Heavens no! This was a simple intrusion detector, not unlike the devices used to open doors at stores. We're talking low cost here, as in a single 2N3866 oscillator and a 1N34 detector. Ability to discriminate advance targets from receding targets enabled elimination of a lot of false alarms because a target that makes no net progress, e.g. something fluttering in a breeze, is not a threat.
John, if what you see is what I see, the lines disappearing down off the bottom of the page are two channels that will be nearly in quadrature over the frequency range Phil specifies. I haven't checked the pole and zero locations of each of the 4 stages in each channel, but I've no reason to doubt that it works as Phil claims. It's a straightforward approach: cascaded all-pass sections of different center frequencies to get the bandwidth, one channel offset from the other so the difference is nearly 90 degrees over the range of interest. Both channels will have significant phase shift and delay from the input signal, but the difference between them is the matter of interest here.
Where did you get that idea? The small-signal bandwidth is one thing, the large-signal bandwidth is another, and only matters if you need a near rail-to-rail signal output.
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
Don Foreman, your current Email Address does not work?! There was a failure when I tried to send an Email to you. I sent It to Phil because of the second part of my writing there...
Don Foreman, your current Email Address does not work?! There was a failure when I tried to send an Email to you. I sent It to Phil because of the second part of my writing there...
Don Foreman, your current Email Address does not work?! There was a failure when I tried to send an Email to you. I sent It to Phil because of the second part of my writing there...
Lots of things confuse me, but not this one. 8 opamps, 2 degree error, wu/wl=1146 is right out of the tables in chapter 7 of the Williams+Taylor filter book; all that's left is to grind out the normalizations and ponder the component tolerances.
What does confuse me is this: If you design, say a 6-pole iir lowpass straight out of the average dsp filter text, you get a nasty mess with a huge span of multiplier coefficients. But if you design a classic state-variable (integrator based) opamp filter, and simulate it digitally, it's a lot more managable, but I don't see people doing this.
I'm playing with the idea of doing a realtime ac power analyzer, with most of the math in an fpga. I'd need a nice 90 degree digital-data phase shifter, inside an fpga, that covers, say, 20 Hz to 1 KHz, to do everything from diesel to jet engine power systems. If I just call up the Xilinx fir Hilbert synthesizer for, say, 16 bit data, and clock it at some reasonable super-Nyquist rate, I'd get some huge number of shift-multiply-add stages, hundreds of them maybe, even with the clever folding and stuff they do. But you can do this with just 6 or 8 opamps using the allpass thing. Digitally simulating an rc lowpass is just
In message , dated Thu, 31 Aug 2006, John Larkin writes
A friend of mine, who is into these digital things, found the same situation, and used my analogue solution instead. Modest specification, well met, and only four op-amps.
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.