SSL "Talkback" Listen Compressor

[tv]

More dicking:

The R11, R12 etc. (base current compensation) should be all bypassed with a 100nF ceramic chip (multilayer) caps. They should be calulated according to the total resistance (DC) that the negative input sees. So, the R12 should be 6,8kOhm, while f.e. R19 should be 22KOhm (because of the C17), and R17 is OK but all should be bypased with a 100nF multilayer each or else each such unbypassed resistor would inject some noise that will be amplified troughout stages by their noise gains (~~2,2 per stage on a quick glance).

Edit: I'd suggest to add 10kOhm pullup resistors to positive rail to each opamp's output for easier marketing (A-class ultraanalog blah blah). IME these cheap additions remove slight edginess off of (presumably TI's) 553x's.

 

[livingnote]

Ok base current compensation - now you've got me.

What's the idea there - is there any reference you could point me to for it?

With the pullup R, do you mean on each and every opamp in the circuit, or just on the bal O/P? What exactly does it do? I know I have that one included on the micro (built one for myself too but also still haven't tried it), but was kinda perplexed - this seems to be the corner of opamp territory where I need to get in...

oh and ps have no fear of mucking around with that micro - if you blow it up, I'll just stick another one in the oven.

Speaking of blowing things up - I've gotten into the habit of sticking in some Schottky-MELFs on the rails for good measure, that way it gets rather impossible to fricassee the chippedy. Might include those in the next one.

 

[tv]

http://www.national.com/an/AN/AN-31.pdf

"R5" in "inverting summing amplifier"

http://www.google.com/patents?id=SsAaAAAAEBAJ&dq=5554958

"Abstract
An operational amplifier with reduced bias circuit noise. The present invention reduces the noise contribution of the non-inverting input bias circuit. This noise reduction is accomplished by reducing the bandwidth of the bias circuit by bypassing high frequencies to signal ground. This reduces the noise voltage developed across the bias circuit and thereby reduces the noise at the output of the amplifier."


pullups should ideally go on every opamp. This forces the output stage into class-a the audible effect being that it reduces the edginess or nasalness in their sound.

 

[tv]

Putting 100nF poly's in parallel to _every_ elco in signal path (330uF ones) would be good as well. Having a couple of un-bypassed ones in the middle of sort-of defeats the effort of having bypassed input and output...

 

[livingnote]

Haven't run any listening tests yet but...do the bypass caps really make a difference to audio or did someone once just get the idea because he used not-so-good elkos and the problem went away with 100ns? I was reading the one thread here once where there was mention that elkos are fine into the HF bands - but sometimes the ears tell a different story. What's going on here?

 

[keefaz]

Sorry for OT but shouldn't the pullup resistors be connected to negative rail instead of positive in the case of inverting op amp
I mean opamp output should be more negative than positive in inverting mode, no ?

 

[tv]

witch-hunt..

I can hear a difference when a no-name al-elco is bypassed. I can also feel it when I practice over headphones (even shitty ones) - the response is different. But for instrument pre's I think tantalums work best..

IOW, if you plan on super-duper audiophile-grade elcos, the bypassing may become a question of being politically correct .. OTOH, if you plan on baking a SMD ditty, with "standard" elcos, then bypassing may be a Good Thing.

Have you tried some tantalums in audio path yet? I know there's 470uF/10V or 6,3V chip that would fit in the schem (slightly more agressive/brittle than elcos but that can be countered by pulling opamps more into class-A region). Tantalums add color to the sound, and they can be bypased as well.


One word: breadboard (and play with it.)


pullups: ime 553x and nat's 4562 work better with pullup to +rail ymmv

 

[SSLtech]

Sorry for OT but shouldn't the pullup resistors be connected to negative rail instead of positive in the case of inverting op amp
I mean opamp output should be more negative than positive in inverting mode, no ?

No. -To both questions.

or, more accurately "Not important" to the first question and "absolutely not" to the second.

Keith

 

[Greg Norman]

I don't know where this fits in the discussion, but I have been playing around with this board a little.  I found that replacing C11 with a larger cap (10uF and up) made the thing sound a hundred times better for music.  Making the gain for T4 variable is super handy as well (feeds a THAT 1646 output driver).
I tried having a line input (via a THAT 1203) drop in before R25, but the stock threshold of the compressor seems to be higher than I want it to be. 
Am I correct in wanting to increase the size of C4 and C7?
I am thinking of simply padding one of the inputs for a "line" in. 

 

[livingnote]

Not sure about that one - but re C11, that was EXACTLY one thing I was wondering - I presume the SSL peeps decided to put in a Locut there because for Talkback you don't want to have all those low frequencies - same thing on my Neumann's little TB mic.

GK, beautiful Eagle Schemo - even the component names are straight. I'll get back to you on it - right after I get through sending Steinberg a pipe bomb, that is...

 

[geetar_king]

Thanks Lukas, do those outputs to your blend chain work for you?

Greg, good to hear someone has played with C11!  I wanted to do a line in as well. I'm setting up a board with a 1:4 transformer for mic input as well as a few relays to reverse the TXR to 4:1 which hopefully will work for line levels...  We'll see. It would be nice to have the flexibility there. 

The schematic is attached (make sure you're logged in or it won't show up below)
We have it set up with a few jumpers to be able to mess with a bit if we need to, as well as a take off point and unbalanced output to go to a Crush and blend circuit and then out.

Note: The relays are left out on the schematic

 

[geetar_king]

...right after I get through sending Steinberg a pipe bomb, that is...

do explain!

 

[livingnote]

3 Metric Halo 2882s. 1 Neumann Mixer. 1 Cubase 4.

1 Deadline.


This has been screwing me for the past year or so - it's not just Steinberg, but this damn Aggregate Devices
system on the mac has me seriously bothered - darn thing keeps falling out of sync. Probably it's Apple along
with it I should be mad at, but the system doesn't seem to self-reset or anything - means if you run into trouble
or change anything or start another application, you end up having to restart the entire digital rig including
the computer, which is a major PITA. This was probably the second time in my life I actually threw a mouse
at a monitor, and the poor mouse can't even help it.

And I am dearly pissed off with all the marketing b.s. by which it would appear that I'm the only one having
issues like this.

 

[livingnote]

Alright, as far as planning goes - I still have some CnB orders I need to process and while I'm at it, I'll just go ahead and make the final Rev3 board with the bias resistors on before I make more poor folks solder resitors to bent-up chip legs Then I'll be able to spend some more time on this.

 

[tv]

@GK: aren't you having messed up +/- inputs on U1A/B opamps in your schem?

 

[geetar_king]

@GK: aren't you having messed up +/- inputs on U1A/B opamps in your schem?

You are right! Good catch. I originally had 5534's and then i went and replaced those two with the 5532 and forgot to flip it!

Revised version is attached...

 

[livingnote]

Lol, that one gets everyone at some point ))

 

[tv]

C7 - 470nF?

 

[livingnote]

Weird, in the SSL schemo they have parallel inputs with all the same components except one 420µ and one 470n. What did they stick on the board (looks like even they screwed up in their production literature...comforting)

 

[geetar_king]

Yes the production boards actually have 470nF caps there for both talkback amps! Strange there is a discrepancy on the SSL schematic.