New 'ultimate' SSL buss comp clone ;-)

[Dr nEon]

[quote author="Dr nEon"]

I'm gonna rack this one up now [/quote]

No I'm bloody well not!..I was jumping ahead...still hadn't fine tuned ratio,meter etc...

But now that's done , ratio's are spot on with 118.7 resistors, in my case!
Levels thru L&R channels of USSL are identical , both in bypass , and with circuit engaged but without compression...and ratios of L and R channels are behaving identically ..

I have only one unsolved issue, which I noticed during ratio tests...

OK , so I'm running a tone out and back into protools via rme interface, nice and direct thru the ussl... metering levels before and after ussl, doin' the standard test..(that is..lookin at in and out lvls and doin the math..), with about 8dBs compression on the ussl meter , at 10:1 , well out of the knee..


SO , while I was checking the ratios of each side of the ussl independently(patching thru only one side at a time) , I noticed that , for any given tone setting , my right channel is 0.4dB down compared to the left.

This is a static 0.4dB difference between left and right that doesn't change with compression amount....

However , take the threshold right up out of compression range , and suddenly L and R are equal !..

This difference cannot be seen if both L and R are plugged up!

So , it appears that the L and R are matched shape and level wise , but the actual onset of compression is happening at slightly different places , 0.4dB apart , in my case..

I'm thinking this could be something to do with the "REAL" or "ACTUAL" threshold detection diodes ..

Does this sound plausible as a reason?


Any thoughts on this will be greatly appreciated!


Cheers everyone!


nEon

 

[Sleeper]

The way you describe it, seems like anywhere along the sidechain path, you could be getting slightly more or less amplification. any one of those TLO72-74 amplifier sections.

I'm sure PRR would tell you not to worry about such a small number, which you would only see when compressing less than .4db...

to fix it, MAYBE, the thing to do would be to put a trim on the 50k and 100k gain resistors on the TLO72 sections that go out to CV to SC and CV to VCA
OR maybe put a little trim on the 100R out resistor... same sections.
Do Ya Think?

Sounds like a lot of work though...
Me I'd probably let it slide...
KElly

 

[Dr nEon]

Hey Sleeper , many thanks, as ever ,for those very useful pointers..

I hadn't thought about those gain stages , but studying schemo further , I think i should look around the cv to s/c part that u suggested...I haven't checked for voltages or anything like that yet, so I'll see if i can find the actual spot where this is happening..or otherwise , put in your trim at the cv to s/c o/p ...

Yes,I know some would say does 0.4dB matter , but I am striving to make this as perfect as I can get ,with my limited resources,as it will be my only mixbus compressor ( seeing as my avalon 747 sucks big time!)...and besides, this fine tuning is the most enjoyable aspect of diy , for me!

which you would only see when compressing less than .4db...

I am sure that the compressor is seeing this difference at any compression setting , but it is not "visible" to the end user , because of the summing to the VCAs ... However it is still there , an in a real world situation , an instrument panned hard right would influence the compressor nearly half a db more than the same guy panned to the left... and when i'm setting up a mix trying to get the image right , I know this difference would bug me..


kElly , there is just one thing I'm not clear on :

MAYBE, the thing to do would be to put a trim on the 50k and 100k gain resistors

I think I've got the 100k , but which 50k do you mean? The 50k pots? Or the 56k on pins 1 and 2 of TL072?

Cheers!

nEon

 

[Sleeper]

Hello Doctor.
Fine tune away.

I looked at the schematic again. Yes I meant the 56k (my eyes need trimming)

I want to let you know that this is all guesswork and trial and error on my part.
I'm no SSLTech or Bcarso Or PRR or one of the many other's here who know
this stuff like they invented it.
That being said , if one of the VCAs is going into compression early, this should fine tune the point where it turns on.

Do report back if this works for you.

Best
Kelly.

 

[Dr nEon]

Nice one Sleeper , I hear ya!

I'm goin' hunting for it now , and I'll definitely report back when I've found the right spot to tweak..

:guinness: :guinness:

cheers!

nEon

 

[Dr nEon]

Ok! So the 100r trim cv to sidechain would work!..

however , I wanted to try and find the exact place where the discrepancy was occurring . I went around in circles for a while studying schemo , and checking voltages around the tl072 and tl074 area.
Measuring at the detector diodes ,testing them independantly, it was interesting to see the discrepancy...their conduction points were at slightly different places on the main 'threshold 'knob .So then, in turn , I set the threshold knob to the place where the diode was just hitting the actual threshold, and measured the potential just before it. I did this on both channels , and the voltage readings were identical..
So I left the diodes alone , and started to suspect the TL074 stages , as you suggested, sleeper.
I also looked at the behaviour of the tl072 cv to s/c part , looking at its influence on the potential before the detector diodes , but without signal present , and it was matching perfectly at any settting..

I had spent ages originally on the build carefully matching everything, so although I could fix the discrepancy with the trim on cv to s/c , I was a bit uneasy about upsetting something that was matched nicely in itself..

So I got to thinking about how the only components I couldn't actually match at the time were the ICs themselves...

The level difference in mVs was quite apparent on pin 14 of the TL074s , so before I got any further into it , I just tried swapping the '74s around between boards to see if the problem moved the other way around, and hence I'd know if the discrepancy was within the ic..

Now maybe we're at the point of diy voodoo here now...but with them swapped , the difference was pretty much cut in half!..

Good news , it results in about 0.2 dB detection difference between L&R , which I feel a lot better about...I'm sure that wouldn't bug me!

In fact, I've now realised that SSLtech makes reference to this in his original posts... (man I'd read this whole thread stacks of times too, and hadn't picked up on that!)

So I'm gonna leave it there!... I had expected either the entire half dB difference to move with swapping the 74's , or stay the same...so I don't really know why it is as it is now , but it's good voodoo anyway..


Cheers for all your help, sleeper!


nEon.

 

[Sleeper]

Hi Neon, glad it's working.
my scope pooped out. new one on the way.
can you confirm something with me please.

On the distortion trim procedure DO YOU?? remove one of the input chips when you trim each VCA.
think I've seen this, but couldn't seem to find it again.

oh yeah
Was the test you did to see the level differences on each side of the compressor done in oxford or GSSL mode?
Kelly

 

[SSLtech]

Remove the OUTPUT chip. (5532).

Keith

 

[Sleeper]

Thanks Keith.
It's an awesome unit, thanks a ton for sharing.
My mixbus/mastering channel just went from flat to fluffy.
wish my hair would do the same.
Kelly

 

[Dr nEon]

Hi Sleeper!

Sorry , just caught up with your question now, Glad Keith chipped in...

Mine is done with 2180s ,so I have no vca trim,

Was the test you did to see the level differences on each side of the compressor done in oxford or GSSL mode?

Oxford Mode!

That's all I need...I didn't make this unit switchable because it will reside only on my mixbus , and I didn't want to mess with the existing front panel,which I was happy with..

Mind you...as soon as Keith's turbo pcbs are available to buy ...I'll be adding one to my other gssl , for some lovely switchabilityness..

[/quote]It's an awesome unit

I'll second that !

cheers

nEon.

 

[Sleeper]

Hi Neon, it's done, trimmed and both sides are very close to equal.
once I had the distortion trimmed out... oh yeah... i found the place where I'd read which opamps to pull for the trimming. IN THE BUILD INSTRUCTIONS :roll:
anyhow, once I had the distortion trimmed out I had a look at the levels and I did have a .5 db difference here and there along the compression curves.
it wasn't a uniform difference, instead at 2db (of compression) both sides were equal and at 10db there was a .5db side to side differential and at 20db compression both sides were equal. clearly theres a bit of curve as the tl074 pushes up the gain.

Swapped in a few different tl074 and I found a pair that were equal all the way through the gain curve.

Nice compressor.

:thumb:
Kelly

 

[vertiges]

About the distortion triming procedure :

[quote author="SSLtech"]Remove the OUTPUT chip. (5532).
Keith[/quote]

I used the outputs :--> Fireface 400 :--> RMAA software for triming the 2181.
I did it without removing the 5532's. What are the consequences ? (I do like to understand what's happening and why... :green: )
Is my method is adequate for measuring the distortion ?

And how practically do you do that ? Have you got to put a wire somewhere ? I mean, where do you get the signal and how ? I don't know if I'm clear... How terrible my English is...

What I've noticed is that the trim could make a huge difference for the THD % (difficult to hear but I'haven't made intensive test) And also for the left channel I had to put the trim 30 degrees left from the center position and for the other one, it was almost full clock wise.

Thanks,

eD

 

[Sleeper]

eD,
you did make an ULTIMATE SSL with the parallel sidechains correct
(2 GSSL boards)
you need to trim the VCA's individually so you have to pull one of the 5532's so that your trimming procedure is only monitoring the output from/through one VCA at a time.

2181... aren't they factory trimmed?
you might need to consult GSSL threads and see, but I think with the pre-trimmed THAT chip you actually need to remove the parts involved with the trimming procedure.
If it's the 2180 that's the pre-trimmed part than disregard this part of the post.

Kelly

 

[vertiges]

[quote author="Sleeper"]eD,
you did make an ULTIMATE SSL with the parallel sidechains correct
(2 GSSL boards)
you need to trim the VCA's individually so you have to pull one of the 5532's so that your trimming procedure is only monitoring the output from/through one VCA at a time.
[/quote]

Of course ! You've just made me remember the Ultmate Design.
No need to take off the 5532 in my case. ("normal" GSSL) :wink:

[quote author="Sleeper"]
2181... aren't they factory trimmed?
you might need to consult GSSL threads and see, but I think with the pre-trimmed THAT chip you actually need to remove the parts involved with the trimming procedure.
If it's the 2180 that's the pre-trimmed part than disregard this part of the post. Kelly[/quote]

That's the contrary : the 2180 are trimmed, the 2181 not.

I would like to know if the RMMA software is as reliable as a conventional distortion meter... Actually, I'm not able to find a reason why that wouldn't be... (EDIT: unless you use it with a poor sound card)

Thanks,

eD)))

 

[Sleeper]

I always get the 2181 2180 thing confused.

I'm on a mac so I can only say that I've read people are using RMAA with good result.

Good Luck
Kelly

 

[emx]

So what's this turbo pcb?

 

[rrs]

[quote author="SSLtech"]
Oh yes.. also I increased the 10Ω resistors going to the 78L12 and 79L12 to about 300Ω, to run the regulators cooler. -Looks like it works very well indeed. (Regulators are dropping from 20VDC instead of 29VDC.. I'm using an 18-0-18V transformer.)

Keith[/quote]

Hi Keith.
If I am using a 15-0-15V Toroid so will a 300 ohm resistor still be suitable?

Also re doubbling the 47K resistors before the sidechain will this alter the overall level coming out of the GSSL, will any other adjusment's need to be made?

 

[rrs]

[quote author="SSLtech"]
Oh yes.. also I increased the 10Ω resistors going to the 78L12 and 79L12 to about 300Ω, to run the regulators cooler. -Looks like it works very well indeed. (Regulators are dropping from 20VDC instead of 29VDC.. I'm using an 18-0-18V transformer.)

Keith[/quote]

Hi Keith.
If I am using a 15-0-15V Toroid so will a 300 ohm resistor still be suitable?

Also re doubbling the 47K resistors before the sidechain will this alter the overall level coming out of the GSSL, will any other adjusment's need to be made?

 

[SSLtech]

[quote author="rrs"]If I am using a 15-0-15V Toroid so will a 300 ohm resistor still be suitable?[/quote]


I think it'll be a bit high. -Try using 10Ω for now.

[quote author="rrs"]Also re doubbling the 47K resistors before the sidechain will this alter the overall level coming out of the GSSL, will any other adjusment's need to be made?[/quote]
It won't affect the signal level, only the sensitivity of the threshold control.

Keith

 

[rick carter]

Somewhat of a newbie question but I haven't seen it addressed in this thread - I'd like to use Oxford mode OR GSSL with sidechain (not both together) - do I still need two sidechain pcb's if I'm switching between the two? Doesn't seem like it to me...just a two pole 8pos switch with Oxford on the second pole set to Purushas "off" position and the requisite resistor mods...or am I just a newbie?