nice demonstration of how to go about developing a filter from a set of response criteria. I especially liked your explanation of how to get the initial coefficient guess into the right ball park on page 2.
Running a company that routinely takes on analogue design challenges including both electronics based solutions and software based equivalents on the other side of the ADC I particularly appreciate seeing examples like this.
I'm looking forward to your next column on doing the same coefficent fit and implementation in the digital domain.
As a suggestion, some of your readers might appreciate the spreadsheets being available for download. I would. No matter how many times I have solved a particular design challenge I find I can always learn something new from studying how another problem solver approaches the problem.
As for a guess at the sensor type. It is a temporal band pass response so it isn't an absolute value reading device like temperature, humidity, radiation, gas detector, accelerometer of the silicon or mechanical stain variety, accelerometer of the capacitive type, distance measurement or light metering.
Some piezo type sensors can behave like this when they aren't in a charge amplifier configuration so it could be a vibration sensor, shock sensor or an accelerometer for delta measurement only but the upper frequency rolloff seems to be at too low a frequency for this.
Siesmic detector is another guess. The roll off is at a bit too low a frequency for this so some normalisation at the upper end would be helpful.
A flow measurement device is another option. Not the spinning wheel variety but maybe a doppler shift or thermal equilibrium version. But again the time domain respose is very low for this application if it was doppler shift based.
I was project leader for the LASER based version of the VESDA Very Early Smoke Detecting Apparatus and a number of the competitor products worked on pollution measurement with time constants like this but that was on the other side of the software metrics system.
Some PIR based personal presence detection systems also have similar filters to take out the background IR and only show the changes but again that was part of the software discrimination and not necessarily the sensor itself.
Getting warm?
Thanks again for a great article on analogue design priciples and solving for a filter response requirement. Keep them coming.
