[sdiy] equalizer

[cheater cheater]

On 10/12/05, Johannes Öberg <johannes.oberg at gmail.com> wrote:
> In reply to cheater cheater:
>
> >no... What I think I'll build would:
>
> > 1. break up audio into several bands using LP/HP filters
> > 2. set the gain of each of the bands separately (so, as you see, it
> > doesn't necessarily "kill" those bands, it just separates them much
> > more sharply)
>
> Yep, but this is _not_ what I'm _not_ suggesting :-) If I understand you
> correctly, you are talking about a shelving three-band eq with a level
> adjustment of -36 dB and a 24dB/oct band crossover separation (for better
> "cuttiness"). My idea is that the mixer is probably easier to build, and
> better to use, if you have separate systems for removing frequency bands and
> adjusting "sound".
>
> > No, it's because many DJs prefer "kill" over "smooth" EQs :)
> > Also, I'd think 2 filters per band (and tuning two filters per band)
> > would be *more expensive* than one (bandpass) filter per band.
> > And more troublesome to tune and fit in the box.
>
> ...which was exactly what I was saying... However, you wouldn't need "2
> filters per band" with my idea :
>

I thought you were saying that kill EQs are used because they're
cheaper. My misread, my bad :P

> You would have 1 normal eq adjusting 3 bands (bass, treble, mid) over a +-
> [insert some usably small amount, say 5 to 10 dB] range, and this eq could
> be the normal 6 dB/oct separation routine, with fixed crossover frequencies.
>
> Then, you'd have one 24 dB/oct filter with adjustable cutoff, and switchable
> between hi and lowpass (and perhaps bandpass, and notch and whatnot).
>

Well... you may find yourself wanting to "kill" different bands very
frequently when playing dynamic music. Then, a simple filter wouldn't
work that well.

> Hopefully, this would add up to less 6 dB filter blocks than having a 3-band
> eq with 24 dB / oct band separation. And imho, it would be much more usable
> in a real mixing situation, since you could first adjust the track sound
> when you cue up a song, and then when you are actually doing the mix, you
> can temporarily use the hipass to kill the bassdrum, while still keeping the
> eq adjustments intact. Preferably the pre-listen function would let you (if
> you wanted to) pre-listen to the cued track without frequencies cut, so you
> would have an easier beat-match situation, while still having the benefits
> of bassdrum kills.

There's much more to music than just BD :P ;-)

>
> Since you can adjust the cutoff frequency of the kill-filter, you get much
> better control over what you are killing than with the normal fixed-cutoff
> kill-eq setup.
>

Who said anything about fixed crossovers? :-)

> > You need to have precision in your fingers :)
> > Also, BIG knobs help.
>
> Yup, but this doesn't go against what I'm suggesting. Further, if you have a
> shelving-type eq and use it to kill bands, it is difficult to go back to
> exactly the same "sound" adjustment settings you had set up before using
> those same knobs for killing.
>

As I see it, there are two kinds of tracks:
A. tracks you tweak slowly with feeling
B. tracks you tweak fast

A. With slow tweaks, you usually have very much time, enough to get
any good EQ setup you might need. Also, you're not doing anything
frantically, so you'll have time to get a good look at where that
digit was pointing.

B. With fast tracks, you want hands-on control, and this means you
might want to kill a band, then another one, then drop in another one
while killing another one. At that speed it doesn't really matter if
you get around 0.3 dB of the previous setting, dynamics of your
performance are what counts. Besides... trust me, you'll learn to
remember how that thing was set up... it'll be in your fingers 8)
You'll become real precise with time.

Also, buttons for killing bands are nice too. And they're going in my
mixer for sure :-)

> > Not that anyone would hear slight differences on shitty audio setups
> > DJs are forced to play on anyways 8P
>
> Now thats a real point, but on the other hand, we're talking "mixer of our
> dreams" here, right? :-) Anyway, it makes a real difference on mix-tapes,
> which you need to get gigs :-)
>

True!

> > I was thinking of a digitally controlled mixer :)
>
> Hmm, no $200 per channel then, I'm afraid...

uCs are real cheap! c'mon

>  >Faders themselves would of course be THE faders.... P&G. no less.
> > {well, perhaps for prototype, but I won't rest until those are in my
> > box 8) } Though I cringe at the thought of having 6 of these in a
> > mixer... or the cost, rather... heh.
>
> Yup, faders are bloody expensive. I, who am only a bedroom DJ, find myself
> afraid of using the damn things too roughly...
>
> However, wouldn't it be fairly simple to make a VCA style fader so you could
> use any sort of linear mono fader? I guess the usability of that would
> depend on the application, but it would cure most problems with having to
> choose between glitchy faders and not eating for a month.
>

That's how the P&G faders have to be used anyways. The ones used as
crossfaders aren't even sold as audio-taper, I think.

of course, I never ever thought of not going with a VCA fader kind of
deal (either the VCA controlled by the fader directly or through the
uC)
However, those faders are just great.
Typical faders would die with my playing style after a
month...................? And start getting glitchy in the first week
:P
They also give you very very much better control, that's why e.g. all
of them scratch wizards use mixers with P&G faders :)

> > Your proposal of making a mixer together sounds very interesting!
>
> Unfortunately, I don't think I'm skilled enough in electronics to design
> this sort of thing. I'm more the rip-other-people-schematics sort of guy.
>

Same with me.
This will be a learning experience!
You can join in if you want to. I'm sure there'll be something you'd
be able to add to the project :)

> > Tell me what you think of my concept.
>
> If you really want all of that digital stuff, you should probably go
> all-software. Just A/D's, then software, then D/A.

No way!
DSP is no fun. DSP costs a lot. DSP breaks. DSP doesn't have an analog
sound. DSP isn't easy to change on your workbench. And I'm not
interested in learning DSP.

> Digital control over
> analog filters seems to me to be a very cost-ineffective way to do things
> these days.

Why?

> And hacking it together as a VST plugin to be controlled from a
> Behringer rotary MIDI controller seems to be more likely to actually being
> done :-) Especially if you're running Live.
>

No - that's exactly why I want an analog mixer.
With VSTs, even with ASIO2 you still realistically get 10-20 ms of delay.
Needless to say that's way too much for very fast fader tricks. It sucks 8(
BTW, for MIDI controllers, go to www.ucapps.de - to hell with behringer :P

> In fact, you could probably get this mixing functionality without even
> having to write a single line of code! Just download some free VST filters,
> a free virtual sound card manager, and set up your MIDI controller properly.
>

Actually, that's where I would do your "set up the permanent EQ for
the track" part. You can get VST equalizers in much better quality
than you'd be able to build with any sane amount of time and effort :P

> > I could take part of the difficult mathematical aspects of this thing.
> > I'm a 3rd year maths student :)
> > Complex calc is yet to come though :/ 2nd semester of 3rd year, though :)
>
> I don't think neither the analog route or the DSP would require any real
> math actually. It's all readily available on the web, except for the
> electronics know-how, which doesn't seem to be so much about math as about
> intuition :-)
>

Bah 8)

> Good luck
> /J
>
Thanks!

Cheers,
D.          8')