Metal Detector Signal Processing Attn: Engineers and inventors

SFLOcenEng

Jr. Member
Apr 19, 2008
69
2
Rio Seco
Detector(s) used
Minelab Excalibur II
My question is has anyone ever tried to process the signal from a metal detector to gain an advantage while hunting? By this I mean looking at the signal from the headphones in the frequency domain or doing time/freq. domain analysis. I have been working on gathering some data from test samples and processing them in MatLab. As soon as I get some results I will post them. What I hope to see is some real difference in the signal when various objects are scanned. If anyone has done anything like this please let me know. I'm using a Minelab Excalibur II.
 

OP
OP
SFLOcenEng

SFLOcenEng

Jr. Member
Apr 19, 2008
69
2
Rio Seco
Detector(s) used
Minelab Excalibur II
I don't think that the detectors can tell the difference between say 6061-T6 Al & 5086-H32. I just doubt that they are that sensitive. So the "infinite no of alloys" will not be the main obstacle. Let me know if you know different.
 

radarwill

Sr. Member
Feb 8, 2008
477
11
Central MA
Detector(s) used
Minelab Sovereign GT
Absolutely. Just had my GT hooked up to a National Instruments ADC yesterday.
You will be surprised to "see" the differences in signals a Minelab will produce.
I've made a 0-180 Pic controlled meter for my GT and am developing a "audio cross check" function to try to better my design.
Good Luck
 

99thpercentile

Full Member
Nov 2, 2006
139
99
Evergreen, CO
Detector(s) used
Geonics EM61-MK2, Geophex GEM-3, GapEOD UltraTEM III, Minelabs F3, Foerster MINEX 2FD 4.500
Primary Interest:
All Treasure Hunting
This is a very important research area for landmine detection and unexploded ordnance detection and discrimination. Look at the papers from the SPIE meeting in Orlando every year. DoD is currently funding lots of universities for the modeling and signal processing work (Duke, University of British Columbia, Colorado School of Mines, Dartmouth, MIT). I use Matlab as well as National Instruments hardware and labview for this type of research at work and for my PhD. This was pone of the main topics of discussion at the magnetoc soils workshop I was at in Cranfield, UK a few weeks ago. The main problem is that consumer metal detectors don't provide enough information to do this effectively. Geophysical sensors such as the Geophex GEM-3, in the frequency domain, or the Geonics EM63, in the time domain, are approaching what is necessary. If you are interested in this subject let me know and I can give you a reading list to get started.
 

Ramapirate

Hero Member
Jul 5, 2006
679
21
Charlotte
Detector(s) used
Primary detector is a Garrett AT Pro
Also have a Garrett Ace 250
Primary Interest:
All Treasure Hunting
Yeah, them danged ol' signals and when you're a' processing them and a coming up with figgers and a running them through them things that make 'um do that thing, ta what ya call "inhance" the ding - ping - pong sounds, and then you got something............

eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeyep.

You 'uns is right smart folks....

HH,
Ramapirate
 

Ramapirate

Hero Member
Jul 5, 2006
679
21
Charlotte
Detector(s) used
Primary detector is a Garrett AT Pro
Also have a Garrett Ace 250
Primary Interest:
All Treasure Hunting
I hope you are laughing!!! Everytime I have people telling me how much I know about computers and radio equipment and freqs and mods and stuff(which I don't really know that much, just enough to be dangerous) I read something like this that buries me and makes me feel like I need to get out my blocks and footy pajamas. No chance of me ever getting the big head with you guys around. :wink:

HH,
Ramapirate
 

Ramapirate

Hero Member
Jul 5, 2006
679
21
Charlotte
Detector(s) used
Primary detector is a Garrett AT Pro
Also have a Garrett Ace 250
Primary Interest:
All Treasure Hunting
And trust me Radarwill, I was talking in first person. I'm the dummy in this equation...

HH,
Ramapirate
 

B

BIG61AL

Guest
You'll have to take in the fact that many targets will send a signal to the detector that is false. A large copper coin laying flat will be read as a good target but so can a chunk of aluminum. Jewelry is another issue. While silver and gold will generally give a consistant signal the wide range of alloys used in jewelry production and the overall various shapes they come in will also cause false signals. This is certainly a very difficult undertaking and I wish you much success.
 

OP
OP
SFLOcenEng

SFLOcenEng

Jr. Member
Apr 19, 2008
69
2
Rio Seco
Detector(s) used
Minelab Excalibur II
Dell Winders said:
SFLOcenEng said:
My question is has anyone ever tried to process the signal from a metal detector to gain an advantage while hunting? By this I mean looking at the signal from the headphones in the frequency domain or doing time/freq. domain analysis. I have been working on gathering some data from test samples and processing them in MatLab. As soon as I get some results I will post them. What I hope to see is some real difference in the signal when various objects are scanned. If anyone has done anything like this please let me know. I'm using a Minelab Excalibur II.

Here's a possible start. It's commonly used and conventional. Converting audio frequency to numerical data. Enter the numerical data into Geophysical software to obtain a computerized graphic image. Dell

LINK to Arc-Geo Metal detector imaging systems: http://www.lrlman.com

Dell your confused and this is beyond the scope of your understand!
 

NGE

Silver Member
May 27, 2008
3,506
119
S.E. Michigan
Detector(s) used
Etrac, Explorer XS II, Fisher 1266-X
Primary Interest:
Other
I think I somewhat unnerstan' what ya all is talking about. When I was working at a Ford dealership back in 1984, there was this guy who was a former Radio/TV tech. And he shared with me something he was working on. He had an old metal detector with an oscilloscope hooked up to it, and he showed me how different metals had different wave lengths, but he hadn't figured out how to power it in the field. In the 60' and 70's I was into CB and Ham big time, and knew all about wavelengths pertaining to frequencies, but, my brain is fried and all I can remember is that 9 feet make one wavelength, and a full wavelength worked better for transmitting. I could wrap 18 feet of wire on a 9ft. antenna and have 3 full waves, therefore allowing me to talk further than the standard 9ft. stainless whip antenna. (wrapped at bottom) into a coil.............NGE
 

vayank54

Silver Member
Oct 11, 2009
2,737
20
Northern VA
Detector(s) used
Whites Blue Gray & Tesoro Cibola
Ramapirate said:
I hope you are laughing!!! Everytime I have people telling me how much I know about computers and radio equipment and freqs and mods and stuff(which I don't really know that much, just enough to be dangerous) I read something like this that buries me and makes me feel like I need to get out my blocks and footy pajamas. No chance of me ever getting the big head with you guys around. :wink:

HH,
Ramapirate

Don't worry Ramapirate you aren't alone :BangHead:
 

jb7487

Sr. Member
Apr 16, 2009
354
19
I think the biggest problem with this approach is that you are working at the wrong end of the data. As someone else alluded to, the sound you get out of the headphones has already likely been processed and information has been lost or altered. The digital detectors these days already do a bunch of signal processing and the audio doesn't really "represent" anything about the original signal that was obtained anyway. It is optimized already to give you as much information as you can probably easily interpret with your ears. Trying to squeeze a little bit more out of it at the expense of additional processing power in the form of matlab is like squeezing blood from a turnip. Even if you could do it you wouldn't find any real practical use from doing so.

Your ears are some of the best signal processing components money can't buy. Your detector has hopefully been filled with optimized circuitry that gives your ears the best signal for the money. Trying to augment that combination is likely a lesson in futuility. I'm not saying that you couldn't possibly squeeze a little more info out of your current detector. But at what cost? Do you intend to carry around a computer with you while you detect?

If you want to look at implementing better signal processing, do it in the actual signal processing stage of the detector. I'm currently building a PI detector and I'm working on improving the target detection characteristics using signal processing techniques. Will it be better than what I can buy. Heck no! I don't do this for a living and the tools I'm using aren't up to snuff. It's a spare time project. But it's really fun and satisfying to use a product that works well that you created yourself. Even if it does ultimately cost 3 times more than what I could have bought one for. ;D
 

dazoff

Sr. Member
Aug 7, 2007
321
3
Lower Mainland Vancouver,BC
Detector(s) used
To many to say
I have a guy working on a Whites MXT for me at the moment modding sound.
He is using a principle of guitar tuning to mod the sound to specific minerals and uses an integrator I believe for guitars.

Conductive properties of different metals are very hard to separate in today's VLF detectors as they distinguish metal objects from each other based on the ratio of their inductance to their resistivity based on transmit signal.
I may not understand him fully but as he explained what he changes is the inductance signal adding a circuit just for this purpose.
Dan
 

Top Member Reactions

Users who are viewing this thread

Top