Filter Solutions

(No relation)

It looks great, but I wonder how powerful it is: can it design this filter?

Constraints: Source impedance 10k || 1pF Load impedance 450 || 5.1pF Gain: I forget; let's try for "maximal" Type: asymmetrical bandpass HPF response: 2nd order, -3dB at 95MHz LPF response: 5th order, -3dB at 135MHz Two zeroes in HF stopband, min attenuation 80dB Lossy inductors (whatever those values work out to) Chebyshev passband, +/- 1dB

It's a rather complicated description, and I suspect a filter design that very few people are aware /can/ exist. (I built it from a couple building blocks, and a lot of fiddling, naturally.)

And yes, it works. This is with the attached VCO:

(Response plotted against CV, of course.)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 

It can do lowpasses and bandpasses with arbitrarily located elliptical notches, up to 15 nulls, using the Amstutz algorithm.

So, probably, yes.

Fiddling is OK until you get "lost in space", which I do past maybe

3rd order. Or maybe 2nd order.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics

On a sunny day (Tue, 09 May 2017 16:24:34 -0700) it happened John Larkin wrote in :

Nice

Springer released a new book last year that shows how SPICE netlists can be generated for ladder network based filters. Then a quick frequency/phase response plot can be generated.

Expensive.

--
 Thanks, 
    - Win

I use their freeware version sometimes, but it's limited to two poles. Two-pole filters are so squishy that they aren't very sensitive to exact component values anyway, unless you put in a big peak.

You used to be able to get GPIB-controlled electric screwdrivers for computer-controlled filter tuning. They're probably USB now. ;)

I've tuned LC filters up to about 7 poles by hand. High-Q ones can be done using Dishal's method, where you basically tune the resonance of each section separately, or (what I've usually done) resonating each section with a grid-dip meter. The response is generally much more sensitive to the f_0 of each section than to the L/C ratio, so the GDO method works fine. That's also why tuning by hand is hard--it would be much easier if you could tune LC and L/C directly, instead of L and C separately.

You generally tune by return loss because nulls are easy to spot. The key piece of wisdom (taught to me long long ago by a crusty old technician of about 28) is first to get the right number of bumps (one per pole, counting both peaks and valleys) and spread them out over the right bandwidth before tuning for the desired shape.

It's fatally easy to tune up a nice-looking filter that's slightly too narrow. If that happens it's best to start all over again because the true optimum isn't anywhere nearby.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 

It has saved us weeks of work and made better filters than we probably would have come up with. So it was worth it to us several times over.

I'm doing a notch filter now. Williams says to pick a lowpass prototype (which is not trivial), convert to highpass, scale to the actual bandwidth and impedance, then resonate all the Ls and Cs at the notch frequency. Then I'd Spice it to get the final response, maybe play with finite Qs and other parasitics. The Nuhertz code does all that instantly and lets me iterate at the top level.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I can fiddle maybe a 3rd order LC lowpass filter; after that, I get hopelessly tangled. Active filters aren't so bad.

A bandpass made of loosely coupled resonators, like a top-coupled LC or one of those screws-in-a-waveguide thing, can usually be tuned by peaking the resonators individually, in a pass or two.

I'm happy to turn stuff like this over to computers.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

Anyone used, and can comment on, the Active Filter Module? ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 

I emailed the guy about getting a deal on the LC and active filter modules together, and he about bit my head off. Talented but crabby.

I think he has free 30-day evals now.

I use FilterPro and such for active filters. Actives don't interact like LCs, so they are generally easier to design.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

I just put up a mirror of the AADE filter design program that I use:

It's not as refined as the Nuhertz app, but it'll do what most people need and then some. And it's hard to argue with the price.

-- john, KE5FX

On a sunny day (Wed, 10 May 2017 16:50:07 -0700 (PDT)) it happened "John Miles, KE5FX" wrote in :

Thanks, downloaded it. Soon as I get wine running on some PC again I will give it a try.

This page:

... was linked from the site with the older (V4.42) copy of the program. Looks like it's possible to get it running under Linux with a bit of work.

-- john, KE5FX

On a sunny day (Thu, 11 May 2017 01:11:18 -0700 (PDT)) it happened "John Miles, KE5FX" wrote in :

Thank you, wine on this PC crashes the system, I try to avoid it # uptime 12:36:52 up 69 days I have an other one with an older version of wine installed that works OK. Will give it a try.

Ah, yeah, yeah!! Neil Hecht!! R.I.P.

--
 Thanks, 
    - Win

Awesome, thanks. (it seems to run fine on my Win 7 machine. I didn't try making any filters) And hey I've already paid for it cause I have his LC meter.

George

On a sunny day (Thu, 11 May 2017 01:11:18 -0700 (PDT)) it happened "John Miles, KE5FX" wrote in :

Thank you, wine on this PC crashes the system, I try to avoid it # uptime 12:36:52 up 69 days I have an other one with an older version of wine installed that works OK. Will give it a try.

PS I tried, and get exception 0000 or something. It is an old version of wine, maybe 10 years old.. I did try the tricks on the website. Looks like I have to wait for the Linux version ;-), or somebody releasing the source.

I want wine of my latest system anyways... although LTspice on the old PC is nice for simple filters.

mmm maybe it will run in win98, wait...

AHA! on this old PC there was Win98 in hda1. Started it, but remembered one day it became chaotic when a new graphics card was placed (works great under Linux), drive issue. So found an old HP mouse (did not get my wireless mouse either, nor my USB extensions, but do not need those) its one of those round connectors, same keyboard :-) and JIPPEE! filter program installed, and I made a bandpass PI filer with it. Old win98, just kept it there as it had a driver for my Canon scanner... Who would have thought that still worked...

Many thanks for the program,. looks cool, simple and fast :-)

Note SS has some filter generators.

The demo download don't restrict any of the filter design stuff.

It places the filter on the schematic, which you can then just run as per normal. You can even use the generated netlist in LTSpice if you so desire :-)

-- Kevin Aylward

- SuperSpice

You're being diplomatic...

Filsyn

is a very capable design program, but it takes some time to learn how to use it effectively.