Filter problem

From the description, I expect John means Williams & Taylor,

Cheers

Phil Hobbs

"Coaxial Resonators with Helical Inner Conductor" may help. Fig. 1 shows a shield less than 2" in diameter and about 4" high. (Its height was eyeballed.)

Danke,

Yes, Williams.

Does Zverev have tables of filter values? I generally nab a table from Williams, denormalize and maybe transform it with my program, then Spice it with part values that I can get, or preferably already have.

Or better yet, use the NuHertz passive filter program. It can design LC filters with standard value parts.

Yes. Zverev was a pioneer—the tables were generated directly by a computer, including the typesetting, which was pretty rad for 1968.

He also has really good plots, as well as a unique design, the equiripple delay filter. That one maintains flat group delay well into the stopband, unlike Bessels and suchlike.

Cheers

Phil Hobbs

For my board full of DDS clock generators, I need a new filter form, the Realibad filter. As in, say, 6 dB passband ripple.

I grabbed the one book I have on the subject ("Filters and helical and folded helical resonators" by Vizmuller). I see a few things which suggest that this could work within your constraints:

- The design examples include a 450 MHz resonator (square shield, 1.5 cm shield width, mean helix diameter .99 cm, resonator height 2.39 cm) and a 220 MHz resonator (cylindrical shield, 3.2 cm shield diameter, mean helix diameter 1.76 cm, resonator height 4.25 cm). These examples use some BASIC-language code to perform the calculations (the code is provided).

- The "Proportions of an optimal conventional helical resonator" diagram in Appendix I give the following for an "optimal Q" design. With "d" being the diameter of the helix,

Shield inner diameter D is 1.82d Helix height is 1.5d Space above the helix (tip, to lid of shield) and below the helix (ground point, to bottom of shield) is 0.46d.

Hence, the total height of the shield is almost exactly 2.5 times the helix diameter.

For a wire diameter of "g", the distance between the centers of adjacent turns is "2g" (i.e. the clear space between turns is equal to the wire diameter). This makes it easy to wind... wind a pair of identical wires onto your form, and then remove one. Scaling up from the examples, I'd guess that a 2"-diameter shield, 5" high, is right in the ballpark for the frequencies you are concerned with. 35mm might be a bit tight - you'd need to wind a narrower helix, using thinner-gauge wire to get the necessary number of turns, and this would reduce the Q and the depth of notch.

Am 13.06.25 um 22:54 schrieb john larkin:

There is no Miller effect when there is no voltage gain between grid and anode that could be fed back via Cga. A shorted grid just shorts the feedback. There is also no feedback when input and output frequencies are really different or if the voltage gain of the input device is kapputted by a cascode stage.

Cheers, Gerhard

Sure. Miller needs gain.

How does the Miller effect know what the input frequency is? Does it disappear when there is no input?

A unity gain common cathode stage has a Miller gain of 2, because it’s inverting.

Cheers

Phil Hobbs

But first you have to achieve unity gain from fin on the input side to fin on the output side, which is both unwanted/not needed and hard to do in a mixer stage with a resonant circuit for the mixing product.

BTW another good filter book:

Randall W. Rhea HF Filter Design And Computer Simulation isbn 0-07-052055-0

R.W.R. is the author of Eagleware Genesys which Agilent found necessary to buy in spite of their own Advanced Design System. His oscillator book is also good.