THAT 4301/4305 compressor design - turbo sidechain...?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

SWAN808

Well-known member
Joined
Sep 13, 2009
Messages
91
I'm working with someone on a stereo compressor using the THAT 4305...I like the compression action, however it uses a summed sidechain like the gyraf SSL...I can hear the stereo image narrowing slightly...has anyone tried the 'turbo mod' style solution used in the SSL project with a 4305 design and does it solve the problem?
 
There's been a lot written about how to do that properly here and elsewhere, try a deep search, it's probably been a decade. 
 
SWAN808 said:
I'm working with someone on a stereo compressor using the THAT 4305...I like the compression action, however it uses a summed sidechain like the gyraf SSL...I can hear the stereo image narrowing slightly...has anyone tried the 'turbo mod' style solution used in the SSL project with a 4305 design and does it solve the problem?
The reason for mono side chain is so the stereo image does not shift around with different gain changes applied to left and right.

JR 
 
EmRR said:
There's been a lot written about how to do that properly here and elsewhere, try a deep search, it's probably been a decade.

ok thanks I will search more, I didn't find that much on the 4301...

JohnRoberts said:
The reason for mono side chain is so the stereo image does not shift around with different gain changes applied to left and right.
JR

thanks, yes as an alternative to dual mono...however it seems an artifact of summing the L/R is phase cancellation which ultimately results in a slight narrow of the stereo image on compression...(If I understand correctly...)...the philosophy of the 'Turbo mod' being that is reads to L and R signal individually avoiding the phase cancellation...
 
SWAN808 said:
ok thanks I will search more, I didn't find that much on the 4301...

thanks, yes as an alternative to dual mono...however it seems an artifact of summing the L/R is phase cancellation which ultimately results in a slight narrow of the stereo image on compression...(If I understand correctly...)...the philosophy of the 'Turbo mod' being that is reads to L and R signal individually avoiding the phase cancellation...
Perhaps ask who ever said/wrote that...

JR
 
SWAN808 said:
thanks, yes as an alternative to dual mono...however it seems an artifact of summing the L/R is phase cancellation which ultimately results in a slight narrow of the stereo image on compression...(If I understand correctly...)...the philosophy of the 'Turbo mod' being that is reads to L and R signal individually avoiding the phase cancellation...

In my understanding when you sum the signals & then put them into one sidechain it compresses bass frequencies more, (they are the bulk of the energy) because they are summed together & create a larger sidechain signal so those frequenciees get compressed more which kind of scoops out the sounds & makes it sound like it's been split wider & has no power like it's been neutered. 

If you do the turbo mod then each signal creates it's on side chain signal & the bigger one of the 2 is what creates the compression.  Thus you aren't compressing the signal twice in the low end, which makes it sound like it has it's balls intact.
 
Rob Flinn said:
In my understanding when you sum the signals & then put them into one sidechain it compresses bass frequencies more, (they are the bulk of the energy) because they are summed together & create a larger sidechain signal so those frequenciees get compressed more which kind of scoops out the sounds & makes it sound like it's been split wider & has no power like it's been neutered. 

If you do the turbo mod then each signal creates it's on side chain signal & the bigger one of the 2 is what creates the compression.  Thus you aren't compressing the signal twice in the low end, which makes it sound like it has it's balls intact.
I am not aware of mono signals sounding more bass heavy than stereo?

There are many alternate side chain designs (lots of them). You can respond to peaks captured from either channel, while still mono-ing the control voltage fed to the gain elements for a stable stereo image.

JR
 
JohnRoberts said:
I am not aware of mono signals sounding more bass heavy than stereo?

There are many alternate side chain designs (lots of them). You can respond to peaks captured from either channel, while still mono-ing the control voltage fed to the gain elements for a stable stereo image.

JR

I didn't say mono signals were more bass heavy.    What I'm saying is if you sum the stereo & then put it in a single detector there is more bass content than if you put left & right through individual detectors & then use one of them to control both left and right.  Therefore the bass gets compressed more with the forst scenario.
 
Rob Flinn said:
I didn't say mono signals were more bass heavy.    What I'm saying is if you sum the stereo & then put it in a single detector there is more bass content than if you put left & right through individual detectors & then use one of them to control both left and right.  Therefore the bass gets compressed more with the forst scenario.
I do not understand the difference but whatever.....

JR
 
Hi Rob thanks for your input...yes there is for sure an audible difference...I think the result of the sum is 'stronger' representation of what is commonly in the centre (bass)...due to the L/R phase cancellation...

I found a good 4301 thread and it looks like this could be what I'm after credit to Script:

Script said:
No. On the THAT 4301, pins 5 do not connect to pins 15.
For stereo operation:
(1) Cross-connect RMS pins 5 of left and right channel THAT 4301
(2) Cross-connect EC+ pins 15 of left and right channel THAT 4301
(3) Entire sidechain of "Slave channel" can be skipped/switched out of circuitry.

Summing before the RMS detector works too, but it will cancel out out-of-phase signals (i.e., original Gyraf GSSL approach). And if the VCA In pins are not cross-connected, the VCAs will see any slight difference in Threshold and Ratio settings on left and right channel, resulting in image shifts. This holds true for any (VCA-based) compressor.
 
The 'bass-heavy' behaviour of the GSSL is a misleading description.

In Aarhus mod (i.e., summing before detectors) that comp is 'blind' to stereo -- which can be just the right thing depending on program material ;)

The GSSL is rather 'aggressive' when bass in L and R are mono-summed before detector  but not scaled down (i.e., more compression action due to hotter SC signal compared to Oxford approach).
 
Script said:
The 'bass-heavy' behaviour of the GSSL is a misleading description.

In Aarhus mod (i.e., summing before detectors) that comp is 'blind' to stereo -- which can be just the right thing depending on program material ;)

The GSSL is rather 'aggressive' when bass in L and R are mono-summed before detector  but not scaled down (i.e., more compression action due to hotter SC signal compared to Oxford approach).

Thanks Script...in terms of retaining the stereo field as much as possible using stereo compression, is your 4301 guidance above similar to the Oxford approach?
 
My electronics knowledge (still) really sucks big time, and there might be subtleties I'm completely missing out on -- but I'd say 'yes'.

Having said that, the mixed FB / FF sidechain approach of the SSL is kind of unique though.

Maybe also consider mono-summing before detector (switchable) for more 'exciting' compression results, replicating GSSL Aarhus behaviour. ;)
 
SWAN808 said:
I found a good 4301 thread and it looks like this could be what I'm after credit to Script:

THAT Corp design note #116 has all the information needed. Page 5, True RMS power summing.

For proper operation, you need to match both detectors 0dB reference point.

http://www.thatcorp.com/datashts/dn116.pdf
 
elskardio said:
THAT Corp design note #116 has all the information needed. Page 5, True RMS power summing.

For proper operation, you need to match both detectors 0dB reference point.

http://www.thatcorp.com/datashts/dn116.pdf

thanks a lot for this, will look into it...
 
Back
Top