clintrubber
Well-known member
I had seen that one and was wondering a bit what the use of a pan-bypass would be. There'll be applications somewhere of course.
Panning and Matrixing deleted
- Thread starter clintrubber
- Start date
Help Support GroupDIY Audio Forum:
Not to be contrary, but in console design there is a constant tug of war between panel real estate, incremental per channel costs, and utility. While I have put mono switches in the control room master section, I have never used one per channel.
That said, if I were to design a per channel mono switch I'd be inclined to target a design approach where the gain doesn't jump between pan pot centered and defeated.
While they manage to deliver the proper 3dB pan law, the network input impedance is not very opamp friendly, and dumping mA of signal current into PCB grounds makes layouts and grounding design to control fader and send kill in channels more difficult, and crosstalk between channels in larger console configurations.
YMMV
'JR
That said, if I were to design a per channel mono switch I'd be inclined to target a design approach where the gain doesn't jump between pan pot centered and defeated.
While they manage to deliver the proper 3dB pan law, the network input impedance is not very opamp friendly, and dumping mA of signal current into PCB grounds makes layouts and grounding design to control fader and send kill in channels more difficult, and crosstalk between channels in larger console configurations.
YMMV
'JR
G
Guest
Guest
Mediatechnology, if to think of each simple thing as a part of a big complex one it looks a bit different...
IMO the pan circuit needs more than just scaling but pan circuits are well explored so I'll stop picking nits.
I would also advise against their figure 2. (the alternate pan bypass circuit) as it will result in a polarity inversion between normal and bypass. Another unwritten rule of (good) console design is that sundry signal paths maintain polarity integrity, not only relative to each other, but input to output.
Whatever..
JR
I would also advise against their figure 2. (the alternate pan bypass circuit) as it will result in a polarity inversion between normal and bypass. Another unwritten rule of (good) console design is that sundry signal paths maintain polarity integrity, not only relative to each other, but input to output.
Whatever..
JR
matthias
Well-known member
R3 and R4 should be 4,99K, otherwise the output level will be a bit louder than the input level.
I made some simluations with that pan pot cirucit the main-problem is that if you turn the pot in the middle the output volume is way louder than the input signal.
If you apply a 0,775 V sinewave, 1khz to the input, you get 0,775 V at every end of the pot an 1,10 V if you rotate the pot to the middle setting.
is it possible to use a lin. dual 1K pot as a standard attenuator?
this should give a more linear curve?
what do you think?
I made some simluations with that pan pot cirucit the main-problem is that if you turn the pot in the middle the output volume is way louder than the input signal.
If you apply a 0,775 V sinewave, 1khz to the input, you get 0,775 V at every end of the pot an 1,10 V if you rotate the pot to the middle setting.
is it possible to use a lin. dual 1K pot as a standard attenuator?
this should give a more linear curve?
what do you think?
NewYorkDave
Well-known member
I'm glad to see new circuits posted and I'm glad to see the critiques, even when it's one of my own circuits being picked apart. As Kev always says, "it's all good."
Wayne, what's the output load on that SSL panpot (the mix resistors)? That has a big bearing on the pan law and should, in effect, be considered part of the pan circuit.
Since we're talking panpots, here's mine:
Image
I developed this little variation when I wanted a panpot with a more constant input impedance than the usual dual-linear-with-pullups type, and less insertion loss than the single-linear "Orban" type. The cost of all this is a 6dB minimum insertion loss, which isn't terrible.
Its constant-power law is good.
Chart
Wayne, what's the output load on that SSL panpot (the mix resistors)? That has a big bearing on the pan law and should, in effect, be considered part of the pan circuit.
Since we're talking panpots, here's mine:
Image
I developed this little variation when I wanted a panpot with a more constant input impedance than the usual dual-linear-with-pullups type, and less insertion loss than the single-linear "Orban" type. The cost of all this is a 6dB minimum insertion loss, which isn't terrible.
Its constant-power law is good.
Chart
[quote author="NewYorkDave"]I'm glad to see new circuits posted and I'm glad to see the critiques, even when it's one of my own circuits being picked apart. As Kev always says, "it's all good."
[/quote]
From a mfg. cost POV using a dual when you could skin that cat with a single is not exactly cost effective, but I like the improvement to insertion loss. My old standby circuit is more like 10dB so that's 4dB less make up gain to boost bus noise.
When performing resistor tricks with pots it's sometime worth reflecting on tolerances. Bulk resistance tolerance for common (screened resist) potentiometers is something like 20%. The ratio metric tracking will be much better than that, so at 50% rotation it can pretty accurately be 50% of the not so accurate 20% overall resistance.
This source of error is true of all typical pan topologies other than the rare, dual custom taper that work "poteniometrically" or pan law is defined by the screening pattern. Tooling up specials is out of the question for DIY and I don't know if anybody is doing this in production.
Note: this is not a big deal. While the 20% tolerance of pot bulk resistance will cause some deviation from predicted insertion loss, in production potentiometers from the same batch will be much closer to each other than the overall tolerance. So you won't typically perceive channel to channel differences.
JR
[/quote]
From a mfg. cost POV using a dual when you could skin that cat with a single is not exactly cost effective, but I like the improvement to insertion loss. My old standby circuit is more like 10dB so that's 4dB less make up gain to boost bus noise.
When performing resistor tricks with pots it's sometime worth reflecting on tolerances. Bulk resistance tolerance for common (screened resist) potentiometers is something like 20%. The ratio metric tracking will be much better than that, so at 50% rotation it can pretty accurately be 50% of the not so accurate 20% overall resistance.
This source of error is true of all typical pan topologies other than the rare, dual custom taper that work "poteniometrically" or pan law is defined by the screening pattern. Tooling up specials is out of the question for DIY and I don't know if anybody is doing this in production.
Note: this is not a big deal. While the 20% tolerance of pot bulk resistance will cause some deviation from predicted insertion loss, in production potentiometers from the same batch will be much closer to each other than the overall tolerance. So you won't typically perceive channel to channel differences.
JR
SSLtech
Well-known member
[quote author="JohnRoberts"]Tooling up specials is out of the question for DIY and I don't know if anybody is doing this in production.[/quote]
If you ever happen to stumble across any Audio & Design B-format pan/rotate boxes, they use TRUE sin/cosine pan pots...
I can't even begin to describe how stunningly useful those things are!!!
Keith
If you ever happen to stumble across any Audio & Design B-format pan/rotate boxes, they use TRUE sin/cosine pan pots...
I can't even begin to describe how stunningly useful those things are!!!
Keith
This would work for a dedicated L-R only mix topology while it would not support multiple track assignments simultaneously. A second issue might be difficulty in executing a proper solo as post pan would be a virtual level. Of course these limitations could be either lived with or worked around (PFL instead of solo for example).
Summing the current outputs would indeed buy you both a freedom from the typical N+1 noise gain common to virtual earth summers, but it would also isolate the channel strip ground from the master section bus ground finessing a often difficult design issue. FWIW and a trick that I did in one old LOFT console where I used synthesized current sources (not VCA) to feed the L/R bus, I also took advantage on the low bus noise gain to use a VCA as the summing amp. The result is a bus structure with remarkable headroom.
Downsides (ignoring cost) are that VCA noise floor may be higher than alternate pan approaches and in a larger bus structure you will have a build up in bus capacitance (manageable in a topology I mentioned to NYD where bus sum amp is turned into a 2 pole LPF). The noise floors would combine incoherently (SQ rt of sum of squares) so increase only 3 dB for every doubling of inputs. Note: the old reality check still applies that this VCA noise floor will be lower than a mic preamp at nominal gain, so noise floor issues will only bother folks too reliant on spec sheets and not real world users.
JR
Summing the current outputs would indeed buy you both a freedom from the typical N+1 noise gain common to virtual earth summers, but it would also isolate the channel strip ground from the master section bus ground finessing a often difficult design issue. FWIW and a trick that I did in one old LOFT console where I used synthesized current sources (not VCA) to feed the L/R bus, I also took advantage on the low bus noise gain to use a VCA as the summing amp. The result is a bus structure with remarkable headroom.
Downsides (ignoring cost) are that VCA noise floor may be higher than alternate pan approaches and in a larger bus structure you will have a build up in bus capacitance (manageable in a topology I mentioned to NYD where bus sum amp is turned into a 2 pole LPF). The noise floors would combine incoherently (SQ rt of sum of squares) so increase only 3 dB for every doubling of inputs. Note: the old reality check still applies that this VCA noise floor will be lower than a mic preamp at nominal gain, so noise floor issues will only bother folks too reliant on spec sheets and not real world users.
JR
SSLtech
Well-known member
Wayne,
Yes, the surround (or more accurately multichannel) application was what has driven me thus far: Specifically B-format to 5.1 ("G-format") and other periphonic decode applications...
That's where I looked into sin/cos, which is perfect. -I did eventually get to well within a dB of it though... (using linear pots)
Keith
Yes, the surround (or more accurately multichannel) application was what has driven me thus far: Specifically B-format to 5.1 ("G-format") and other periphonic decode applications...
That's where I looked into sin/cos, which is perfect. -I did eventually get to well within a dB of it though... (using linear pots)
Keith
G
Guest
Guest
[quote author="mediatechnology"]
Now what have we done here? We have created a subtractive lossless pan circuit. [/quote]
No. We compensated the loss by an extra amp per channel adding noises and distortions.
What I dream about, is a joystick that controls time delays...
Now what have we done here? We have created a subtractive lossless pan circuit. [/quote]
No. We compensated the loss by an extra amp per channel adding noises and distortions.
What I dream about, is a joystick that controls time delays...
G
Guest
Guest
[quote author="mediatechnology"]
And your joystick doesn't?
[/quote]
No, it is digital. It simulates relocation of a sound source in the naturally reverberating environment, between 4 walls.
I can slightly increase gain of the stage before pan control instead of adding one more whole opamp per each channel. :grin:
And your joystick doesn't?
[/quote]
No, it is digital. It simulates relocation of a sound source in the naturally reverberating environment, between 4 walls.
I can slightly increase gain of the stage before pan control instead of adding one more whole opamp per each channel. :grin:
SSLtech
Well-known member
[quote author="Wavebourn"]What I dream about, is a joystick that controls time delays...[/quote]
Not enough.
There's more to it than that. Along with time delay there is ALSO still an amplitude shift, along with a head-shadowing frequency/amplitude modification.
Combined, they do realy start to sound like real motions.
It's been done, it's fabulous, and it comes standard on the Studer D950 and Vista series consoles. in addition, there's also a reflection pattern generation (VSP... Virtual Surround Panning) which fills in rear reflection patterns from changing front incidence angles, AND distance... don't forget the triangulation available to the brain to do with proxmity/distance.
Simple delay will only make mono compatibility laughable, while doing intrinsically little to help the perceived angle of incidence be more forcefully resolved. Likewise front-to-back shifts.
Basically the (truly wonderful) Studer offers several paning options: HRTF1, HRTF2 (Head-Related-Transfer-Functions: 2 options to better suit differently-scaled speaker placement and headphone alternatives) as well as amplitude-only and VSP (the full DSP option, with complete reverberant field and triangulation, HRTF incorporated... Takes a little processing power to do in real-time, but it's VERY good.)
Sadly for all their advantages, each also has at least one significant drawback. For those who insist that amplitude-only panning is garbage, I'd point out that NO other option has anything even approaching the same level of fold-down compatibility. HRTF can sound great on headphones, but very blurred and muddled over speakers. Another HRTF can sound spacious on speakers, but sort of mono-ey over headphones. VSP can also help to destroy location clues already contained in a multi-mic recording... and yet each has a mighty benefit to them.
So there's no magic bullet. No one solution, and -since we listen on differently-spaced speakers in different envoronments- there never will be.
Amplitude panning is the simplest DIY and quite frankly the ONLY analog option. It also folds down wonderfully. It's served us for decades, and it's very useful. As soon as you introduce more variables, you introduce more difficulties, and not every listener is placed to take advantage of the benefits more recently available panning alternatives can offer... and if youre not getting advantages, you're getting disadvantages instead.
Keith
Not enough.
There's more to it than that. Along with time delay there is ALSO still an amplitude shift, along with a head-shadowing frequency/amplitude modification.
Combined, they do realy start to sound like real motions.
It's been done, it's fabulous, and it comes standard on the Studer D950 and Vista series consoles. in addition, there's also a reflection pattern generation (VSP... Virtual Surround Panning) which fills in rear reflection patterns from changing front incidence angles, AND distance... don't forget the triangulation available to the brain to do with proxmity/distance.
Simple delay will only make mono compatibility laughable, while doing intrinsically little to help the perceived angle of incidence be more forcefully resolved. Likewise front-to-back shifts.
Basically the (truly wonderful) Studer offers several paning options: HRTF1, HRTF2 (Head-Related-Transfer-Functions: 2 options to better suit differently-scaled speaker placement and headphone alternatives) as well as amplitude-only and VSP (the full DSP option, with complete reverberant field and triangulation, HRTF incorporated... Takes a little processing power to do in real-time, but it's VERY good.)
Sadly for all their advantages, each also has at least one significant drawback. For those who insist that amplitude-only panning is garbage, I'd point out that NO other option has anything even approaching the same level of fold-down compatibility. HRTF can sound great on headphones, but very blurred and muddled over speakers. Another HRTF can sound spacious on speakers, but sort of mono-ey over headphones. VSP can also help to destroy location clues already contained in a multi-mic recording... and yet each has a mighty benefit to them.
So there's no magic bullet. No one solution, and -since we listen on differently-spaced speakers in different envoronments- there never will be.
Amplitude panning is the simplest DIY and quite frankly the ONLY analog option. It also folds down wonderfully. It's served us for decades, and it's very useful. As soon as you introduce more variables, you introduce more difficulties, and not every listener is placed to take advantage of the benefits more recently available panning alternatives can offer... and if youre not getting advantages, you're getting disadvantages instead.
Keith
[quote author="mediatechnology"]John - You could still do a destructive in-place solo though right. And PFL. And why would the grounds not be able to common? Remember this is not a 56 input desk 30+ foot long desk but a simple rack-mount summing mixer concept driven from a D/A. No reason why the panner VCAs for each module couldn't be located in the master section at the output I-V. Pan and Level control by DC. And the big deal is no N+1 noise gain.
[/quote]
I guess I wasn't clear... VCAs located at each input channel would be properly referenced to the local channel ground and since their output is a current they would ignore any (voltage) potential differences between channel and bus master ground. This is a good thing. If course if you locate all the VCAs in the master section then you need to deal with the relative ground potentials.
JR
[/quote]
I guess I wasn't clear... VCAs located at each input channel would be properly referenced to the local channel ground and since their output is a current they would ignore any (voltage) potential differences between channel and bus master ground. This is a good thing. If course if you locate all the VCAs in the master section then you need to deal with the relative ground potentials.
JR
G
Guest
Guest
[quote author="SSLtech"][quote author="Wavebourn"]What I dream about, is a joystick that controls time delays...[/quote]
Not enough.
There's more to it than that. Along with time delay there is ALSO still an amplitude shift, along with a head-shadowing frequency/amplitude modification.
Combined, they do realy start to sound like real motions.
[/quote]
It is true, however! What I mean, a combined pan + reverberation control, where in addition to traditional amplitude pan control via DSP reverberation is controlled (initial delay shifts between right and left, amplitude balance of direct/reverberation far-near).
I had an experimental console with "far-near" sliders that regulated balance between direct and reverberated signals, it was very convenient to play with dynamically. But it was not enough, since reverberator was a tape loop type mono reverberator, so I dreamed of the way to shift recording heads for each channel when controlling panoramic pot... Thecnicaly it would be possible, but too expensive!
Now I believe it is possible when DSP is so cheap and well developed, but unfortunately I have no experience with it.
It is different. I don't mean audio tracks for movies, I mean simple concerts where I can simulate environment of a big nice hall.
Not enough.
There's more to it than that. Along with time delay there is ALSO still an amplitude shift, along with a head-shadowing frequency/amplitude modification.
Combined, they do realy start to sound like real motions.
[/quote]
It is true, however! What I mean, a combined pan + reverberation control, where in addition to traditional amplitude pan control via DSP reverberation is controlled (initial delay shifts between right and left, amplitude balance of direct/reverberation far-near).
I had an experimental console with "far-near" sliders that regulated balance between direct and reverberated signals, it was very convenient to play with dynamically. But it was not enough, since reverberator was a tape loop type mono reverberator, so I dreamed of the way to shift recording heads for each channel when controlling panoramic pot... Thecnicaly it would be possible, but too expensive!
Now I believe it is possible when DSP is so cheap and well developed, but unfortunately I have no experience with it.
It is different. I don't mean audio tracks for movies, I mean simple concerts where I can simulate environment of a big nice hall.
SSLtech
Well-known member
[quote author="Wavebourn"]What I mean, a combined pan + reverberation control, where in addition to traditional amplitude pan control via DSP reverberation is controlled (initial delay shifts between right and left, amplitude balance of direct/reverberation far-near).[/quote]
You're describing VSP almost perfectly.
Yes, as long as you KNOW the playback/listening envoronment, you can do WONDERFUL things with it. -Go look it up on the web!
Of course, being Swiss, STUDER have researched this to an astonishing degree, and gone way beyond reflection/direct balance, to the point where the reflection triangulation angles all change with near/far proximity adjustments, and other factors far too numerous to dopcument here, even if my menory would allow... :wink:
I'll try to find a link, ( http://www.studio-systems.com/audiofeatures/JulyAug2003/Multichannel/multi90.htm will get you started) -but they published a paper on it, and it's utterly brilliant. -However, the drawbacks when you fold down for different envorinments mean that nobody is following Studer down that path licensing-wise... TOo much money for too little benefit, (not to mention the disadvantages) almost certainly.
Keith
You're describing VSP almost perfectly.
Yes, as long as you KNOW the playback/listening envoronment, you can do WONDERFUL things with it. -Go look it up on the web!
Of course, being Swiss, STUDER have researched this to an astonishing degree, and gone way beyond reflection/direct balance, to the point where the reflection triangulation angles all change with near/far proximity adjustments, and other factors far too numerous to dopcument here, even if my menory would allow... :wink:
I'll try to find a link, ( http://www.studio-systems.com/audiofeatures/JulyAug2003/Multichannel/multi90.htm will get you started) -but they published a paper on it, and it's utterly brilliant. -However, the drawbacks when you fold down for different envorinments mean that nobody is following Studer down that path licensing-wise... TOo much money for too little benefit, (not to mention the disadvantages) almost certainly.
Keith
[quote author="Wavebourn"]I had an experimental console with "far-near" sliders that regulated balance between direct and reverberated signals, it was very convenient to play with dynamically. But it was not enough, since reverberator was a tape loop type mono reverberator, so I dreamed of the way to shift recording heads for each channel when controlling panoramic pot... Thecnicaly it would be possible, but too expensive!
Now I believe it is possible when DSP is so cheap and well developed, but unfortunately I have no experience with it.[/quote]
I once wrote a software that extended an existing amplitude based panning system (3 dimensional over arbitrarily positioned speakers) by adding a 'distance' control. In the end it was just adjusting the balance between direct and reverberated signal (for each speaker with an extra delay for each speaker) depending in the 'distance' you dialed in. Combined with the XYZ movements the basic system supports (see http://www.acoustics.hut.fi/research/cat/vbap/ for the original system) this was quite impressive.
The first version didn't do any DSP and instead was just using MIDI signals to adjust signal and reverb levels in a Yamaha digital mixer.
Maybe this could be coded completely in analog, but would need at least VCAs as far as I can see.
Olaf
Now I believe it is possible when DSP is so cheap and well developed, but unfortunately I have no experience with it.[/quote]
I once wrote a software that extended an existing amplitude based panning system (3 dimensional over arbitrarily positioned speakers) by adding a 'distance' control. In the end it was just adjusting the balance between direct and reverberated signal (for each speaker with an extra delay for each speaker) depending in the 'distance' you dialed in. Combined with the XYZ movements the basic system supports (see http://www.acoustics.hut.fi/research/cat/vbap/ for the original system) this was quite impressive.
The first version didn't do any DSP and instead was just using MIDI signals to adjust signal and reverb levels in a Yamaha digital mixer.
Maybe this could be coded completely in analog, but would need at least VCAs as far as I can see.
Olaf
SSLtech
Well-known member
Another link:
http://www.adtweb.it/download/articoli/Studer_D950.pdf
Go about 2/3rds of the way into the pdf, it was intended to be a powerpoint with talk/information delivered alongside it, but it does list the panning parameters... distance is one.
Ah... HERE is the one I was searching for!
Start at page 5 and you'll start to see what they did...
Keith
http://www.adtweb.it/download/articoli/Studer_D950.pdf
Go about 2/3rds of the way into the pdf, it was intended to be a powerpoint with talk/information delivered alongside it, but it does list the panning parameters... distance is one.
Ah... HERE is the one I was searching for!
Start at page 5 and you'll start to see what they did...
Keith
G
Guest
Guest
Keith, thank you for you link!
It is indeed very interesting, but my variant is simplier and more portable, including folding of channels.
It is indeed very interesting, but my variant is simplier and more portable, including folding of channels.
SSLtech
Well-known member
Yes, anything which is subtractively derived is re-combinably compatible, providing that no subsequent delays or UNWANTED phase shifts (i.e. those not intrinsically associated with WANTED equalisation) are added along the way... in other words, if you wanted to EQ the "side" of a Mid-Side matrix, don't insert any A/D conversion or other such delays which would destroy most instantaneous polaity/phase relationships.
But essentially yes. Now, -what was the question again? :wink:
As far as mono LF (a BIG deal in the days of stereo vinyl cutting!) recent papers published in ambisonics are suggesting that actually INCREASING the LF content of the 'side' (or L-R subtractively-derived matrix channel) actually increases the sens of 'spaciousness' in most mixes or recordings. I'd have to do some digging to locate the references, but I know that I've tried it and it's a parlour trick which seems to work well until the brain realises that it's being 'faked-out'... and that different people seem to have massively different thresholds of tolerance.
But no doubt that matrix panning/matrix processing is a VERY powerful tool... If a compelling benefit presents itself, then it becomes attractive to get into and back out of a matrix, (mastering is a killer app) but most people are sadly reluctant...
I derived a logical matrix a while ago, which people are starting to use in film mixing... there's nothing really complicated about it, but it's being called the "Andrews Matrix" by those who have taken to it. It's Dolby Pro-logic compatible, and unwraps stereo source quite nicely for inclusion in motion picture mixing. I won't go into it here, but it's another FINE example of how matrix thinking gets you solutions to problems you didn't know you even had! :wink:
Keith
But essentially yes. Now, -what was the question again? :wink:
As far as mono LF (a BIG deal in the days of stereo vinyl cutting!) recent papers published in ambisonics are suggesting that actually INCREASING the LF content of the 'side' (or L-R subtractively-derived matrix channel) actually increases the sens of 'spaciousness' in most mixes or recordings. I'd have to do some digging to locate the references, but I know that I've tried it and it's a parlour trick which seems to work well until the brain realises that it's being 'faked-out'... and that different people seem to have massively different thresholds of tolerance.
But no doubt that matrix panning/matrix processing is a VERY powerful tool... If a compelling benefit presents itself, then it becomes attractive to get into and back out of a matrix, (mastering is a killer app) but most people are sadly reluctant...
I derived a logical matrix a while ago, which people are starting to use in film mixing... there's nothing really complicated about it, but it's being called the "Andrews Matrix" by those who have taken to it. It's Dolby Pro-logic compatible, and unwraps stereo source quite nicely for inclusion in motion picture mixing. I won't go into it here, but it's another FINE example of how matrix thinking gets you solutions to problems you didn't know you even had! :wink:
Keith
