with tone ID. After 3 pulltabs and a full 12 inches I pulled this copper bracelet out of the hole. All I did was interpret the tones from the Golden. Ok I know this is an Eric Foster PI forum but I just wanted to show if you know the detector your using then you will do well. So how are you guys doing lately? I hope y'all keep posting stuff like this because it is very interesting to me and keeps the forum lively. Ok CU guys later and God Bless.

John Tomlinson, CET


GOD BLESS OUR TROOPS ALL AROUND THIS WORLD, THEY NEED IT!

 

[Kev]

n/t

 

[wyndham]

Yea I agree, got to keep the grey matter from harden up like cement.
BTW I'm sure some of you have seen the TV show Connection where the host James Burke traces back thru history to find the connections between say how the pull tab was invented and finds it started when someone a 1000 yrs ago invented a type of corn mill(not real example)
It would be interesting to trace back Eric's PI to some event 500-1000yrs ago.... or what will come from the PI 500 yrs from now.
Wyndham

 

[Wirechief]

Hi Wyndham, good to hear from ya and I hope your close to a breakthrough in these PI instruments haha! Well you know Wyndham I have been watching and posting at times on various forums and I have never seen so much nit picking over these detectors. I'm refering to VLF type detectors. I like gadgets and blinky lights too but either a fellow is serious about getting dirt under the fingernails and being down on those knobby knees or he is just dressed for church. I am pretty fed up with a some of these forums being dominated by a chosen few and I am going to disappear from them right now. But I am for sure going to stay with this one because there is a bunch of great people here. It is very doubtful I will ever be able to add anything to this forum because all of the guys here are much above me. I just hope to gain some knowledge and hopefully mostly wisdom. Ok I will stand down now and Eric I hope I can eventually get one of your new instruments to put to use in a goldfield someday. Thanks and God Bless all.

John Tomlinson, CET
John's Detectors

 

[wyndham]

I understand the knobby knee thing, and the back that don't bend right, so what we need is a robot remote controlled PI/VLF/IB/ect detector/digger/retriever go getter or.. a time machine to take 20-30 yrs off our hides. Wyndham

 

[Wirechief]

n/t