twin T circuit

Lots of web sites. Here's one:

Ed

** Groper Alert !

** Got a low drive impedance and high load impedance ??

........ Phil

"Phil Allison"

** For high Q factors - ie a sharp notch use this circuit.

....... Phil

Few people will have the expensive Theade book. Best if you could post the circuit somewhere, as there are any number of possible TwinT arrangements. Would suspect the wide stopband is down to not enough gain from a feedback opamp, leading to a low Q notch. Use a pot' for the feedback resistor. Q's >

say 5 are usually quite critical as to the gain setting resistor value. Fractionally too much gain and the whole lot oscillates.

--
Posted via a free Usenet account from http://www.teranews.com

Thanks for the replies. I am looking for a circuit with active element not passive. The Lm102 looks good but there is not enough tuning. My bandstop extend from 1k to 10Khz at 2Khz i want it to be at -30dB

here is a link to my circuit wi th the results, Its almost there, But I think this is the maximum i can obtain right ?

the %7E is a tilt, works either way :)

ken

Circuit seems massive overkill. Just one stage could give better results than what you're showing. (also need a lot more data points in your sim). Is that overwritten "RU4841" gain resistor on the left really "90" ohms and not 30k?. Why R160 and C90?. Seems no reason for 'em.

--
Posted via a free Usenet account from http://www.teranews.com

"lerameur"

** Groper Idiot Alert !

** Then f****ng say so - you moron. ** Fucking Say What YOU want in the FIRST POST -

YOU ASININE , ASD FUCKED , WOG IDIOT !!!!!!!!!!!!

......... Phil

actually they are pretty usefull, AS for the 90 Ohm resistor, It do not matter to what I chosse, it is again. From what I calculated, it did make any sens, I obtained a negative value

I have joined on the web site the steps that I used to obtain my circuit (in a pdf).

ken

and

The TwinT acts like a series resonant LC circuit and rejects only -one- particular (designed) frequency. Rolloff slopes dependant on designed Q. You will not achieve the 1k-10k bandstop using it. You should be designing something like a third order 10kHz low pass followed by a third order 1kHz high pass.

--
Posted via a free Usenet account from http://www.teranews.com

Would that create a bandstop or bandpass ??

k

(also

ohms

You

followed

Oops, yes, bandpass. (my bad!) Having reread your wants, it looks you're after a helluva bandstop, needing at least 5th order filters plus some kind of (probably inadequate) summation. Maybe yes, that 3 section TwinT is looking the way to go. Fired up the filter prog and poked around a bit. There's lots of arrangements that will do the job nicely but they all use thousands of bits. For minimum components and in line with what you've already built, I looked at the TwinT arrangements. As you calculated, your arrangement needs a negative resistor in the first stage, so that's out. Using inverting biquads gets rid of the negative element but then there's a problem with poor rejection. Further poking around turned up a nice 3rd order Elliptic design that will nicely do the -30dB at 2kHz (1% ripple). Total time steering the prog' was about 15 minutes. Wouldn't have dreamt of doing the same manually!. If of interest, I could put the circuit diagram on A.B.S.E. or email it.

--
Posted via a free Usenet account from http://www.teranews.com

Component tolerance is a killer with Twin-T circuits ... that's most likely why your simulation works and your breadboard does not.

The passives determine the frequency in the Twin-T. Mismatched passives will minimize the notch depth. The last time I built one of these, the math predicted great results, but real world components yielded disappointing results. By analogy, you may be focussing on improving the transmission in your car, when the real problem is that the engine doesn't work.

Ed