Panning with the Law
- Thread starter analag
- Start date
Help Support GroupDIY Audio Forum:
SSLtech
Well-known member
What's the function of the top & bottom op-amps?
...and the two right-going resistors?
Curious, that's all.
Keith
...and the two right-going resistors?
Curious, that's all.
Keith
NewYorkDave
Well-known member
It looks like a take on Doug Self's (patented) idea of using negative impedance converters to bend the law of the pots. To wit:
The negative impedance circuitry can provide -3dB at 50% and good end points but is it constant power everywhere else? I'm too lazy to do the math...
In the real world if your going to throw a dual pot at what most designers do with a single, you would just dial in the taper you need. I appreciate for one-off or low volume production we are constrained to work with off the shelf parts.
This approach will be obscure and unusual, but few people appreciate the clever details under the hood. For a high performance design I am a little concerned that the impedance seen by the bus is varying with pan position and to a lesser extent additional noise contribution from extra added opamps. At least in Self's version the bus sees a low constant source impedance so you can properly forward reference the channel ground.
------
Has anybody considered using something like a 12 pos wafer switch and just switching in precision values for pan law? This could deliver a lower bus noise gain when channel is panned down. An extra wafer would be needed to maintain proper ground forward reference impedance, so this is not cheap or easy, but it could be quite good.
A little food for thought. I'm sure this has been done, perhaps with less switch positions on old consoles.
-----
If one was serious about addressing the price is no object tweaky console market (not me) I suspect one could build any law they want using 02x01 smt resistors on a substrate with standard pot wiper to tap into the precision resistor divider. A single pot section design could be populated with an infinite variety of tapers as needed. This would be more precision than the best screened parts.
JR
In the real world if your going to throw a dual pot at what most designers do with a single, you would just dial in the taper you need. I appreciate for one-off or low volume production we are constrained to work with off the shelf parts.
This approach will be obscure and unusual, but few people appreciate the clever details under the hood. For a high performance design I am a little concerned that the impedance seen by the bus is varying with pan position and to a lesser extent additional noise contribution from extra added opamps. At least in Self's version the bus sees a low constant source impedance so you can properly forward reference the channel ground.
------
Has anybody considered using something like a 12 pos wafer switch and just switching in precision values for pan law? This could deliver a lower bus noise gain when channel is panned down. An extra wafer would be needed to maintain proper ground forward reference impedance, so this is not cheap or easy, but it could be quite good.
A little food for thought. I'm sure this has been done, perhaps with less switch positions on old consoles.
-----
If one was serious about addressing the price is no object tweaky console market (not me) I suspect one could build any law they want using 02x01 smt resistors on a substrate with standard pot wiper to tap into the precision resistor divider. A single pot section design could be populated with an infinite variety of tapers as needed. This would be more precision than the best screened parts.
JR
SSLtech
Well-known member
Ah... I see it now!
-Modifying the law that way, if you CAN get constant power, you could eliminate the need for sin/cos custom tapers... -I wonder...
I wonder...
There's something different about where the feedback occurs around the first op-amp: The R which feeds the top of the cross-wired pot gangs looks like it's effectively removed from having the dividing effect which it does in Doug Self's version, and -as I see it- the negative impedance converter actually 'forces' a slight level change there at different wiper positions... I'm not sure what changing that feedback point does in terms of advancing the design, but I'm certainly eager to learn!
Hey Rowan, -can you run some sim. data for the curve? -Can you get a true sin/cos approximation over the wiper rotation? -let me dig out what curve I came up with last time...
Keith
-Modifying the law that way, if you CAN get constant power, you could eliminate the need for sin/cos custom tapers... -I wonder...
I wonder...
I'll have to try and run that through an excel spreadsheet or something similar... I did that trying to match a sin/cos for a soundfield decoder, and got close... -ish... but never spot on. -Good enough for most purposes though, but if this can get me even closer, -now there'd be a thing!
There's something different about where the feedback occurs around the first op-amp: The R which feeds the top of the cross-wired pot gangs looks like it's effectively removed from having the dividing effect which it does in Doug Self's version, and -as I see it- the negative impedance converter actually 'forces' a slight level change there at different wiper positions... I'm not sure what changing that feedback point does in terms of advancing the design, but I'm certainly eager to learn!
Hey Rowan, -can you run some sim. data for the curve? -Can you get a true sin/cos approximation over the wiper rotation? -let me dig out what curve I came up with last time...
Keith
SSLtech
Well-known member
Ah... -here it is!
Keith
Keith
![Soldering Iron Kit, 120W LED Digital Advanced Solder Iron Soldering Gun kit, 110V Welding Tools, Smart Temperature Control [356℉-932℉], Extra 5pcs Tips, Auto Sleep, Temp Calibration, Orange](https://m.media-amazon.com/images/I/51sFKu9SdeL._SL500_.jpg)
[quote author="JohnRoberts"]
Has anybody considered using something like a 12 pos wafer switch and just switching in precision values for pan law? This could deliver a lower bus noise gain when channel is panned down. An extra wafer would be needed to maintain proper ground forward reference impedance, so this is not cheap or easy, but it could be quite good.
A little food for thought. I'm sure this has been done, perhaps with less switch positions on old consoles.
JR[/quote]
Altec and Langevin both built 22 position T and ladder type 600 ohm panning pots. Both in rotary and straight-line, AND in two output and L / C / R output.
Has anybody considered using something like a 12 pos wafer switch and just switching in precision values for pan law? This could deliver a lower bus noise gain when channel is panned down. An extra wafer would be needed to maintain proper ground forward reference impedance, so this is not cheap or easy, but it could be quite good.
A little food for thought. I'm sure this has been done, perhaps with less switch positions on old consoles.
JR[/quote]
Altec and Langevin both built 22 position T and ladder type 600 ohm panning pots. Both in rotary and straight-line, AND in two output and L / C / R output.
NewYorkDave
Well-known member
http://www.groupdiy.com/index.php?topic=120047#120047
Other discussions on panpots:
http://www.groupdiy.com/index.php?topic=251485
http://www.groupdiy.com/index.php?topic=46036
http://www.groupdiy.com/index.php?topic=214767
http://www.groupdiy.com/index.php?topic=201390
Other discussions on panpots:
http://www.groupdiy.com/index.php?topic=251485
http://www.groupdiy.com/index.php?topic=46036
http://www.groupdiy.com/index.php?topic=214767
http://www.groupdiy.com/index.php?topic=201390
![[silent:arts]](https://cdn.imagearchive.com/groupdiy/data/avatars/m/0/172.jpg?1620371656){kind=avatar}
[silent:arts]
Well-known member
Neumann. somewhere I must have a schematic of the Pan Pot module. will have a look tomorrow. works - just great.
NewYorkDave
Well-known member
As you probably know, Bob Orban's "Notes on Panpots" is the standard reference for the single-pot pan approach.
Here's a practical example, used as part of an Ambisonic decoder:
The math of stepped panpots is covered in "Precision Pan Pots" by Richard Cabot.
I wish I could post the Orban and Cabot papers but the AES wouldn't like that :roll:
[silent:arts]
Well-known member
Wayne: Not to reopen an old debate but constant voltage (-6dB law) is not constant power (-3dB). I believe mono compatibility is not the primary issue but perceived loudness while panning L to R. There seem to several differing POV on this.
YMMV
JR
EDIT: this is confused by panned mono being a coherent signal, but whether that coherence is maintained all the way to the listening position is suspect.
YMMV
JR
EDIT: this is confused by panned mono being a coherent signal, but whether that coherence is maintained all the way to the listening position is suspect.
Your math is impeccable, and it demonstrated the 3dB difference in total power between pan centered and hard L/R.
My target for constant loudness pans is total power from L+R adds up to unity at all pan positions. For example, centered -3dB + -3dB = 0dB, hard panned 0dB + -infinity= 0dB.
There is plenty of debate about coherent summation vs. incoherent combining as this will not add up to one when summed to mono. Some, have argued for -6dB or even compromised with -4.5dB law to balance perceived levels with mono compatibility.
This is pretty much a subjective or human factors issue so there isn't really a right or wrong but common practices and AFAIK defacto standards favor the constant power law over constant voltage law.
It's been a while since I designed a console so don't know if fashions have changed and there's probably examples of all three to be found. I must concede I don't spend much time looking at other designs, especially last couple decades.
JR
My target for constant loudness pans is total power from L+R adds up to unity at all pan positions. For example, centered -3dB + -3dB = 0dB, hard panned 0dB + -infinity= 0dB.
There is plenty of debate about coherent summation vs. incoherent combining as this will not add up to one when summed to mono. Some, have argued for -6dB or even compromised with -4.5dB law to balance perceived levels with mono compatibility.
This is pretty much a subjective or human factors issue so there isn't really a right or wrong but common practices and AFAIK defacto standards favor the constant power law over constant voltage law.
It's been a while since I designed a console so don't know if fashions have changed and there's probably examples of all three to be found. I must concede I don't spend much time looking at other designs, especially last couple decades.
JR
[quote author="mediatechnology"]Thanks John.
Doesn't this one meet that definition: L+R meaning "mono?"[/quote]
My phrasing was sloppy.. What I mean is Left power + Right power adds up to unity for all pan positions. As I have stated summed mono is NOT constant power and thus the basis of much discussion. There is no solution for both unless you could somehow game phase across the full bandpass to maintain constant voltage and power in mono but that is heavy lifting even with DSP and may cause other artifacts worse than the benefit.
I'm inclined to design for comfortable mixing to stereo, and review the results in mono for problems after the fact. These days mixes are probably getting messed up by mixing for MP3 compatibility, or whatever is the common codec in use. Mono compatibility remains a factor but probably less so with more and more stereo delivery channels.
Of course I could be wrong.
JR
Doesn't this one meet that definition: L+R meaning "mono?"[/quote]
My phrasing was sloppy.. What I mean is Left power + Right power adds up to unity for all pan positions. As I have stated summed mono is NOT constant power and thus the basis of much discussion. There is no solution for both unless you could somehow game phase across the full bandpass to maintain constant voltage and power in mono but that is heavy lifting even with DSP and may cause other artifacts worse than the benefit.
I'm inclined to design for comfortable mixing to stereo, and review the results in mono for problems after the fact. These days mixes are probably getting messed up by mixing for MP3 compatibility, or whatever is the common codec in use. Mono compatibility remains a factor but probably less so with more and more stereo delivery channels.
Of course I could be wrong.
JR
SSLtech
Well-known member
I took a look at some of the posted links, and -wow!- it seems that people -including me!- tend to get pretty worked up about pan laws... -who knew? :wink:
But yes, Soundcraft, SSL and may others used "minus-4-point-something" pan laws, as a reasonable compromise between 1-channel-summed-mono and discrete-2-channel playback. I can provide a photograph of a Harrison module front panel with switchable pan laws of -3dB and -6dB, and the Amek M3000 also had that.
Here at work we have a Harrison MPC-3D console, which had LOTS of panning options (DSP) and the law in all cases (left-to-right, left-to-center and center-to right, front-to-back, front-to-mid and mid-to-back etc.) is -3dB, but then it's a MOVIE console, for multi-channel playback, and there is NO downmixing, or LT/RT or mono monitoring facility ANYWHERE on the console... That may also be true of most or even all film dubbing consoles, I can't say with absolute certainty. (by the way, we just had some photos taken of that room for the front cover of 'Mix' magazine, FWIW...) and I can say that 3dB works VERY well for multichannel film.
For music MOST panning is static, and levels will often be set or tweaked after final positioning, so the law is not TOO critical in the final analysis.
However, TRUE sin/cos ("equal power") is the 'holy grail' of most ambisonic or 'B-format' decoders or controllers.... and pretty light work for DSP applications.
Rowan, -you ever get some sim sweeps done yet?
Keith
But yes, Soundcraft, SSL and may others used "minus-4-point-something" pan laws, as a reasonable compromise between 1-channel-summed-mono and discrete-2-channel playback. I can provide a photograph of a Harrison module front panel with switchable pan laws of -3dB and -6dB, and the Amek M3000 also had that.
Here at work we have a Harrison MPC-3D console, which had LOTS of panning options (DSP) and the law in all cases (left-to-right, left-to-center and center-to right, front-to-back, front-to-mid and mid-to-back etc.) is -3dB, but then it's a MOVIE console, for multi-channel playback, and there is NO downmixing, or LT/RT or mono monitoring facility ANYWHERE on the console... That may also be true of most or even all film dubbing consoles, I can't say with absolute certainty. (by the way, we just had some photos taken of that room for the front cover of 'Mix' magazine, FWIW...) and I can say that 3dB works VERY well for multichannel film.
For music MOST panning is static, and levels will often be set or tweaked after final positioning, so the law is not TOO critical in the final analysis.
However, TRUE sin/cos ("equal power") is the 'holy grail' of most ambisonic or 'B-format' decoders or controllers.... and pretty light work for DSP applications.
Rowan, -you ever get some sim sweeps done yet?
Keith
