I have been looking around the web for anyone to come up with a "real DSP" design for a metal detector. I have seen some that use the DSP just for audio processing.. Who will be the first with True DSP manipulation of the ground and target signals? True DSP can show signals way below the noise floor of a conventional metal detector, so forget about the analogue domain, it will not deliver the performance of the "True" dsp design.
Announcement
Collapse
No announcement yet.
Real DSP Metal Detector
Collapse
X
-
Originally posted by detectormods View PostI have been looking around the web for anyone to come up with a "real DSP" design for a metal detector. I have seen some that use the DSP just for audio processing.. Who will be the first with True DSP manipulation of the ground and target signals? True DSP can show signals way below the noise floor of a conventional metal detector, so forget about the analogue domain, it will not deliver the performance of the "True" dsp design.
When will you enter the elegant DSP world?
Aziz
-
-
Originally posted by detectormods View PostIf you put your information on the internet then kiss it good bye to the big boys and their patents.
Everyone gets the same chance to develop the ultimate detector DSP software.
You too. Good luck.
Aziz
PS:
I have prepared a big trap for big boys & patent trolls stealing my ideas! They won't be very much amused about it.
Comment
-
Hi woody,
I think the word "Digital Signal Processing" is misinterpreted.
If we go by it's traditional term, then any processing done to a digitized signal can be considered DSP. And you're right, most of the processing appears to be done at the later stages ( audio level).
However, when you say "true" DSP, then I'm inclined to think analytics performed at the signal level in real time....I don't currently know of any portable machines that perform this trick. Part of the reason has to do with the horse power involved, which until recently, has not been found in the current generation of stand alone microprocessor solutions. Laptops have been running in the multi-Gigahertz range for a long time now, but not ARM. ARM processors have only hit the 1.5Gigahertz barrier recently.
For what is worth, I don't think we need "true" DSP in your intended sense of the word. Simple, precise timing captures of fast fet switching will do us just fine.
Take a look at the processors found on the later Minelabs. They're only running at a wimpy 4Mhz(!!!), yet their detectors remain very capable.
In other words, I think "true" DSP is over kill ...
Comment
-
I have been working with DSP in down hole formation evaluation tools since 1997. In fact resistivety tools are like metal detectors in a way. They have transmit and receive antenas (coils) on the drilling collar with different spacings for different depths of investigation. They usually transmit on 500khz and 2mhz. The conductivity of the formation is determined by the phase shift and amplitude of the received signal. Originally the A/D converters were not that fast so the frequency was reduced using a mixer then some filtering and into the A/D and DSP.
A DSP is a power hog. The faster you clock the more current you draw. Some of the newer projects that I have been working on use TI DSP microcontrollers and they seem to work well but the TI 430 family of low power microcontrollers has plenty of processing power for many applications and they are much easier on the batteries. Matlab is a very helpful tool if you plan on working with DSP.
I have also noticed that most of the metal detector schematics I have seen basically have a complete analog circuit all the way up to the end off the filters and they process the final levels from there. Then again I have not seen everything and it is pretty hard to get a look at the newer stuff unless some one traces one and shares it. I would think they are probably using multi layer bds now so even that would be quite a task. The metal detector business is really small potatoes and they are probably limited as to what they can spend on development and still get a good return.
Comment
-
Originally posted by Aziz View Post*LOL*
I have a real DSP metal detector. It is processing the pre-amp output directly.
Aziz
Me too! It's powered by a 400MIPS XMOS XS1 L1, It also process the pre-amp output directly via a ultra low noise 16 bit ADC. I have enough power to take 16 samples during the flyback decay curve.
Comment
Comment