SSL "Talkback" Listen Compressor

[livingnote]

More like this:

Or like this:

Any thoughts on which is better Keef? I kinda swing to the second version because there's
more buffering in there and less work for T3, if you're gonna use a 5532 might as well...

Cheers,

Lukas

 

[geetar_king]

Oh okay.  I like the first version.   If you had the second version, and you then cranked the gain at T3 would the blend section be mixing a weak dry signal and a hot, amplified compressed signal?  I dunno. I just thought it would be best to take the dry side for blend after T3..?

Thanks for the schemo edit Lukas

 

[livingnote]

Sure, no prob - did you see what Keith was getting at? If you stick the same signal into the + and - of any opamp, you'll be getting 0 at the output.

On second thought I like version 1 better myself too for exactly the same reason, just occurred to me and then I read your post

 

[geetar_king]

Okay sounds good. As for the 10k resistor before pin 2 on the 5532 does it matter what that resistor is? The original schemo had a 10R there. I just replaced it with a 10k and put 10k on the feedback loop of the 5532 for unitiy but I don't know what the difference would be if I used 2k resistors there for example.    I don't really understand what a series resistor does to the signal. -reduces amplitude?

 

[livingnote]

Yes. It's a voltage divider with respect to the opamp's output. The reason it can be so high is because you don't need current, all you need is voltage for the opamp to "sniff and trace". The 10R is your typical thingie that dangles off every output, in order to limit current should the output opamp's output be shorted to ground - otherwise it would just pound out what it can with its theoretical 0R output impedance and it doesn't like stuff like that when it just got off work

 

[geetar_king]

Okay, right on.  So would I be okay with 10ks there then.  Might have to adjust the resistors around the potted divider though and around U3 eh?  Any way to calculate those or do I have to play with it.  I have no problem setting it up with sockets and switching some resistors around just wondering about the logic there?

 

[livingnote]

I've been through that mill. No need for you to have to go there too.

But if you want to know the way there - here goes:

The potted divider is a divider made up of two resistors - and their ratio determines the output voltage - R1/R2.

[Edit here these are R6/VR1b and R7/VR1b in your schemo]

Depending on the value of R1 (the one that isn't the pot) you'll get different voltage behaviors when you twiddle
the pot. In the setup above, you will get a sharp climb the first quarter turn and a very shallow climb thereafter-
not much feel to that. So here I plotted out some curves for varying R1's:

The higher R1, the flatter the curve, which is yay because you're getting closer to linearity that way. Of course,
the higher the resistance of R1 is, the more impressed the electrons will be by R1 and the more voltage will drop there,
so the less voltage will drop across the pot in turn, which means you need to fudge the OPA's feedback to deliver more bang in order to restore more or less a unity output.

You can save yourself the entire merry-go-round by simply using the R values specified in my schematic for the CnB, they work and imho they feel right.

 

[MartyMart]

Guys, The less/more in the SSL plug-in is going to do the same thing as a crush-&-blend.

Keith

oops !!
Thanks Keith, sorry about that but it did seem to "respond" in that way IIRC - it's been a couple of
years since I had that one working ( no AU version )

Marty.

 

[geetar_king]

You can save yourself the entire merry-go-round by simply using the R values specified in my schematic for the CnB, they work and imho they feel right.

So from your schematic you're saying 39K for "R1" and 10K for "R2" and then 91K on the OPA feedback should work well? Should I set the layout up with sockets still? Probably a good idea

Should I be sticking a capacitor in parallel with the 10k on the feedback loop of the 5532??  Like a 100pF or 22pF or something? I've seen that before but of course I don't know why it's there. Just for stability?

 

[livingnote]

Yah, not a bad idea, it's for RF suppression and I just have one off all the opamps in the internal crush for good measure. You can always just leave them out if you don't want them on the board, but it's cool to have options to mess with.

Those are the R values I meant, and you should be good to go with them, as well as the feedback and input R's to the OPA. That said, it is wise to keep with the little sockets for the proto to be able to solder once and fudge ten times

 

[SSLtech]

Hang on everyone... Before going into the C&B, the make up gain for the FET-compressed path should happen.

So in the first version for example, the inverting input of the op-amp should go to a pot wiper, one end of the pot track should go to the feedback resistor and the other end of the pot track should take the incoming signal. -That way you get plenty of range. -Play with the ratio of pot value to input and feedback resistances to get the desired range.

Don't forget also that the R31:R34 ratio also gives the FET stage some significant makeup, so you might need to adjust things down or up... or play with the gain in the 'dry' path in order to get it to match.

You could just play with it, or -to help match things up- it might help significantly to add some simple GR indication.

Keith

 

[livingnote]

Hey Keith,

would it be a good starting indication for GR to measure the voltage across R30? I assume that that
could be a more or less absolute indication of what's happening in the FET...also (don't know if that's
been adressed before) is C11 not a bit small to be right in the signal path? (I can imagine that for Talk-
back you wouldn't need too much low freq stuff).

I assue TP 1 and 2 would be connected (I think I need to read up here to make sure I'm not asking stuff
twice...)

Anyway, makeup gain - is this what you mean?

Cheers,

Lukas

 

[geetar_king]

I understand the reasoning behind makeup after the compression side of the signal.  Does the setup you're suggesting keith use the 5532 for makeup? ie instead of unity gain, get a little gain outta the 5532 to match?

edit:   lukas- yes TP1 and TP2 are joined together.    I was thinking of the pot being added to the 5532 maybe. Like change the 10k after C16 to a 5K and put a 10K trimpot in place of the 10k resistor on the 5532 feedback arm?

 

[livingnote]

Well...yeah, but that would give you a max makeup gain of x2. I have this hunch that Keith's setup there around T4 can do some real damage...it looks like a x10 amp to begin with, and fudge that and you should get some serious control. Not sure if there's sense in making it controllable though or just trimmable - if there's no threshhold control, what would be the point (except maybe the input level being a threshhold of sorts, exploiting the FET's characteristics - then it would make sense, like 3 knobs input-gain-cnb)...

 

[SSLtech]

You probably don't want to fiddle or shift the equivalent impedance around T4 very much though, since it's involved with the GR element buffering.

No, I'd do it around the 'added' 5532 inverting stage (which RE-inverts the signal... if the signal was inverted on each end of the summing pot, the middle would be a null).

Keith

 

[SSLtech]

Voltage across R30? -I don't think so... the voltage across C14 might be a rather better indication. R30 is part of a super-high impedance -6dB  distortion trim AC arrangement. Any circuitry reading across both ends will incorporate some AC signal component, and possibly run the risk of messsing up the distortion canceling in addition.

The DC component of the modulating voltage signal is basically smoothed across C14. that's where you should read it, I reckon.

Keith

 

[geetar_king]

Well...yeah, but that would give you a max makeup gain of x2.

Yeah for sure...  Decrease from 5k to 1k or increase the pot resistance to get more gain ratio there?

So the setup would have input gain on T3, and if you increase that you will get a hotter clean signal and more GR on the comp signal, so then you have to increase the makeup right to match. Yeah I think you'd have to make it controllable for sure.  Input Gain pot / Makeup Gain pot / CnB pot like you said Lukas.

It would probably be useful to have some sort of GR indicator then, so you can read gain reduction and increase your makeup by about the same.  I guess you could do it by ear and just adjust to taste.  You think GR would be read across C14??  what about the voltage difference between T3 output and the 5532 input, or does that not make any sense.

 

[geetar_king]

Keith, your description makes me imagine this pot placement (below), but I'm not sure if this is what you mean....

 

[geetar_king]

I was thinking of something like this maybe for the 5532 (below):

 

[livingnote]

Awww man Keith, and I was already thinking you meant T4 and that was some kindofa magical mystery arrangement I didn't understand yet *pop*

Now I see what C14 does - Of course, the detector input comes from down there, duh *el forehead slap* - now I dig this circuit. Thanks!

Maybe take 1K's instead of 10Ks and make it a 10 or 20K pot, or take a 1K for initial input and a 5K as feedback R (I see some major fudgery coming our way - better keep them sockets handy...)

The voltage difference between T3 out and 5532 in...lemme think...of course you'd do it with a peak or avg detector but what would you get...well yeah, GR would get measured...but along with it the entire unit's gain. If I get this right, what you'd be seeing on the meter if you measure out to in you'd first get some fixed LEDs that generally give you the difference from in to out, and 1 or 2 that would blink on above that as the output is higher in the input's dips - certainly an interesting approach because you could set your gain to unity by making all the LEDs go out that are constant, except for the ones that keep blinking, which you'd keep on the scale. Funny stuff. But me being a meter nut I'd go with an in meter, an out meter and a GR meter across C14...and there's yet another funny meter idea I have...

But here comes another nifty idea (uh-oh...) - mess around with a timing network in place of C14 à la GSSL and you'd get yourself controllable attack and release times, no?