Stamp BS2 Microcontroller metal detector

See what I mean Brian? To the uninitiated, the BS2 looks like a solution when in fact it is nothing more than a toy. There's not really enough RAM available to do any kind of look up table because most of that RAM will be taken up by program. I have a few BS2s, dozens of Basic Stamps and hundreds of PIC devices. As an electronic design engineer (now semi-retired) and former Microchip third party consultant, I would advise the young designer to skip over the BS2 when trying to use the device for complex tasks. Would I like to hear more? Consider this. In a world full of treasure hunters and metal detector users, what are the odds that someone else with a background in electronics has not already considered your intellectual ideas?

As to the BS2, RAM issues aside, there is also the concern of processing speed. The BS2 uses an interpretive language of BASIC. He would be past any target by a figurative mile before he got a meaningful analysis. Designers need to consider speed and not just processor clock speed in megahertz. You need native machine code running at processor speeds to get the job done. Doing it all with the BS2 is my objection. Use the 555 or some other device to do the hard slogging scut work of developing the basic frequencies, timing and gating. That is what I suggest. Use the BS2 only when you must and only for the right reason.

If he decides to use the BS2 to drive an LCD display, for example, he should use another along with the external timing and frequency devices to do the detecting and the target comparisons. Remember that a ring has an infinite number of positions relative to the coil in three dimensions. Same for a coin. Target mass and structure need to be considered as well. Just those two parameters put it outside of any look up table by any design standard.

This is a golden opportunity for you Bassman. If you like the PIC device and you like the simplicity of BASIC, consider getting a PIC BASIC Compiler and use a raw PIC. The compiler changes your code to maching language and native code making it considerably faster than the BS2. I like the CCS C Compiler myself. I prefer to code in C and export to machine code. There are two good PIC BASIC Compilers out there. Do a Google search for them and see what you come up with.

Brian, regarding a 4 bit word, it takes way too much BASIC code to process a 4 bit word. Just creating a mask and executing one more instruction to do it will cost about an inch if coil swing over the target. The cost here is measured in milliseconds but they do add up.

Best of luck with your project. To both of you.

Linkback:

You are not allowed to view links.
Please Register or Login

http://www.thunting.com/smf/index.php/topic,6345.msg38551.html#msg38551

Brian,we have to stop agreeing like this. First, I will clarify that I am not discouraging anyone from trying something out but look back to the original question and YOU chill out. I was giving Bassman some sage and sane advice (he asked for it) about the BS2 and somehow you leaped in with some good but irrelevant to the discussion ideas which you thoughtfully posted here.

To Bassman, go ahead and play around but do try and use dedicated components for some of the constants in the circuit. The BS2 just will not do it all and will disappoint you in the end. Try a higher end microcontroller instead. There are much better devices out there.

What we need to do is go to SourceForge and set up a collaborative project where all of the things an ideal metal detector could do would be listed. SourceForge is not private so perhaps another place would be better; something password protected and open by invitation. Then, the ideas you are having could be discussed among the people who can design such things. What you are discussing has been tried and tested, Brian. The basic concept was used in military radar first; a monopulse resolution enhancement receiver which would receive various frequencies and decode the results giving a radar return that could literally show a 3D image of aircraft and even determine the type of paint on the plane.

To do something like that in the frequencies that we discuss for metal detection in the earth would take a lot of time for switching. You know that huge PONG sound you get when you turn on your detector? That's the circuits all being turned on at once. Some detectors blank out that signal so some do not hear it but it still happens as the circuits tune and balance. Imagine doing that and switching between a half dozen or so frequencies to use that as target ID. Your coil would be in the next Zip Code before a response was analyzed. Yes, I mentioned the FFT and you know why. The spectrum of the return (the various harmonics) can be analyzed in as few as 14 machine cycles with the right processor and an FFT subroutine. It is also much easier to do than switching frequencies and analyzing each result. Easier in computing terms.

I maintain that this discussion has gone to the point at which it is impractical to discuss it here. We have moved into the domain of engineering and design; speculative and established. It is well and truly beyond the scope of the original question of "Has anyone tried designing a BS2 detector?" and into the world of very difficult circuitry. We should have mercy on the casual passers by or their brains will explode.

Linkback:

You are not allowed to view links.
Please Register or Login

http://www.thunting.com/smf/index.php/topic,6345.msg38589.html#msg38589

Brian, every metal that enters a magnetic field reacts differently based on mass, molecular content and shape. Believe it or not, shape has the least effect. Molecular content and mass have about equal effect. The easiest way to determine the signature of the object in a magnetic field is to use a Fast Fourrier Transform analysis from the return spectrum after subtracting the original spectrum. That is why I refer to the FFT as the solution. Yes, I brought it up because the original questioner implied that he wanted to determine his target before digging.

If an FFT was the ultimate solution by itself, I would be a happy man but it is not and that is where additional research is necessary. Such factors as two metal objects in proximity, depth, soil condition, soil mineralization and even temperature can effect the result of an FFT analysis. Any one of those tasks is beyond the capability of the BS2 and most other processors as they come from the manufacturer. I have done years and years of research on this subject for my own personal use with an intent to future profits, of course. I have volumes of notebooks and waveform pictures from my spectrum analyzer along with piles of circuitry built and tested. The "pie in the sky" is not going to be easy to reach but the path so far has had some very funny and rewarding moments.

When I suggested SourceForge, it was about the premise behind the site, not the site itself. This forum is a good place to discuss things like this but to be honest, I see most of this discussion getting the <IGNORE> button by a lot of people here because of the complexity. A private area where collaboration and ideas can be exchanged without rancor is the ideal solution. I am looking towards setting up such a place. Bear with me as I am quite busy these days. When it is ready, you are on the top of my list to invite.

You also asked a while back about using BASIC for the Stamp in a PIC16C54 and the answer is no. Parallax used every byte of available ROM inside the PIC to store their interpreter. They use external RAM to execute the program interpretively. If you could find a Basic Stamp compatible compiler, the code could be used and it would be much, much faster. There are PIC Basic compilers out there that are close and you could certainly rewrite the code to be compatible. Using any one of the Flash PIC devices would make your applications scream and you would be able to reprogram it without having to erase it.

Linkback:

You are not allowed to view links.
Please Register or Login

http://www.thunting.com/smf/index.php/topic,6345.msg38693.html#msg38693

You do not need years of research into FFT. It is only a math formula executed on the spectral return of a waveform. It works. That is why metal detectors use it. However, as I mentioned, there are outside influences that render it less than accurate. Less so than just looking at a waveform on an oscilloscope. Which, by the by, when viewed on an oscilloscope is already transformed by an FFT. The true, raw waveform of the return is far different than what you see.

How was lunch? It is not even breakfast time here yet - 03:50 on Wednesday here. I have always been an early riser.

Linkback:

You are not allowed to view links.
Please Register or Login

http://www.thunting.com/smf/index.php/topic,6345.msg38696.html#msg38696

But Brian, it happens to be guys who have the notions that drive the engineers to design stuff. Seriously. The better mouse trap and all that.

The desire to figure out what is in the ground is what drove the target ID revolution in detectors. I wonder sometimes how many coffee cups were emptied during those discussions over the past 40 or more years. And pinpointing? And target depth? And now, true target ID that is not simply a guess.

I want a 4 ounce machine that has a little 3D TV mounted on the handle that shows me a picture which I can then electronically scan around a buried object to see what it is. Shall I order you one too?

Lunch sounds great. Nearly 08:00 here and time for breakfast.

Linkback:

You are not allowed to view links.
Please Register or Login

http://www.thunting.com/smf/index.php/topic,6345.msg38724.html#msg38724