Passive filter panel

[emrr]

I don't know that attenuation variance is such a big deal since we're not talking about lab equipment. I'd be more concerned about ringing issues near the cut-offs with changing impedance.

 

[NewYorkDave]

Well, if the scale says "-20" and it's actually -15, that could annoy some people :wink:

Ringing comes into play with the LPF and HPF if driven from very low source impedance. But with typical real-world values of source Z, it's not a big problem.

 

[emrr]

[quote author="NewYorkDave"]Well, if the scale says "-20" and it's actually -15, that could annoy some people :wink:
[/quote]

Don't make a scale! Many things don't. Everyone who messes with any kind of T attenuator outside of a loaded 600 ohm environment deals with the same problem.

 

[NewYorkDave]

Oh yeah, you gotta run it into a 600-ohm load. That's a must... both for this filter and for a 600-ohm T-attenuator. But people are used to having some kind of dB scale on an EQ or notch filter, don't ya think? I could just put in a pot with the legend "0" and "full cut" but I'm not sure if that would... um, "cut it." :wink:

 

[emrr]

I dunno, this is all supposed to be done by ear, right? I'd be tempted to label it without being specific, or label it exactly but with a disclaimer. The latter may be unwieldy, but might have good 'lab' vibe.

I think you sent me this one before, which everyone should have a look at concerning loading and ringing efffects with filters:

http://alteclansingunofficial.nlenet.net/publications/techletters/TL_192.pdf

 

[Jim Zuehsow]

[quote author="NewYorkDave"]Oh yeah, you gotta run it into a 600-ohm load. That's a must..:[/quote]

Passive filters were all the rage back when ALL equipment had trannys in and out and you had the impedance nailed down. You could normally patch anything to anything and never have issues. Lots of people built filters similar to what Dave is proposing using modules available from Altec. They were available in both fixed and variable models, and you just daisy chained them together.
Today though, I would probably put a balanced line reciever or opamp at the input with a 600 ohm series resistor into the filter, and a balanced line driver on the output, with a 6oo ohm resistor on it's input terminal.
I still have a pair of Little Dippers in the rack which I used to use when working with people doing film work. Notching out hum, industrial noise, and the ever present noise from un-blimped cameras were the main uses for me. An amazing piece of gear for the time, however, I now use a VST plugin that gives me 5 notches, with the first one at 60 HZ being 60 db down. Talk about surgical!

 

[emrr]

I was thinking this afternoon about using either an input driver amp for fixing the input impedance, or a 500 ohm input rheostat in series for trimmable ringing. The rheostat is nice as an extra effect sometimes.

And an output load resistor on a switch.

 

[clintrubber]

[quote author="NewYorkDave"]It's bad enough that the filter output must be terminated in 600 ohms to work correctly--and working from a 600-ohm source on top of that means an automatic 6dB insertion loss in the passband. That was commonplace in the old "power matched" days but might scandalize today's users.[/quote]
Redesigning for unequal terminations could alleviate that somewhat:

http://en.wikipedia.org/wiki/Bartlett's_bisection_theorem

http://michaelgellis.tripod.com/bartlett.html

I'n not fully sure this could indeed be a free lunch though, but what the trade-off is I couldn't tell right now... should polish up my passive filter stuff again.

I'm also wondering why both URLs I gave don't properly show up in blue....

 

[clintrubber]

[quote author="emrr"] I think you sent me this one before, which everyone should have a look at concerning loading and ringing efffects with filters:

http://alteclansingunofficial.nlenet.net/publications/techletters/TL_192.pdf[/quote]
FWIW, had never realized you could reshape a T-network from 3rd to 2nd order by means of the loading-resistor (case II). Understandable though, the rightmost coil is made ineffective w.r.t. filtering.

 

[NewYorkDave]

[quote author="emrr"]Ior a 500 ohm input rheostat in series for trimmable ringing. The rheostat is nice as an extra effect sometimes.[/quote]

I was thinking of something similar, except it would be stepped and labeled "source impedance" (referring to whatever's plugged into the input). For example, the "600 ohm" position would be a short, the "low Z" position would be, say, 550 ohms and there'd be a couple of steps in between. But making it continuously variable is cool too.

 

[emrr]

Stepped probably makes the most sense. Seems like the 200-390 ohm range is where I end up usually when I try to match specific pieces.

 

[NewYorkDave]

All right, this is back on my drawing board for the moment. Lemme ask you all something: ya think 12dB cut would be adequate for the BRF? I rarely have to use more than that unless something is seriously wrong with the source material. 12dB is a lot easier to realize with real-world inductors and can be done in 2dB steps wtih a 2P6T switch, which isn't hard to source. 20dB in 2dB steps would entail 2P11T (harder to source) and would require very high-Q inductors. (The switch would have to be double-pole for a bridged-T filter section, which I'm strongly inclined to use because it will give the proper value of attenuation as long as one of its ports is terminated).

 

[Joechris]

I think 12 db is fine. How broad will the curve be? One octave?

jo

 

[emrr]

12 sounds adequate.

 

[Guest]

Dave;
it is more than you think of: adjacent reactances will interact

 

[NewYorkDave]

Yes, I am aware of that.