Purusha
Well-known member
Very cool!But hey, let us hear some A/B test samples between yours and the Gyraf's version first.
I am usually more impressed with the greatness of the sound than a few more boards inside the box... :wink: :green:
New 'ultimate' SSL buss comp clone ;-)
- Thread starter SSLtech
- Start date
Help Support GroupDIY Audio Forum:
Very cool!But hey, let us hear some A/B test samples between yours and the Gyraf's version first.
I am usually more impressed with the greatness of the sound than a few more boards inside the box... :wink: :green:
G
Guest
Guest
Keith
Just how high do you want to set the bar
Bloody scousers
nice one
Simon
oh - nice 2 pole EAO switches!
Just how high do you want to set the bar
Bloody scousers
nice one
Simon
oh - nice 2 pole EAO switches!
SSLtech
Well-known member
Okay, test numbers are in:
THD+N (1kHz input @ 0dBu) Threshold adjusted for 10dB gain reduction, 2:1 compression ratio.
Ch1: 0.012%
Ch2: 0.014%
Not "straight-wire", but pretty good.
I figured out that you can cross-link the outputs and still trim the VCAs individually, -you just pull the output 5532 from it's socket on the channel that you're NOT trimming... As usual, make sure the VCAs are warmed up to operating temperature before you adjust though...
There's about 0.2dB inter-channel threshold sensitivity difference, but I'd never actually had cause to measure the inter-channel matching on the real console bus compressor before now... -It seems that if I patch a tone into the pre-VCA left front, center, right and surround channels of the two SL9000's that I've tested so far, there is a slight variation in terms of threshold sensitivity, so this is not abnormal.
Ratio measurement: 2:1 measures as about 2.1:1, 4:1 measures as 3.95:1 and 10:1 measures as about 9:1. I could adjust the 1kΩ resistors to get a closer match at 10:1, but then 4:1 would be off a little, and 2:1 would also be off... It's more than good enough for me! (didn't have the test set in the control room, so I can't give you SL9000 distortion numbers or ratio calculations just yet...
Time to screw the lid on.
I did make some more links and changes to get it to track better:
Link point 'D' on both boards.
Link point 'E' on both boards.
Link point 'F' on both boards.
Remove the 3M3 resistor to ground on the slave
Include the TL072 in the slave build, do not omit it as I previously tried... -This way works better.
I think that's everything.
One of my VCAs died. I had one spare (mercifully!). I actually thought that I had a second spare somewhere, but I can't find it, and now I've had to use the one spare that I definately did have, I definately no longer have a pair of 202XTs. -Someone was asking if I had a pair to sell, but I definately don't for now... :sad:
Anyhow... This bad 202XT probably has one bad 2151 in it, and probably seven good ones, so I'll pull them and test them. Maybe then I can 'resurrect' it with a 2151 if I can get hold of one.
One more thing: the unit has 6dB gain in bypass. I reckon that replacing the 15K resistor in the first half of the 5532 output driver with a 7.5k resistor (or just adding another 15K resistor in parallel) should sort out that problem. That's one more thing to look at before I sign off on the project.
Keith
THD+N (1kHz input @ 0dBu) Threshold adjusted for 10dB gain reduction, 2:1 compression ratio.
Ch1: 0.012%
Ch2: 0.014%
Not "straight-wire", but pretty good.
I figured out that you can cross-link the outputs and still trim the VCAs individually, -you just pull the output 5532 from it's socket on the channel that you're NOT trimming... As usual, make sure the VCAs are warmed up to operating temperature before you adjust though...
There's about 0.2dB inter-channel threshold sensitivity difference, but I'd never actually had cause to measure the inter-channel matching on the real console bus compressor before now... -It seems that if I patch a tone into the pre-VCA left front, center, right and surround channels of the two SL9000's that I've tested so far, there is a slight variation in terms of threshold sensitivity, so this is not abnormal.
Ratio measurement: 2:1 measures as about 2.1:1, 4:1 measures as 3.95:1 and 10:1 measures as about 9:1. I could adjust the 1kΩ resistors to get a closer match at 10:1, but then 4:1 would be off a little, and 2:1 would also be off... It's more than good enough for me! (didn't have the test set in the control room, so I can't give you SL9000 distortion numbers or ratio calculations just yet...
Time to screw the lid on.
I did make some more links and changes to get it to track better:
Link point 'D' on both boards.
Link point 'E' on both boards.
Link point 'F' on both boards.
Remove the 3M3 resistor to ground on the slave
Include the TL072 in the slave build, do not omit it as I previously tried... -This way works better.
I think that's everything.
One of my VCAs died. I had one spare (mercifully!). I actually thought that I had a second spare somewhere, but I can't find it, and now I've had to use the one spare that I definately did have, I definately no longer have a pair of 202XTs. -Someone was asking if I had a pair to sell, but I definately don't for now... :sad:
Anyhow... This bad 202XT probably has one bad 2151 in it, and probably seven good ones, so I'll pull them and test them. Maybe then I can 'resurrect' it with a 2151 if I can get hold of one.
One more thing: the unit has 6dB gain in bypass. I reckon that replacing the 15K resistor in the first half of the 5532 output driver with a 7.5k resistor (or just adding another 15K resistor in parallel) should sort out that problem. That's one more thing to look at before I sign off on the project.
Keith
SSLtech
Well-known member
Indeed, Wayne.
In fact the two different methods both have benefits to consider: the "highest side only" method is useful for broadcasters, who need to ensure that NEITHER side clips, while maintaining maximum modulation. The "Sum of both" takes into account that a signal SOUNDS louder when it appears in both channels, even if neither channel is louder.
In both cases, the stereo image is rock-steady, since the control voltage is ALWAYS the same (I have it hard-wire-linked) but the decision about when to reduce gain to all channels, and by how much, is differently-weighted by the two methods.
Here's a tip everybody: The two 15kΩ resistors on each board: The ones closest to the screen-printed "Output" text on the component side. Replace them with trim-pots. If you're using the balanced output, the gain is 6dB in bypass if you build it as per Jakob's original Gyraf version. It'll be 0dB if you use it unbalanced and take the high-leg only, ignoring the low leg, but for most of us who use it balanced, it'll be 6dB too high. Basically insert the GSSL and you have more gain... even in bypass. 7.5K or thereabouts (my unit actually exhibited more like 5-point-something dB gain per VCA) is the number, but a trim pot will allow you to trim out slight side-to-side variations.
In bypass (still going through the VCAs, but with GR disabled) I'm currently measuring 0.007% THD+N (1kHz @ +4dBu)... :shock: That is -I think- the lowest I've ever seen on a GSSL. -I think that Gil (BR) should be able to confirm that: -I've calibrated several which he's brought in here after he's finished building them. -In fact at +22dBu into a 100kΩ load, the THD+N is down to 0.004%.
-Now THAT's a little closer to straight-wire! :twisted:
I know, I know... 100K load is not really too much of a trial of the output stages, but I'm just surprised that the VCAs are that clean. -It's actually a lower number than I'd hoped for!
Maybe I should condense this into a single-post set of instructions to add into the META... or maybe just add a link for this thread into the META...
Keef
In fact the two different methods both have benefits to consider: the "highest side only" method is useful for broadcasters, who need to ensure that NEITHER side clips, while maintaining maximum modulation. The "Sum of both" takes into account that a signal SOUNDS louder when it appears in both channels, even if neither channel is louder.
In both cases, the stereo image is rock-steady, since the control voltage is ALWAYS the same (I have it hard-wire-linked) but the decision about when to reduce gain to all channels, and by how much, is differently-weighted by the two methods.
Here's a tip everybody: The two 15kΩ resistors on each board: The ones closest to the screen-printed "Output" text on the component side. Replace them with trim-pots. If you're using the balanced output, the gain is 6dB in bypass if you build it as per Jakob's original Gyraf version. It'll be 0dB if you use it unbalanced and take the high-leg only, ignoring the low leg, but for most of us who use it balanced, it'll be 6dB too high. Basically insert the GSSL and you have more gain... even in bypass. 7.5K or thereabouts (my unit actually exhibited more like 5-point-something dB gain per VCA) is the number, but a trim pot will allow you to trim out slight side-to-side variations.
In bypass (still going through the VCAs, but with GR disabled) I'm currently measuring 0.007% THD+N (1kHz @ +4dBu)... :shock: That is -I think- the lowest I've ever seen on a GSSL. -I think that Gil (BR) should be able to confirm that: -I've calibrated several which he's brought in here after he's finished building them. -In fact at +22dBu into a 100kΩ load, the THD+N is down to 0.004%.
-Now THAT's a little closer to straight-wire! :twisted:
I know, I know... 100K load is not really too much of a trial of the output stages, but I'm just surprised that the VCAs are that clean. -It's actually a lower number than I'd hoped for!
Maybe I should condense this into a single-post set of instructions to add into the META... or maybe just add a link for this thread into the META...
Keef
SSLtech
Well-known member
[quote author="mediatechnology"]Does your highest value hard-link look anything like this?[/quote]
Not quite... the GSSL has a detector which outputs through a 'backwards' diode (it's a negative-going voltage at that point) and so the two diode outputs can be hard-linked without any further complexity... Of course the individual detector/ratio/threshold functions have to be linked in addition, to ensure that the two detectors actually generate the same signal given the same input and control setting conditions:
Here's the schemo:
http://www.gyraf.dk/gy_pd/ssl/ssl_sch.gif
I think I'm reading it that the detection is peak detection rather than RMS... Am I looking at that correctly?
The hard-link is at point "C" just above the meter in the schematic. In order to make both unit sidechains behave equally under all setting conditions, the seven resistors, the switch and the diode associates with the ratio control are duplicated, and point 'E' is linked (to match makeup gain), point 'F' is linked (to match threshold sensitivity -or sidechain gain-) Point "D" is linked (return from shared attack & release time constants... slave 3.3MΩ resistor is removed).
That's the 'broadcast' mode. In "Gyraf" mode (L+R sum) the slave channel sidechain is silenced to remove its influence, by disconnecting the 47kΩ resistor into the sidechain. It's then added to the summing point after the 47kΩ resistor at the input of the sidechain opamp for the master channel. Then a single sidechain works, acting on a summed signal, just as in Jakob's variant.
Keith
Not quite... the GSSL has a detector which outputs through a 'backwards' diode (it's a negative-going voltage at that point) and so the two diode outputs can be hard-linked without any further complexity... Of course the individual detector/ratio/threshold functions have to be linked in addition, to ensure that the two detectors actually generate the same signal given the same input and control setting conditions:
Here's the schemo:
http://www.gyraf.dk/gy_pd/ssl/ssl_sch.gif
I think I'm reading it that the detection is peak detection rather than RMS... Am I looking at that correctly?
The hard-link is at point "C" just above the meter in the schematic. In order to make both unit sidechains behave equally under all setting conditions, the seven resistors, the switch and the diode associates with the ratio control are duplicated, and point 'E' is linked (to match makeup gain), point 'F' is linked (to match threshold sensitivity -or sidechain gain-) Point "D" is linked (return from shared attack & release time constants... slave 3.3MΩ resistor is removed).
That's the 'broadcast' mode. In "Gyraf" mode (L+R sum) the slave channel sidechain is silenced to remove its influence, by disconnecting the 47kΩ resistor into the sidechain. It's then added to the summing point after the 47kΩ resistor at the input of the sidechain opamp for the master channel. Then a single sidechain works, acting on a summed signal, just as in Jakob's variant.
Never thought of that... but of course I'd expect the feed-through to be like-pole, so yes I suppose it should... -So much the better! :green:
Keith
Top job, Keith. Impressive. I love the idea of being able to use two sidechains like the original.
I think it would be quite a useful job if we did a new Gyraf VCA Buss Compressor PCB (that's how I'd like to call it from now on) that had mono/dual sidechain selectable configuration like yours, also with HP filter incorporated on board.
I'd love to give it a try sometime myself, if I can spare a time after next month.
So, would we own the copyright if I just did that? And then it would be Jakob's turn to slam me for my mistakes.
I kid I kid :twisted:
See how it goes.
What value are the R networks that feed the 218xs on 202XT, would you be able to read them for us?
mk
I think it would be quite a useful job if we did a new Gyraf VCA Buss Compressor PCB (that's how I'd like to call it from now on) that had mono/dual sidechain selectable configuration like yours, also with HP filter incorporated on board.
I'd love to give it a try sometime myself, if I can spare a time after next month.
So, would we own the copyright if I just did that? And then it would be Jakob's turn to slam me for my mistakes.
I kid I kid :twisted:
See how it goes.
What value are the R networks that feed the 218xs on 202XT, would you be able to read them for us?
mk
SSLtech
Well-known member
[quote author="mkoz"]What value are the R networks that feed the 218xs on 202XT, would you be able to read them for us?[/quote]
I can take pics and/or scan the circuit board if I write this one off and opt to part it out into 2151's... Then the (slightly nicer) 2181's that I have as sidechain VCAs in this one can be promoted to better things... :twisted: Oh, and yes, I can measure the resistances of the R-packs when the VCAs are out.
I think that the idea of a redraw with differential VCAs is a worthwhile project. Of course, the 'optional VCA' PCB layout arrangement which Jakob included is VERY flexible... something like that would be very useful, allowing 202C's, 202Ts, 2001's and a few other 'can' type VCA choices, in addition to 215x/218x SIL VCAs.
I think -to be proper- it might be 'protocoligorically correct' to touch base with Jakob as well... if a 'team effort' is acceptable to everyone, then everyone's input can help make an über-version available for everybody... Perhaps we could agree on a "slave" unit connector standard: like A D-sub, which could be used to chain-up slave versions for quad/5.1 applications?
♫
Team playing is what it's all about... That's what they keep screaming at me every brass band rehearsal, but ... you know what.... ?ûçk 'em! :twisted:
:wink:
Keith
I can take pics and/or scan the circuit board if I write this one off and opt to part it out into 2151's... Then the (slightly nicer) 2181's that I have as sidechain VCAs in this one can be promoted to better things... :twisted: Oh, and yes, I can measure the resistances of the R-packs when the VCAs are out.
I think that the idea of a redraw with differential VCAs is a worthwhile project. Of course, the 'optional VCA' PCB layout arrangement which Jakob included is VERY flexible... something like that would be very useful, allowing 202C's, 202Ts, 2001's and a few other 'can' type VCA choices, in addition to 215x/218x SIL VCAs.
I think -to be proper- it might be 'protocoligorically correct' to touch base with Jakob as well... if a 'team effort' is acceptable to everyone, then everyone's input can help make an über-version available for everybody... Perhaps we could agree on a "slave" unit connector standard: like A D-sub, which could be used to chain-up slave versions for quad/5.1 applications?
♫
Team playing is what it's all about... That's what they keep screaming at me every brass band rehearsal, but ... you know what.... ?ûçk 'em! :twisted:
:wink:
Keith
SSLtech
Well-known member
Okay... I'll try it!
So -to be absolutely clear- You said you "put 1V RMS into both". -Was it the same signal (i.e. correlated) or a pair of 1V RMS non-correlated signals? -Or would you like both variants of the test to be carried out?
Keith
So -to be absolutely clear- You said you "put 1V RMS into both". -Was it the same signal (i.e. correlated) or a pair of 1V RMS non-correlated signals? -Or would you like both variants of the test to be carried out?
Keith
[quote author="SSLtech"]I think -to be proper- it might be 'protocoligorically correct' to touch base with Jakob as well...[/quote]
He may be a bit allergic to my touch these days, like "a touch a touch too much" but yeah, absolutely
Damn I want to get AC/DC'd in the middle of the night, I'll stick HWTH on now...
"...seems like a touch, ga-gowww ga-gowww, a touch too much, ga-gowww ga-gee-ga-go-gowww..."
MK
He may be a bit allergic to my touch these days, like "a touch a touch too much" but yeah, absolutely
Damn I want to get AC/DC'd in the middle of the night, I'll stick HWTH on now...
"...seems like a touch, ga-gowww ga-gowww, a touch too much, ga-gowww ga-gee-ga-go-gowww..."
MK
SSLtech
Well-known member
HWTH?
-Have you heard the version by "Hayseed Dixie"?
Fan-friggin' tastic! -Bluegrass, complete with flamin' banjo pickin'!
Keef
-Have you heard the version by "Hayseed Dixie"?
Fan-friggin' tastic! -Bluegrass, complete with flamin' banjo pickin'!
Keef
[quote author="SSLtech"]HWTH?
-Have you heard the version by "Hayseed Dixie"?
Fan-friggin' tastic! -Bluegrass, complete with flamin' banjo pickin'!
Keef[/quote]
Ehehe, if you like that one then try Southern Van Halen's "Strummin' With The Devil" too.
http://www.youtube.com/watch?v=kX9qcggRo18
Talking about AC/DC btw, my brother and I have done a Turkish 9/8 version of You Shook Me All Night Long (we changed the time signature so to speak and somehow made it work).
We've changed the words to "You Shook It All Night Long" as well, cos the plan is that the video will feature forty belly dancers shaking it all night long.
Hilarious.
If we can get anywhere with it, I'll ask if Brian Johnson would be interested in playing the sultan and Angus Young playing some saz solo over it.
Not likely to happen, but would be a nice try though.
Damn, I keep giving the ideas away. I'll never learn.
MK Oz
-Have you heard the version by "Hayseed Dixie"?
Fan-friggin' tastic! -Bluegrass, complete with flamin' banjo pickin'!
Keef[/quote]
Ehehe, if you like that one then try Southern Van Halen's "Strummin' With The Devil" too.
http://www.youtube.com/watch?v=kX9qcggRo18
Talking about AC/DC btw, my brother and I have done a Turkish 9/8 version of You Shook Me All Night Long (we changed the time signature so to speak and somehow made it work).
We've changed the words to "You Shook It All Night Long" as well, cos the plan is that the video will feature forty belly dancers shaking it all night long.
Hilarious.
If we can get anywhere with it, I'll ask if Brian Johnson would be interested in playing the sultan and Angus Young playing some saz solo over it.
Not likely to happen, but would be a nice try though.
Damn, I keep giving the ideas away. I'll never learn.
MK Oz
veermaster
Well-known member
Hi Keith,
I´m really impressed about your ideas!
About that sidechain mix resistor thingy: I implemented Steffens HP PCB in my GSSL, so the resistor I would have to change is probably the 47K just before the PCB´s "OUT", right ?
Thanks in advance.
[quote author="mkoz"]Top job, Keith. Impressive. I love the idea of being able to use two sidechains like the original.
I think it would be quite a useful job if we did a new Gyraf VCA Buss Compressor PCB (that's how I'd like to call it from now on) that had mono/dual sidechain selectable configuration like yours, also with HP filter incorporated on board.
I'd love to give it a try sometime myself, if I can spare a time after next month.
mk[/quote]
I guess everybody around here would be really happy to have a PCB with all the new ideas implemented! :thumb:
But of course all people involved have to agree on that...
Emre
I´m really impressed about your ideas!
About that sidechain mix resistor thingy: I implemented Steffens HP PCB in my GSSL, so the resistor I would have to change is probably the 47K just before the PCB´s "OUT", right ?
Thanks in advance.
[quote author="mkoz"]Top job, Keith. Impressive. I love the idea of being able to use two sidechains like the original.
I think it would be quite a useful job if we did a new Gyraf VCA Buss Compressor PCB (that's how I'd like to call it from now on) that had mono/dual sidechain selectable configuration like yours, also with HP filter incorporated on board.
I'd love to give it a try sometime myself, if I can spare a time after next month.
mk[/quote]
I guess everybody around here would be really happy to have a PCB with all the new ideas implemented! :thumb:
But of course all people involved have to agree on that...
Emre
clintrubber
Well-known member
[quote author="mkoz"]Damn, I keep giving the ideas away. I'll never learn.
MK Oz[/quote]
Simple, add a copyright notice to your ideas and all will be fine :wink:
(Sorry, couldn't resist. No intention to re-open any discussions or participating. No offense. Batteries not included.)
Have a good day,
Peter
MK Oz[/quote]
Simple, add a copyright notice to your ideas and all will be fine :wink:
(Sorry, couldn't resist. No intention to re-open any discussions or participating. No offense. Batteries not included.)
Have a good day,
Peter
SSLtech
Well-known member
The detector seems to produce the same voltage with both channels fed as it does with either single channel fed with the same RMS signal level. This appears to be true for both correlated and uncorrelated signals (I blew the dust of an old Loftech TS-1 and fired it into the other channel at about 1.5kHz after first using the second A2 output... both at 1kHz.)
So with your unit, does the GR metering indicate a GR increase when a second channel is added at the same level... and at Full-limit, what is the displayed GR increase? 3dB? 6dB? -That'd be informative... The A2 in 2-channel RMS signal level mode shows no decrease in chennel level when the second channel is added at he same level, also the GR metering stays constant, no matter whether channel 1, channel 2 or channels 1&2 are unmuted at the input to the unit.
So it appears that it's the "greater of two", which is what I was aiming for, though I rather like the "equal power" detection as a third option... Hmmm.... I wonder how that might be integrate-able...
With this option combination, it should be quite possible to evolve a 'GML-beating' (or at least "GML-competing"!) design... Using beyond-reproach signal path integrity, and a selectable behaviour control (greater-of-two peak, RMS power-sum, and full-sum influence)
NOW we're talking!
Keith
So with your unit, does the GR metering indicate a GR increase when a second channel is added at the same level... and at Full-limit, what is the displayed GR increase? 3dB? 6dB? -That'd be informative... The A2 in 2-channel RMS signal level mode shows no decrease in chennel level when the second channel is added at he same level, also the GR metering stays constant, no matter whether channel 1, channel 2 or channels 1&2 are unmuted at the input to the unit.
So it appears that it's the "greater of two", which is what I was aiming for, though I rather like the "equal power" detection as a third option... Hmmm.... I wonder how that might be integrate-able...
With this option combination, it should be quite possible to evolve a 'GML-beating' (or at least "GML-competing"!) design... Using beyond-reproach signal path integrity, and a selectable behaviour control (greater-of-two peak, RMS power-sum, and full-sum influence)
NOW we're talking!
Keith
sahib
Well-known member
[quote author="clintrubber"][quote author="mkoz"]Damn, I keep giving the ideas away. I'll never learn.
MK Oz[/quote]
Simple, add a copyright notice to your ideas and all will be fine :wink:
(Sorry, couldn't resist. No intention to re-open any discussions or participating. No offense. Batteries not included.)
Have a good day,
Peter[/quote]
© mcoz & mkoz
:grin: :grin: :grin:
MK Oz[/quote]
Simple, add a copyright notice to your ideas and all will be fine :wink:
(Sorry, couldn't resist. No intention to re-open any discussions or participating. No offense. Batteries not included.)
Have a good day,
Peter[/quote]
© mcoz & mkoz
:grin: :grin: :grin:
SSLtech
Well-known member
Void where prohibited. Must be 18 years or older. Non-terrestrial participants excepted. Please use responsibly. All designs are the property of Keef unless specifically excepted in writing. Applicable taxes, duties, excise and liens remain the responsibility of the constructor. SSLtech is an equal-opportunity-insulter. Constructors waive any and all rights to property, income, chattels or dowry hereby or hitherto attached or in any way suggested. Party of the first party (hereinafter referred to as 'SSLtech') reserves the right to share his opinion of the party(/ies) of the second party (hereinafter known as 'idiot', 'fool', 'bird-brain', 'berk', 'toerag' or similar) with parties of the third party, either publicly, privately using messages conveyed via a publicly-viewed-but-privately-owned internet communications network, or to his mates down the boozer. All rights reserved, gawd bless the pensioners.
SSLtech
Well-known member
[quote author="mediatechnology"]I'm sorry Keith could you read that fine print in :10?[/quote]
Hmmmm.... I could tighten up the "birdbrain" a little... what's it coming in at right now?
Oh well, here goes... (deep breath)
:green:
Keef
Hmmmm.... I could tighten up the "birdbrain" a little... what's it coming in at right now?
Oh well, here goes... (deep breath)
:green:
Keef
SSLtech
Well-known member
[quote author="mediatechnology"]Another experiment for Keith if he's curious...[/quote]
People say that about me all the time...
-Oh that Keith... -he IS curious!!!
Ripple in terms of envelope modulation caused by the difference frequency? -Maybe I should try closer frequencies?
Keith
People say that about me all the time...
-Oh that Keith... -he IS curious!!!
Ripple in terms of envelope modulation caused by the difference frequency? -Maybe I should try closer frequencies?
Keith
SSLtech
Well-known member
Well, a quick test in "higher-of-two" mode shows it to be essentially immune to difference frequencies, even at quick attack/release settings.
Flip it to "1+1" (summed-single sidechain) and you get a modulation. As the beats (difference ?) get down to single-digit, then it appears to be a fast-pulsing 6dB-swell/infinite-dip envelope modulation, and the sidechain starts to pulse as fast as the attack/release controls will permit. -Of course, the individual channel outputs will not be summed, so they just show envelope modulation of a single input ?, but the reason for the modulation isn't clear until you hear the other channel as well.
On paper this sounds like it should be wrong, but in real life the acoustic summing of both speaker outputs makes the compressor action sound justified... though for broadcast transmission, there'd be no need for any GR activity, because both channels can be carried at 100% modulation with no need for GR...
All very interesting...
Oh, and that Loftech is a drifty thing... I keep having to trim the fine ? control to keep the ?diff from drifting all over the place!
Keith
Flip it to "1+1" (summed-single sidechain) and you get a modulation. As the beats (difference ?) get down to single-digit, then it appears to be a fast-pulsing 6dB-swell/infinite-dip envelope modulation, and the sidechain starts to pulse as fast as the attack/release controls will permit. -Of course, the individual channel outputs will not be summed, so they just show envelope modulation of a single input ?, but the reason for the modulation isn't clear until you hear the other channel as well.
On paper this sounds like it should be wrong, but in real life the acoustic summing of both speaker outputs makes the compressor action sound justified... though for broadcast transmission, there'd be no need for any GR activity, because both channels can be carried at 100% modulation with no need for GR...
All very interesting...
Oh, and that Loftech is a drifty thing... I keep having to trim the fine ? control to keep the ?diff from drifting all over the place!
Keith
SSLtech
Well-known member
I'm at home sick at the moment; away from the super-digit test gear.
However, I do have a DK-audio vectorscope at home... also a decent X/Y scope capability...
However I've never perceived any width alteration with either detection method. -Certainly the left-to-right volume ratio is utterly unaltered using either approach, as previously observed. The influence of center-dominated sources in left-right mixed program material can change the 'dipping' threshold, but that may be perceived as 'pump-influence' rather than any change in angle on the vectorscope (which requires a relative gain difference). -Remember that the DK audio actually has a ganged AGC on its input, and that it is in fact set for "higher-of-two" peak detection (so that the maximum signal display size without clipping is always used, unless the signal is really low... about 20dB gain range -observed emprically when viewed next directly to SIFAM moving-coil true-VU's) -If this used sum-detection (even peak-acting sum-detection) then the angle should not be changed (angle=ratio of left-to-right) but the signal would either clip at left-or-right or not attain full range at left-plus-right for a fully-in-phase sine wave... (which should always be a single vertical or diagonal line, vertical only at equal level.)
Keith
However, I do have a DK-audio vectorscope at home... also a decent X/Y scope capability...
However I've never perceived any width alteration with either detection method. -Certainly the left-to-right volume ratio is utterly unaltered using either approach, as previously observed. The influence of center-dominated sources in left-right mixed program material can change the 'dipping' threshold, but that may be perceived as 'pump-influence' rather than any change in angle on the vectorscope (which requires a relative gain difference). -Remember that the DK audio actually has a ganged AGC on its input, and that it is in fact set for "higher-of-two" peak detection (so that the maximum signal display size without clipping is always used, unless the signal is really low... about 20dB gain range -observed emprically when viewed next directly to SIFAM moving-coil true-VU's) -If this used sum-detection (even peak-acting sum-detection) then the angle should not be changed (angle=ratio of left-to-right) but the signal would either clip at left-or-right or not attain full range at left-plus-right for a fully-in-phase sine wave... (which should always be a single vertical or diagonal line, vertical only at equal level.)
Keith
