atulasthana

I have posted a concept at http://ballz.ababa.net/atul

One of my metal detector designs is at ballz.ababa.net/atul/index.html However, I will still like to build a complex (intelligent) metal detector ;-)

Thanks for the reply. This is not a college project. But could be taken up by college students. Well, the schematic is very simple, the important part is the microcontroller and the chirp algorithms. The schematic depends on the microcontroller used (or the one you are familiar with). Since a large number of functions can be created in software if the microcontroller is powerful, else these functions will have to be done by analog | digital circuits out side the processor. My email id is [email protected]. Please write a mail to me. Regards

Friends, This metal detector is better than PI or VLF | IB type detector. The transmitter will transmit a chirp (linear or expo) and the receiver will amplify the response, process the signal and identify the target. All this needs to be done using a (or more) microcontroller. I feel TI 16 bit processors will be more suited. We need to generate and process atleast 200 pulses (chirps) per second. Technically, the MD should send out linear | exponential chirp signal from 4KHz to about 40KHz. Individual chirp pulses could have different start and stop frequencies. The received response may also be converted to time domain for Human beings, though frequency domain response (along with phase) could be used for target identification. The received response is the sum of MD system response plus ground response plus target response. MD system response is largely from the amplifier, head and Rx pre-amp (and the noise introduced by wave generator, sampling and computational eroors due to finite word length). This can be gauged from operating the MD in air, away from any object. This response can be stored and later used. This will not vary unless the system parts are changed or the head receives a severe knock. Ground response can be gauged from running the MD close to the ground. And with a balanced differential receiving head, will be minimal. It could also be gauged from integrating multiple responses (on the fly and using a window of 1 to 2 seconds) and can again be stored. Target response could then be calculated by subtracting system response and ground response from the received response. The target identification is more of a statistical learning process, and the MD can be trained to identify the target. Any takers...... Atul Asthana