Passive filter panel

[NewYorkDave]

This is something I believe many would find useful for tracking and mixing. Here's the general concept as it stands now:

I'm working on the particulars.

 

[NewYorkDave]

Oh sure! That was more of a surgical tool, though, whereas this is just for general shaping--not for precision search-and-destroy missions. It's a conceptual descendant of the old "sound effects filter" panels.

 

[emrr]

Funny; I just shot response curves for a bunch of different Langevin and Altec passive filters just last week.

I'd be inclined to drop that BRF an octave to get 150 in there. That would likely see more use than the 2400. In fact, IF I WERE DESIGNING IT, I might not even pick full octave jumps for the BRF, and I'd probably have all steps under 1K. Maybe 125 /250 / 350 /500 range, even if maintaining octave wide rejection. What ARE your thoughts on width of BRF?

If this is a PRODUCT, then you have to pick the final values. If it's a kit or DIY offering then it would probably be useful to provide info for recommended and optional frequency choices.

 

[NewYorkDave]

It's a DIY thing, although someone here expressed an interest in offering custom-made inductors for it once the design is finished.

 

[SSLtech]

Less like the "little dipper" (535?) perhaps more like the Pultec HLF(?) filters (...which are pretty mighty things, by the way!!!) with an added band reject.

The little dipper is a wonderful thing, but the killer advantage lies in switching between band-pass and notch-reject... it allows you to "hunt" and then "destroy", with fully variabel frequency, and adjustable bandwitch on the notch... Delicious! -FANTASTIC for film work... not so much at all for music.

Keith

 

[NewYorkDave]

RE: the Pultec filter--Yes indeed, that's the kind of thing I have in mind with this design. But those old HP/LP passive filter sets usually had a slew of cutoff frequencies available--which is cool, but not necessary for most day-to-day studio recording tasks. To avoid the need for many inductors--or a few inductors with a lot of taps--and a shitload of caps and switch positions, I tried to keep it to a handful of frequencies that would be generally useful.

The Little Dipper was 565, or 565T with output transformer.

 

[ioaudio]

i agree that more frequencies at the BRF would be nice.
i think 150 200 250 300 400 and 600 would be useful for recording, especially for tracking drums and akustik guit.
-max

 

[NewYorkDave]

Well, since I promised myself I wouldn't put a lot of time into "Internet giveaways" anymore, I'll present the basic circuit and data so people can go wild with choosing their own frequencies. I should point out that since the BRF isn't super-narrow, there's little benefit to close-spacing the center frequencies.

 

[blandman74]

Is there a reason why you suggest a LPF setting of 20k *and* an "OFF" setting? Or, for that matter, a 25-ohm HPF setting *and* an "OFF" setting?

 

[NewYorkDave]

Response is 3dB down at the cutoff frequencies.

The -3dB point is standard practice for specifying the corner frequency of a LP or HP filter.

 

[skipwave]

Thanks for formulating this Dave.

Things always crystallize whilst looking at your illustrations. Like the puzzle pieces just leap together.

 

[emrr]

[quote author="NewYorkDave"] I should point out that since the BRF isn't super-narrow, there's little benefit to close-spacing the center frequencies.[/quote]

I find in practice that there is benefit to something closer than octave spacing with octave width shapes; at least the 1/2 octave points since the shape usually varies with depth. Your choice of octave centers suggests that sort of width.

I find that 90% of BRF application with recording is below 1 kHz, if not 600 Hz. PA is another story, and that's where it's apparent to me that Mackie designed the 8 buss mixer with PA filtering in mind rather than recording, as their fully adjustable band is the high band with a 500 Hz bottom frequency.

There's a lot of variations running through my mind, not being well versed in the exact circuit differences between different types of BRF.

 

[NewYorkDave]

The Q of the BRF as I've sketched it is about 1. So with a center frequency of, say, 250Hz, the notch is 250Hz wide between -3dB points. I'll see what I get when I rework the L-C values to raise the Q a bit... In my own work, I've found a Q of 1 to 2 useful for mud-busting mid-dips or general taming of offensive bumps. Beyond that, you're getting into the "scalpel" realm and you'd probably be better off with active circuitry.

Like I said, I'll just present the design data when I'm ready and let others take it from there.

 

[blandman74]

[quote author="NewYorkDave"]Response is 3dB down at the cutoff frequencies.

The -3dB point is standard practice for specifying the corner frequency of a LP or HP filter.[/quote]


Sure, but what I mean by asking is that wouldn't the difference be so subtle that it would make more sense to use a (say) 18k or 16k -3dB point? I.e., a 20k LPF would sound the same as "OFF" except that there would be overall loss across the audio spectrum.

 

[NewYorkDave]

Remember, you're not always going for an audible effect. Sometimes you want to attenuate anything that's outside the audible bandwidth (subsonic rumble, ultrasonic hash, etc.).

"20k LPF would sound the same as "OFF" except that there would be overall loss across the audio spectrum."

No, there is no insertion loss in the passband outside of a very small amount from parasitics. This is not a passive equalizer, but a filter--meaning, there is no "boost" function in either the absolute or relative sense.

 

[emrr]

[quote author="NewYorkDave"]Remember, you're not always going for an audible effect. Sometimes you want to attenuate anything that's outside the audible bandwidth (subsonic rumble, ultrasonic hash, etc.).

"20k LPF would sound the same as "OFF" except that there would be overall loss across the audio spectrum."

No, there is no insertion loss in the passband outside of a very small amount from parasitics. This is not a passive equalizer, but a filter--meaning, there is no "boost" function in either the absolute or relative sense.[/quote]

It will have a strong phase component which will affect the sound, and may be useful even if not apparent in regards to frequency change. 20K LPF also makes an AD converter's job easier since it keeps some info out of the converters filters.

 

[blandman74]

Gotcha. All those points make sense. Thanks.

 

[Michael A]

I've got a couple of 2AR variable LC filters that consist of independent low pass and high pass sections. They are mentioned in the Audio Cyclopedia 2nd ed. pg 356. It includes a graph of the bandpass operation for numerous bands under 1kHz. Anyway, I was wondering if you could hook the sections in parallel rather than series and get a band reject function?

Michael

 

[emrr]

[quote author="Michael A"]I've got a couple of 2AR variable LC filters that consist of independent low pass and high pass sections. They are mentioned in the Audio Cyclopedia 2nd ed. pg 356. It includes a graph of the bandpass operation for numerous bands under 1kHz. Anyway, I was wondering if you could hook the sections in parallel rather than series and get a band reject function?

Michael[/quote]

Yes. Several lab type filter sets do exactly that with a switch.

 

[NewYorkDave]

The design of the BRF is proving a bit problematic. The simple series type shown in the original drawing is very sensitive to source impedance. Attenuation varies by several dB when going from a 10 ohm to a 600 ohm source.

Of course, I could always specify that "this filter must be operated from a 600-ohm source impedance" but people have a tendency to ignore such directives. 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.

A bridged-T BRF is much less sensitive to source Z; but with realistic values of DCR for the inductors, it's hard to achieve the full target value of attenuation (20dB) without a lot of dicking around with the resistor values. If someone uses inductors with a different DCR (compared to the design value) it throws the whole thing off. Plus, a bridged-T equalizer requires twice as many switch contacts, inductors and caps.

Hmmm... The original idea (along with a healthy dose of "**** 'em if they can't follow directions") suddenly seems more appealing :wink:

I'll let ya know when I've figured out how I'm gonna approach this.