Modifications on my GSSL

[strangeandbouncy]

Aarhus mode is good on drum bus for me. Not always tho'. If you have extreme panned toms for instance, Oxford  can be better.


  Kindest regards,



      ANdyP

 

[SSLtech]

It's been covered a lot before but here's what you should do to ANY home-made GSSL:

Turbo mod. (Oxford/Aarhus switch). This is -quite simply- essential.
C/R/C filter mod. (and don't listen to ANYONE who says that you can get anywhere NEAR the same result by upping the rail capacitance... you just CAN'T. -Talk to ANYONE who's built it and they'll confirm that this COMPLETELY eliminates the inherent power supply hum sensitivity which the GSSL has)
Buy the VERY BEST meter which you can find.

Possibly the sidechain filtering, if that's what you like (I see that you already had that at the beginning)
Possibly Feedback mod, but be aware that this makes the threshold and make-up gain controls interact annoyingly...
Possibly VU meter mod, if you want it to 'feel' different.

Re: what anyone tells you about the regulation using 78/79 series versus 317/337 series regs, the last two units I built were one of each, and there was no perceptible or measurable difference once the C/R/C was fitted. ZERO. -None whatsoever.

Keith

 

[radiance]

Here some mods for loosing some caps in the signal path...

http://www.groupdiy.com/index.php?topic=27.0

http://www.groupdiy.com/index.php?topic=8174.0

Most links are dead but I think most info can be found in the group DIY gmail account (see meta for info ...)

 

[bassculture]

hi all,

didn't have the time do get on my solder iron, but i got all components, just need some time
and i do the one knob mod

by the way, you say (reanimatorstudio) you have more treshold on turbo mode...
i got also.
and i also recently be aware that it is only on one side (the one with the turbo board)
i posted on that on the GSSL help thread.

can you check that on yours ?

thank you

 

[bassculture]

do you mean you makes mistakes ?

or do some values on the turbo are wrong ?

i'm using that 2181LA, and i've done the resistor changes needed.
same changes on gssl & turbo

thx - happy christmas if you celebrate it

edit > i have found the topic with keith procedure... will try that & post results

 

[SSLtech]

To fine-adjust compression threshold characteristics between the two channels in Oxford mode, replace either 15K resistor (doesn't matter which) with a 10K resistor and a 10k trim pot wired as a variable resistor in series. This allows you to adjust the series resistance between 10kΩ (at minimum) and 20kΩ (at maximum).

Start with the resistor at the middle of its travel, switch to Oxford/Turbo mode, connect a tone to ONE input and set to a given amount of compression (somewhere around 8dB GR is ideal) at 10:1 ratio. Move the signal to the OTHER input, then adjust the trim pot until the GR matches between channels.

There you go; -sorted!

Keith

 

[SSLtech]

Oh that...

Well there is... and there isn't. It is program dependant, so you can't compensate/adjust for it.

To prove it, set the unit for 10:1 ration with a fair bit of limiting. take a mono signal and send it to one side only. Now Oxford and Aarhus are the same. Likewise for the other side.

Now we'll go to a Mono signal and send it to BOTH sides, and you'll see that Aarhus mode has ~6dB more compression.

Now we'll feed it two unrelated signals. a 1kHz tone on one side and a different frequency entirely on the other... same level though.

Now you'll see that the Aarhus mode is only ~3dB more 'compressive'.

Now we'll feed it a 1kHz tone on one side and a 90° phase-shifter 1kHz tone on the other... now the Aarhus mode is the SAME as Oxford.

Now we'll feed it a 1kHz tone and an opposing-polarity 1kHz tone on the other... Now Aarhus mode doesn't compress AT ALL, and Oxford mode still compresses the same.

So hopefully you'll see that in MOST cases Aarhus compresses more than Oxford... but NOT all. -Because you can't say for certain what the phase relationship between the two channels will be, it's best to just adjust the threshold to accommodate which mode you prefer for any given task.

Keith

 

[bobschwenkler]

Great thead!

 

[SSLtech]

Nope...

Won't be right for all cases. -NOTHING will.

You can do it if you like, you'll see what I mean.

Keith

 

[spongebob]

i did the feedback-mod...really like it.....thanx mate ;D

can anybody tell me how to archieve ratios like 1:1,2 or 1:1,5  instead 1:4 and 1:10??????????????????? ???

 

[spongebob]

oh thanx...i´ll try tomorrow...now i understand a bit more...

1.about your feedback mod: i just built the sidechain filter and took the source from the output ...works very well....but as i tried to shorten point x to gnd over the 330 k nothing has changed....can you explain that part?

 

[spongebob]

ahh..ok...

anyway ...now my gssl has the behavior i need....i also changed some attack and release resistors...now i can archieve very long attacks...and ultra fast releases...

thats a special way i learned for mastering ...to make compression as unaudible as possible...

and thanxxx...i changed the ratio resistors for a ratio of 1.3:1 ...special as well...works very nice on the hole mix

i changed all op amps to burr brown 2604 and 604 but it still sounds a bit
op-amp-like....dull...dont know how to explain....

i just still did not removed the bypass caps...what do you think is the best way to improve its overall sound????

 

 

[tv]

Perhaps try to add resistors from audio-path opamps to negative rail (aka class-a bias). Cca 5K6 to 3K3 resistors usually work ok.. This usually removes some of the opamp-ness

What were your "best" resistor values for time constant and ratio?

 

[spongebob]

uuhhh...i actually don´t remember for attack ...i think it was ca 600k...(with a 4,7M for balance)..i could measure ca 60ms


for release i used a 47k for ultra slow and ca 80k for slower rate

i can´t tell you what i did for ratio, maybe around 68k ....but i archieved 1,3:1(measured)

my experience is to use a compressor (on the mix) with rates like 1,2:1 is pretty cool....you can get deep in compression with the threshold but things are not squeezed ....i heard some guys use ratios like 1,12!!!!:1

also a trick is to use attack as long as possible and release as fast as it works...this will lower squeezing as well

but with most units this is not possible...so i had to built/mod one

if you watch the manley for instance (often used for mastering).....it works around 1,5:1...but it has a soft knee....so you can get more or less ..depends on the input-pot


 

 

[spongebob]

Perhaps try to add resistors from audio-path opamps to negative rail (aka class-a bias). Cca 5K6 to 3K3 resistors usually work ok.. This usually removes some of the opamp-ness

can you explain that for beginners? f.instance....resistor from pin x to -15v...etc...??????

 

[tv]

I know these bus compression techniques from software. Cant speak for gssl, but my probably favorite is to use short time constants (both attack and release), cca 12dB knee, 4-6dB reduction and paralleling so that the compressed bus has peaks @ -2dBfs and uncompressed @ -20dBfs (with slight variations).

I attached a "retouched" GSSL schem with 3K9 "biasing" resistors in audio path, routed to negative rail. Only did it for one channel, so that the difference shows up clearly - and because I was too lazy to do both.

One thing I must mention: don't use blindly the 3K9 value: experiment with the values and use the ones which sound best in your device.. f.e. start with 6K8 and go down to 3K3...

But, with 553x opamps, there is another option: IMO these chips like these resistors to go to --positive-- rail instead of more commonly used negative rail. Again, experiment. (IMO 553x sound good with 10K or 8K2 going to positive rail.)

ymmv

 

[spongebob]

yes ....parallel bus compressing is important as well.....

ahh now i understand....think i allready heard about
is that describen in datasheets and apps for some chips???

so what you said about 3k3 - 6k8 is for burr browns opa2604 and 604???

i think i can try a trimmer as well???!!! or not? dont have lots of resistors left...



recently i modded my old alesis 3630 comp...i was surprised how better psu-caps(low esr) can improve performance ....did you made this in your ssl?
think i have to open every box here i got/made and change this

didnt  you  say you used the mnats psu?

 

[tv]

I don't have a gssl.. These "biasing" techniques are "universal". The values I mentioned are better tried "in vitro". They aren't neccesarily limited to a particular type of opamp, but certain opamps like some values better than others.

You can safely use a trimmer (but don't SHORT the opamp output pins!) to try and determine the best sounding resistor value - and then buy the required resistors..

A well designed and executed PSU can make a lot of difference (not limited to a gssl). But for gssl-related PSU issues, better ask actual gssl owners.

 

[tv]

IIRC the cap between pin 5 & 8 has to be larger.

47pF (or something between 33pF-68pF) works with these chips (generally).

 

[tv]

How are you going to implement the "transistor" in release circuit?

Are you going to inline it in a config. similar to a "vbe multiplier"?