bandpass filter

Well is this sci.physics.design or sci.engineering.design! ;-)

I try to be bilingual. Some of my best friends are EEs, really they are, despite certain nasty habits such as slagging each other off on Usenet. (I put a footnote about the sign convention in one of my earlier posts, and mentioned it in my arm-waving causality discussion.)

The way that the causality condition appears in the frequency domain is very pretty. When you take the inverse FT of H(f) to get h(t), you have to close the contour on a semicircle at infinity. The inverse Fourier kernel exp(j 2 pi f t) becomes exponentially large on the upper half circle at negative times and on the lower half circle at negative times.

To evaluate the integral you need to be able to ignore the infinite semicircle's contribution. Thus you have to enclose the lower half plane at negative times, and switch abruptly to the upper half plane at positive times.

The value of h(t) equals 2*pi*j times the sum of the residues of the poles inside the contour, so requiring that h(t) = 0 for t

Ok, thanks. I got that now.

Ok. I think it's beginning to make some sense. I appreciate your help.

Thanks, miso. I'll continue studying this stuff. I think it's beginning to sink in.

One reason I need to educate myself is that I'm going to attempt measuring current using a PC trace as a shunt. Yes, I know it will change with temperature. I won't know how much accuracy I need until I whip something together to test it. This is where I will need to filter.

I sometimes build current shunts into PC boards, so I can measure FPGA supply currents and such. Sometimes I use traces or power pours that just happen to be there; you have to calibrate both.

Another way to measure FPGA supply current: connect a bench supply to the power rail and *very carefully* crank up its voltage until it takes over from the supply on the board, then just read its current.

Yes. QUITE LOW AUDIO. Damn! It is so-o-o-o much fun getting you riled up ... please have an apoplectic fit so the rest of us can return to civil discussions of engineering >:-} ...Jim Thompson

OK, you did a pll on the 15 or 135 Hz bearing signals. Whoopee. That's downstream of a slew of input RF bandpass filters, mixers, IF strips (more bandpass filters!) and detectors.

So why are you insisting that I not use a bandpass filter ahead of my PLL?

You make no sense any more. You bring up my wife when you get flummoxed, because you think you're being clever, or you think you are hurting me somehow. Wrong and wronger.

Makes sense. I have tried a breadboard with an uncalibrated #30ga bus wire. Of course, I used Kelvin connections. It appears to work very well. I had to use a lot of averaging to get the result steady. I have seen that IIR filters have a slight advantage over averaging. Any easy way to improve what I have is good. It is why I am studying the approaches to digital filtering.

My requirements are not speed dependent as are yours. I use a processor instead.

Thanks, John S

I just use a Fluke DVM, but it sounds as if you may want to embed the measurement into a PCB. That will need some sort of diffamp.

You might consider adding real milliohm range shunt resistors in your power distribution. That avoids calibration and TC problems.

PCB copper resistance/thickness isn't very repeatable.

I'm not insisting you NOT USE... I just don't see the necessity for such a tight filter (Q=50, as you specified... remember my suggestion of a low-Q filter)), since the PLL does all the phase restoration you're fretting over.

It's always hard to discern your actual application, since you are purposefully vague... is clock restoration what you are trying achieve?

Naaaw. Just to aggravate... which obviously is successful >:-} ...Jim Thompson

That statement just clicked in my head. A BPF won't necessarily eliminate all false zero crossings, so go right ahead with your approach. ...Jim Thompson

You insist on being both rude and useless. Your joy at creating aggravation is sick and stupid, especially since it doesn't work. The only thing that's distressing is how many mean, stupid, insecure people there are in the world.

The TACAN pll is probably like your "invention" of the phase detector charge pump. You happened to be around when someone else did it.

Hmmm, no reaction. Maybe I did violate causality and cause the universe to end.

Oh, yes. That is the goal. And the shunt is fed to an Op-Amp with gain set to 100. The shunt is somewhere in the area of 4 mohms.

I have that now. I want to eliminate the more expensive 4 mohm shunt resistors. Sorry if this sounds like Joerg. (No offense, Joerg).

I know. I am actually looking for a curve change. I am sorry that I cannot divulge the particulars as I have a sensitive client.

You are a good source of information and you make me think in other directions. Thanks.

As for invention, the TACAN circuit was just an _application_ of the PLL circuits already patented in my name, as in...

which was the ANALOG-style phase detector used.

You can make snarky remarks all you want, the patent _is_ in my name, but, if you continue posting lies about my competence, you will get sued. Ask your attorney about tortious interference and tell him I've saved every post you've made that qualifies for that charge.

In court I'll simply have Ron Treadway come in and testify about the charge-pump... and add your penalty losses to my retirement fund ;-) ...Jim Thompson

You're a silly twit. Sue me for that, too.

Except that you *have* deserved it for a long time running now. Hell, I even deserve to go off his kill filter compared to you at this point.

And that is saying something.

This scenario is much closer to being a feature of *your* immature legacy, asshole.