duffy
Well-Known Member
One fairly easy thing that might be of some use is a "peak detect" (sample and hold) circuit. You send the signal through a diode to a cap on a separate A/D port pin. When a spike comes along, it goes through the diode into the cap and is trapped - the diode blocks it on one side, the port pin effectively presents an open circuit. So the cap holds this peak voltage until the PIC has a chance to read it. Then you briefly turn the port pin from an input to an output, and write a "0" to it - this discharges the cap to 0V. Make it an input again, and you have reset the trap for the next read. Just a couple of cheap parts and an extra A/D input, but this thing can catch fast transients that might otherwise fall between reads.
There's a "quick and dirty" trick I've used with detecting noise in sound-sensing circuits. You place a charge pump (two diodes, two caps and a bleeder resistor) on a separate A/D channel. I found that fricatives and glottal stops would register with this thing before the spectrum analyzer could get its boots on.
Something that concerns me with all this is that the water is going to act like a powerful damper to any high-frequency resonance. What are you trying to detect? Small rocks hitting the bumper? You have a mechanical system with a mass and a spring, but any natural resonance of the mass of the bumper and its main suspension springs is going to be low and require a big hit. The square rod bumper itself might ring on dry land, but that high-frequency resonance is probably damped out of existence when submerged.
When you were working with the PC-based system in the materials lab, did you look for high-frequency signals from the accelerometer while the bumper was submerged? This would be a good place to start - tap it with a hammer while it's underwater, find out what you can detect. If there's any ringing, an "active filter" could be made for that frequency. If there's noise, the charge pump circuit might do the job. If there's a spike that can fall between your read cycles, a peak detect circuit may what you want.
There's a "quick and dirty" trick I've used with detecting noise in sound-sensing circuits. You place a charge pump (two diodes, two caps and a bleeder resistor) on a separate A/D channel. I found that fricatives and glottal stops would register with this thing before the spectrum analyzer could get its boots on.
Something that concerns me with all this is that the water is going to act like a powerful damper to any high-frequency resonance. What are you trying to detect? Small rocks hitting the bumper? You have a mechanical system with a mass and a spring, but any natural resonance of the mass of the bumper and its main suspension springs is going to be low and require a big hit. The square rod bumper itself might ring on dry land, but that high-frequency resonance is probably damped out of existence when submerged.
When you were working with the PC-based system in the materials lab, did you look for high-frequency signals from the accelerometer while the bumper was submerged? This would be a good place to start - tap it with a hammer while it's underwater, find out what you can detect. If there's any ringing, an "active filter" could be made for that frequency. If there's noise, the charge pump circuit might do the job. If there's a spike that can fall between your read cycles, a peak detect circuit may what you want.
Last edited: