Turbo Gssl help please..

[SSLtech]

Not really... For sidechain use either way is perfectly suitable. Use exactly what you have on the main board for reassurance, -since the circuit at that part is identical.

However, feel free to cut/lift pin 4 if you like, but the slight improvement in distortion is not actually relevant here, because we're not LISTENING to the signal here; we're just looking at how big the ENVELOPE of the signal is at any given time. -A 0.1% change in the signal due to any improvement in distortion ccan by definition only have -at the VERY greatest- a 0.1% increase in amplitude... and since this is a fraction in the order of a hundredth of a dB, the WORST CASE error is teeny-tiny... and utterly swamped by resistor tolerance errors in any case.

If we were LISTENING to the sidechain as a full-merit signal output I'd say yes, go ahead. but otherwise you don't need to worry.

But if you want to push for every last tiny fraction of a dB, just pull the 47Ω resistor which goes to pin 4 if you're using a trimless. -Just be sure to do it on BOTH boards (main and turbo) and then you can sleep 0.1% better at night! :wink:

Keef

PS: on the subject of the C/R/C/R/C board: You found a significant improvement too, huh? -It's not just me then!

I'm in the process of building one as part of a custom summing amp/mastering chain for a client at the moment. I've built a C/R/C/R/C board just out of reassurance, but I intend to test it with the board bypassed. just to get a second set of test data... -I'm starting to think that this too might be a worthwhile upgrade to the GSSL.

If you LISTEN to the noise floor, you should probably find that the "hash" and "Hiss" is largely unchanged, but the 100Hz component (120Hz in the USA etc) is VASTLY diminished.

-If we carry on making little add-on boards at this rate, then before long we'll have more bolt-on upgrades to the GSSL than you can imagine!

:wink:

 

[user 10827]

HI Keith,

Re the C-R-C-R-C

Infact i did not have enough caps for the last two caps so mine is CRCR and was going to add them in the afternoon but decided to try it any way and it still worked!!

My Neutric Analyser was showing lots of 100HZ component which in the end worked out it was adding 15dB or so to my noise floor.

now it is completely silent, im really happy (once i get the side chain working!) that i can use this accross the buss with out fear of ruining a mix due to bad noise floor...

BUT i never did this to my first GSSL, that was quiet straight off..around
-80 dB with no 100hz component.

 

[SSLtech]

See, that's wierdly similar to my experiences:

The first one that I built was the 'ultimate' which drew twice as much current because there were two complete GSSLs running from the same transformer. -it needed no C-R-C-R-Cs and I don't recall there being any 120Hz (in 'American money'! :wink component.

The 'test mule' which I bought as a non-working project from another lab member (then fixed and completed) always had a slight 120Hz ripple in the audio, somewhere down by the noise floor. -Eventually the ripple filter board solved the problem after I'd exhausted all the "usual suspects" (you know, rotating the transformer, alternately linking and seperating chassis and audio grounds, etc). -Likewise, my Neutrik A2 told me that there was noise present, and I could hear it irritatingly on super-quiet passages if I cranked the gain up.

Keith

 

[Harpo]

[quote author="SSLtech"]Not really... For sidechain use either way is perfectly suitable. Use exactly what you have on the main board for reassurance, -since the circuit at that part is identical.

However, feel free to cut/lift pin 4 if you like, but the slight improvement in distortion is not actually relevant here, because we're not LISTENING to the signal here; we're just looking at how big the ENVELOPE of the signal is at any given time. -A 0.1% change in the signal due to any improvement in distortion ccan by definition only have -at the VERY greatest- a 0.1% increase in amplitude... and since this is a fraction in the order of a hundredth of a dB, the WORST CASE error is teeny-tiny... and utterly swamped by resistor tolerance errors in any case.

If we were LISTENING to the sidechain as a full-merit signal output I'd say yes, go ahead. but otherwise you don't need to worry.

But if you want to push for every last tiny fraction of a dB, just pull the 47Ω resistor which goes to pin 4 if you're using a trimless. -Just be sure to do it on BOTH boards (main and turbo) and then you can sleep 0.1% better at night! :wink:

Keef[/quote]

Keith, thanks for your answer. :thumb:
This question came to my mind when comparing schematic and the That notes. Contrary to That DN124, DN137 and 2081 datasheet this 47R resistor connecting to pin4 was mentioned `Connect pin4 only to a high impedance source..´ and in the range between 130k and 680k depending on grade A/B/C (for 15V supply), so a 2081 in the gssl circuit, obviously working well, for better or worse, might behave a little different here. Overriding the pretrimmed 2080 with a 47R contrairy to That's suggested NC for pin4 might lead to a different behaviour than the 2081. I had no idea if this difference would be noticable for my crappy ears and have no different VCA handy to check this myself.

 

[user 10827]

ok im a plum! im a plum! im a plum!



I got my 101 & 104's mixed up, must learn my capacitor codes!

Still at least this thread might be useful to some one in the future.

Thanks Keith for your time, Please forgive me.

Thanks Harpo for the much needed lesson.

:sam: :sam:

 

[EEMO1]

i got some troubles with my turbo...


while it compresses and sounds alright, the threshold for the turbo is 20db lower.

rightside -40db sinewave and the meter is vertical to 12 o'clock. left side i have to boost the signal to -20db to get the vertical.

i think it's exactly 20db...

i tried to lift resistors to check the values, but they seem to be right. maybe i missed something or im tired... (dbx2151 vca on both sidechains...).

any tips are welcome!

 

[SSLtech]

Run through the troubleshooting sequence as described on the previous page. -That should narrow it down and show you where it goes.

Keith

 

[EEMO1]

2k resistor in place of a 20k resistor that connects to the pin 8 (if i remember correctly) of the 74...


works fine now. extreme left and extreme right give the same gain reduction.

i guess i should not solder resistors when tired.

 

[SSLtech]

Great news!

Now after you've had some rest and had a little time to assess the nature of the units character, post some thoughts about it.

-I think you're going to really dig what it does for the unit.

:thumb:

Keith

 

[EEMO1]

heh, sure will!

I handpainted the frontpanel markings aswell... finally. after having a unit for 3 or so years.

my cat helped, I have some good pictures.

 

[Svart]

Before I go tearing my box back apart to troubleshoot it, does anyone know what the problem is if the unit now compresses even with the threshold turned completely up(no compression) BUT still varies in compression with the threshold knob twiddled?

 

[jrfred453]

Check to be sure you didn't lift the 100r? resistor and just lifted the 47K resistor. Check the wiring from the turbo board back to the GSSL board.

 

[Svart]

nope didn't lift the 100r. The voltages on both TL074s are the same per the troubleshoot guide.

 

[SSLtech]

[quote author="Svart"]Before I go tearing my box back apart to troubleshoot it, does anyone know what the problem is if the unit now compresses even with the threshold turned completely up(no compression) BUT still varies in compression with the threshold knob twiddled?[/quote]
Isn't that a very common affliction for the GSSL when you put loud sounds into it? -It shoudl do it rather less in Turbo mode, but every GSSL that I've seen has been a little "eager" to compress, and the threshold control has needed to be set pretty high.

Keith

 

[Svart]

compared with my other 3, it's far worse! A/B'd with another set to the same settings it compresses much much more. If I turn the threshold completely up(no compression on the other units) it's still compressing deeply.

 

[SSLtech]

You have a scope?

Mult a signal into two units and look at/compare/contrast:

1) the AC signal levels going INTO the 47kΩ resistors,
2) the DC levels around the threshold control and subsequent op-amps.

Keith

 

[Svart]

I have scopes and sig-gens here at work but I failed to bring the unit with me today. I'll troubleshoot on monday. I was hoping that someone could tell me something like, "oh yeah you have XXX wires backwards" or something of that nature.

EDIT:

I'm still unsure of how having parallel sidechains can make much difference. I understand the explanation you give for how this works but I'm not sure how it would make a difference other than how the additive/subtractive signals at the 100r node could affect things.

Think of it this way, on the original GSSL, the channels are summed before the detector and any loud signal would cause either channel to compress the same. If you have two detectors which are separated but still sum at the 100r node and the channels VCAs are still driven by a single CV, then wouldn't the effect be exactly the same? One of the sidechain detectors will override the other and still control BOTH VCA channels exactly the same, giving you the center compression that we were trying to get away from.

Maybe I'm dense but I just don't get it..

 

[SSLtech]

well, the only "common" error for that fault which I can call to mind right now, is if you forget to fit the (47kΩ?) resistor in series with the threshold control...

...-you did fit the resistor in series with the threshold control, -didn't you? :wink:

Keef

 

[Svart]

The 47k resistor that's part of the original summing pair? Yep it's there. If you mean a different 47k then no, I didn't and I don't see anything in the documentation that mentions this either!

 

[SSLtech]

It's a resistor which goes at one end of the threshold pot, IIRC...

Look again at the Gyraf page:

http://gyraf.dk/gy_pd/ssl/ssl.pdf

Look at the top end of the threshold pot in the first page... then see how it's mounted off-board in the second page.

If you don't have it, then it's behaving EXACTLY as it should with that resistor omitted.

EDIT:

I'm still unsure of how having parallel sidechains can make much difference. I understand the explanation you give for how this works but I'm not sure how it would make a difference other than how the additive/subtractive signals at the 100r node could affect things.

Think of it this way, on the original GSSL, the channels are summed before the detector and any loud signal would cause either channel to compress the same. If you have two detectors which are separated but still sum at the 100r node and the channels VCAs are still driven by a single CV, then wouldn't the effect be exactly the same? One of the sidechain detectors will override the other and still control BOTH VCA channels exactly the same, giving you the center compression that we were trying to get away from.

Maybe I'm dense but I just don't get it..

You're not "thinking in stereo" perhaps. :wink:

-If the 'loud' signal only happens on left OR right channels, then you're thinking correctly... there will be NO difference.

-But if it happens in the middle, then both detectors give the same result, but instead of summing resistively, they are DIODE-linked, so there's no "addition" as such... just the stronger signal gets through.... in the case of a tie, either one gets through.

Maybe look at it this way: 1V + 1V = 2V when summed additively, (using resistors), whereas 1V + 1V = 1V when summed using diodes. (-Ignoring losses, which are equal in both instances, since they ALWAYS go through a diode.)

Keith