My little Beast (Updated - Panel Time)

[livingnote]

You know this is just giving me so many new ideas and what all you can make it do
I guess it's time to bring this project to a "finished" position...I kinda ended up exploding
the 1RU concept in my head so I'll have to postpone newer things to the next generation
of this box.

For this one, I have basically just the standard design goals before I wrap it up:

1. This comp + Double bargraph with peak hold for output metering.
2. Experiments with the audio chain (departure from standard 5532/THAT I/O).
3. HPF in all its configurable glory for any style user-moddable into what-you-like 4-position thing.
4. CnB as standard implementation the way it's on the table.
5. CV 0-5V I/O to make them gang-uppable so with thing A you can compress thing B
and even hook up a synth LFO output or whatever craziness you come up with.

And then full stop. Rest gets completely new box design or it just won't work ;-P

 

[tomcat]

Thanks 


>how did you do all that!??

Illustrator.

I get the black base material from Bungard here in Germany, and etch in an oversized
bubble tank with my modified A3 scanner doing exposer duty. It's really addictive once
you're into it because you can just haul off and design something and then print to
PCB when you're done (and change your sleep rhythm 180°..."oh yeah cool now let's
build it")

THAT would be a nice new thread for you!

Cooking - PCBs

Seriously!

 

[TheGuitarist]

So whats the plans for future revisions just out of curiosity?

 

[livingnote]

Well probably some major sidechain adulteration experiments...there's so much you can do
between where you get your CV and the point where you stick it back to the VCA. One generally
intersting thing is to fish out the transient and control that one separately, so you have a
normal compressor but do selective transient-smashing or accentuation. Or make the whole
unit do everything backwards and expand...feed forward, feed back...aw god. Lotsa stuff.

But explodes any and all 1RU ideas...

Interesting basis for general thinking is that Scientology's e-meter in Mr. Keller's adaptation
does this (ignore the digital blocks, all they do is count stuff - "AVM" is meter 1 btw):

You have "general body resistance" known as the "tone arm", and through a cap he separates
this into "ticks, falls and rises" and "tone arm movements" by separating the two through
a cap, like:

Human body -> INA Opamp -> 1. Pre Cap, 2. Post Cap -> 2 different meters (which would be our VCA CV input).

1. -> Slow rises and falls (blowup/blowdown)
2. -> Ticks and quick rises and falls

And you get these two scales: One über-sensitive meter indicating the little changes (transients)
and one medium-sensitivity one (Crest), feed those two back to the VCA controllably and it puts you
in "SPL transient designer" territory with something like a quad THAT VCA block and nice I/O module.

Just dreaming around...but what about some major mishmash machine where you can do surgery on all
aspects of transient and signal behaviour with the self-stabilizing SSL principle sidechain thing in it to boot?

The schematic of the VK1 (not the Chiswick ) is so way complicated because it's got 10'000 different
ultra-stabilizer circuits in it to make everything long-term drift-free on zero and temperature stable etc...
just beautiful.

Btw f.i. me not ST or any religious system, I just get my inspiration from everything and anything...as in:

"There's a blaze of light in every word, it doesn't matter which you heard, the holy, or the broken hallelujah"

 

[livingnote]

It almost gives you stupid ideas like this one:

Imagine extracting overall volume level from a signal and then inserting that at the end of a chain
on your stereo where you listen to your music...now you get this thing where instead of iTunes
pissing you off with its constantly varying volume level, you have this watchdog circuit at the
other end of it that makes sure the output of your speakers is held at a constant SPL.

Thing is - would you have to get it to do funky stuff like integrating over a 5 second time bite
so that it doesn't contort things too much, but just a smooth autoadaption for comfort?


Ah yeah ... cooking+etching waaaay cool

 

[livingnote]

Oh next stupid idea:

Like - on that VK1, making C3 switchable...you could maybe get different size transients to extract and feed back?

 

[tv]

Err... why don't you just mod the existing "auto-release" circuit (the two RC time-constants stacked atop of each other)?

F.e., you could make a "three-store" RC circuit, i.e. three RC-constants atop of each other, with caps in 1:10 ratio, f.e. 0,47uF, 4,7uF and 47uF.

This way, you could shape the release curve to react to fast transients on 0,47uF cap, on mid-speed vol. fluctuactions at 4,7uF and "slow integration time" volume changes (aka "replay gain"), "agc"-style with 47uF.

Of course, this is from top of my head, and you would have to figure out the exact cap/resistor values, but the simpler "2-stack" is already used in the ssl-comp variants.. iow this works, just needs to be figured out to "sound-good".

There is a more complex but simpler than "scientologist" method used in the "what" compressor, see the dual time constant circuitry with all those opamp/diode rectifiers and summation:
http://dt.prohosting.com/hacks/what.html
http://dt.prohosting.com/hacks/what1.gif
http://dt.prohosting.com/hacks/joecheep_schem.gif

 

[livingnote]

Ah...you know sometimes I'm happy I do etching because it keeps my feet pretty grounded.

Otherwise you'd probably see my breadbording sidechain x100 and seeing what you can get
the pony to do 

 

[tv]

Did you alter the detector circuit as well? (not only the filters/vca)

If you invert the polarity of the "what compresor" detector dual time-constant circuit, you could probably "mod" it into the existing gssl detector circuit - and use "triple" constant, if you add another "leg". I mean using it to replace the whole time-constants circuit (the diode and the whole circuitry after the point "c" on gyraf schematic to the point "d").

Imo this could be a rather effective and not too complicated method. Of course, speaking from the top of my head.

 

[Igor]

This is most nice looking SSL build I ever saw.

For another idea on crest factor, look at JBL7110 (Urei La 10/11/12).
They did 2 detectors, there's "threshold" shift amp for peak detector,
and sum which can be done easier imho with 2 active diodes.

Pek/avg implementation is not as hard as voltage-controlled attack time, if you ask me

Ah yes, scientology is shit

BTW, very important thing in peak/avg detector is led which changes from red to green
depending on which detector used. With active-diode based summing circuit, some resistors
and dual comparator can do the job.

 

[[silent:arts]]

Aw thanks

Haw cool, checkitout now I can post as "hey I'm Volker"

Just visiting here...hmm you know this might secure another wacky crazy
design phase in my life...who knows. I just love the more I understand it
and get in it, the more ideas crop up...sorry if I don't have like "finally a
product output" but all these ideas just float around and anything new is
interesting. And of course I wanna be sure it rocks before I share anything.

But yeah I really oughta wrap this thing up before we can totally go nuts I
guess...just a standard "that's how I wanted my SSL" would be nice so that
I can say "finished" and do something with "hey wow cool idea".

I wanted to do something with it like a "learning kit", you know? Like the things
I used to play with back as a kid, Kosmos or so, where you actually know that
thing inside and out by the time you're done. But I guess for the more advanced
peeps who just want one I can offer them through my shop at the usual PCB rates
plus a standard "hey I did a lot of work" charge.

Cheers, Lukas

 

[tv]

silent arts, eh?

there may be a little issue with the JBL/UREI circuits, because they have their own log conversion circuits while THAT chips have their own built-in. Perhaps they could be simplified?

Use of a controled current sink in release control is inspirational. Perhaps it could be translated into using a opamp/bjt current sink.

However, for starters, the "dumb" circuit from the "what compressor" could be substituted very quickly and with little effort - just flip the diodes and cap. polarities. If you add the third "time-constant", you will use a single TL074...

 

[TheGuitarist]

Lukas......


This isn't a GSSL anymore....

its like a..... i don't even know!


Lukas' Little Beast.



I like it.

Can't wait.

 

[livingnote]

Yeah - it's getting there ever so slowly

But this one's still quite tame w.r.t. pushing the "envelope", ha ha...

<cuts in from help thread>

...that if you were gonna do an implementation, at the same time just
to add one more set of controls. For a neurotic wants-everything-just-so
guy like me there's always this need that gets in there. But then I thought
well if you're gonna be 2RU anyway, and just need 4 more XLRs basically
to pull the stunt, what are we talking about and back to 2x stereo compressor
it was.

 

[ruckus328]

Dude, this is very nice indeed.  That's some serious thinking right there.  Well done.

 

[livingnote]

Cheers, it's the SSL addict section, eh?

Do you know by any chance what series Grayhills I could go with to replace those Lorlins?

 

[[silent:arts]]

Do you know by any chance what series Grayhills I could go with to replace those Lorlins?

What about Elma Series 01 ?

available at Distrelec

 

[Igor]

http://www.grayhill.com/catalog/Rotary_71.pdf
Digikey imho is the best source for these switches
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=GH7106-ND
for xample;
parametric search for Greyhill 71 series

 

[livingnote]

Ah great, guys thanks

Oh Distrelec...our old stomping ground. Makes me all misty right along with Elpro
(who have the most fun client relations girl you can goof off with while talking about
x1000 prices).

Considering sizes and formats, those Elmas do look really tempting. Grayhills like the
sideways-mount style on the other hand would be fantastic for a sandwich or clamshell-
style design, but tough to bridge across for the LEDs...

But if you had a 2x8 switch you could do something like:

SCF off - HP1 - HP2 - HP3 - Tilt1 - Tilt2 - KEY - Ext. CV

In a matter that would be debatable, you could do KEY as a separate illuminated
pushbutton switch, so that if you're actually using the key in you could still have
access to the hipass filter section for using on the key in signal. Question is, is it
worth it in practice - apart from ducking and making drums pump the bass, what
would you go using the key in for that would warrant sending it through the HPF
at all?

 

[tv]

Hpf on Key:

If you use kick with long decay as a Key signal, with variable HPF you could "design" the "decay" portion of the ducking. But you would need "parametric" HPF in this case (f.e. sallen-key). Perhaps better leave this to the channel that actually sends the Key signal to the compressor?


Ext. CV:

http://www.expert-sleepers.co.uk/silentway.html


also motu has a cv plugin IIRC


perhaps usable for "designer" side-chaining, and for VCA trickery not otherwise practically usable in analog world.
Crossed my mind when I saw the plugins, but dont have any practical experience with it.