Sine generator IC solution?

I read in sci.electronics.design that Spehro Pefhany wrote (in ) about 'Sine generator IC solution?', on Mon, 31 Jan 2005:

Holy city street, Batman! (;-)

uk>) about 'Sine generator IC solution?', on Mon, 31 Jan 2005:

Wrong ! The effective Q of a Wien bridge varies inversely with the frequency . A typical Q at 1kHz with a 4 Mhz bandwidth is 400!

... inversely with frequency? Come on folks, let's put the level- servo loop-gain in there where it belongs. E.g., the "Q" of our fab AoE 2ppm-distortion Wien-bridge oscillator is well over 400. Err, "fussy little high Q and amplitude unstable thing" it may be.

Q = 1/3 implies a lot of phase noise to me. Anybody know for sure?

John

Wrong! The Q of a Wien bridge is 1/3 (I'll accept JW's number). The Q of a Wien bridge _filter_ varies inversely with _feedback_, regardless of frequency. The Q of a Wien bridge _oscillator_ is (by definition) _infinite_, regardless of frequency.

jp

Yes as I show in my reply below to Larkin Q wien bridge = GAIN-BANDWIDTH PRODUCT * Q of the frequency selective network /3 * f

regardless

Yeah , yeah . I did the math ... I can't spend an hour rewriting all my work when people here keep continously calling me an asshole ... The 1/3 those guys keep talking about is related to the 2nd order filter around the opamp and has nothing to do with the actual Q of the bridge WHEN operating as an oscillator. The actual Q is the one I have called Qtotal below AND is frequency dependent. In short .... The oscillators phase stability at w = wo is -2Q/wo ... so the Q of the frequency dependent part of the bridge Qf is -1/2 * wo

but you actually deal with 2 feedback loops so Qtotal = -1/2 * wo *d (Phaseangle)total /dw

if you're a bit off resonance conditions Qtotal/Qf = delta phaseangle f/ delta phaseangle total

The transfer functions for the 2 feedback loops are related by the phase variations as |G(jw)| *sin delta phaseangle f =| G(jw) - K| sin delta phaseangle total for small phi's you substitute sin angle = angle

for oscillation Ao*|G(jw) -K| = 1 also

its elementary to show that for the bridge G(jw) = 1/3

also Ao= GainBW /freq do the substitutions of the above and you get Qtotal = Gain BW *Qf /3*freq.

I read in sci.electronics.design that Tony Williams wrote (in ) about 'Sine generator IC solution?', on Wed, 2 Feb 2005:

Apologies. I didn't recognize the true origin of the text because it doesn't contain any abuse. (;-)

I did that a couple times. It worked well, but the aliasing behavior of switched-cap filters made them not like the square-wave input and resulted in some output distortion. A tiny R-C antialiasing filter at the input (just a cap to ground for the Maxim circuit) can help a lot.

But these days DDS is a whole lot nicer, especially as regards frequency resolution, range, and parts count. Switched-cap filters can be noisy, too.

John

The switched capacitor filters make an application like this virtually digital- no analog involved, no tuning, no fussing with voltages, a simple embedded programming application: Digitally Controlled Sine-Wave Generator-

Cheer up. If there's nothing critical on top, it is actually possible to dig down a through couple of layers to get that buried signal ;-) Otherwise quick-service exist also for buried via / micro via/ multi,multi layer designs ;-P And for the parts a good microscope will make 0402/0.4mm pitch rework much more fun! I never had tools for BGA rework - and agreed: that's rather frustrating!

Have a nice weekend, Anders