How does the signal flow in a console work in regards to panning?

[Tillmann]

I'm trying to plan out a summing mixer and ran into the issue of panning.  I have a couple of card options for the make-up amplifier (Neumann V475, Tele V675, API style makeup amp and then dual 325's as the post stereo-pot booster amps), all of which are ACN's, to my knowledge.  Anyhoo, after much searching in the meta, search function, google, wikipedia, and even looking at my studer 169 manual block diagrams and schematics, I still can't seem to get a grasp on how panning works as far as signal flow is concerned.  Also, a couple of technical questions have surfaced. 

First of all, where in the signal chain of a channel strip in a console is the pan pot placed and where does the signal go thereafter?  It would seem to me, using a simplified example of a small console that doesn't have "sub bussing" (don't know the technical term, but when say on a 48 channel console, channels 1-8 are summed to separate L-R summing amps, and then the next 8 to their own summing amps, and so on, requiring in the case of a 48 channel console 6 pairs of L-R amps that are then bussed to a pair of L-R summing amps which are your final main stereo buss amps) that the pan pot would be the very last thing in the chain, even after the fader.  Is that true?

This brings the question, how does a linear taper pot work as far as signal flow and physical properties- meaning, signal goes out the channel into the pot, and the pot determines the amount of level that goes left/right.  Full left panning means that the left side gets all the signal and the right side none and vice versa.  Center detent means that it goes to both sides equally.  So the signal from the channel is basically sort of variably "split" by the pan pot and is then sent to the main left bus and main right bus at varying levels depending on how you have it panned, which then finally go to the main left and right summing amps? 

If that is the case, how does the linear taper pot work physically?  It's obviously altering the impedance like any pot, but what is the result in the signal level and voltage in a left/right comparison.  For example lets say that we have a pot that pans from center to left 10 and center to right 10.  The number 10 is used just as a relative level and does not reflect any real unit of measurement.  If you pan the kick drum center, then each output of the pot will receive a value of 5 at it's output. If you pan the guitar track a bit left of center, say to 7.5, the left output on the pot will send 7.5 and the right side 2.5.  If that is the case, what is physically going on in the pot?  When the left side is at 7.5 from a center detent of 5, has the voltage reading gone up on that side or down?  How about the impedance?  Then, I suppose I can assume that the opposite is true for the right side in this instance?

I feel like I am very close to understanding this, but need this bit of info to be able to conceive the box.  Anyone have some explanations or educational resources available?

Cheers,

T

 

[keefaz]

I see the pan pot as a double attenuator, in the case of single gang pot used as pan pot, the wiper is grounded, in the case of double gang pot, the CCW leg of one gang and the CW leg of the other are both connected to ground. In each case, that forms an attenuator for each channel. The linear law is approximate, ideally I think the resistance curves should be cos and sin curves

 

[jsteiger]

Here is what I did for my API console when rebuilding. To be clear, it's not done yet...one of these years.

I found this voltage power ratio chart from NYDave during my searching. I used it as a guide.
Dave shows -3.38db for both outputs when at 50% rotation.

I ended up with dual linear 10K pots that are slugged with a 2.21K resistor from input to wiper on each pot. To be clear, there also is a 4.99K resistor that hangs from each output to ground immediately after the panpots. I think this is partially to eliminate drastic loading when switching the 16 buses in and out.


I think I read somewhere from Geoff Tanner that Neve used dual 10K linears slugged with 5K1 resistors.

This works pretty well in my circuit. I am closer to -4.25db when at the center detent. I went for -3db first but the the -4.25db seemed to "move" less when panning and monitoring in mono, at least to my ears. This configuration may not work perfectly in your circuit but may get you close. My results are not spot on to NYDaves chart but very close. Close enough for sure.

Your slug values may be different. This will give you a starting point.

Cheers, Jeff

 

[nielsk]

Typically the channel fader output goes to a buffer amp that feeds the post fader sends and pan pot(or pan circuit, you can get pretty fancy, but the most basic is a dual pot, one log and the other reverse log). Then the pan pot output goes to a L&R buffer to feed the mix buss.
You are referring to what is often called "bucket mixing" where the channels in each bucket (usually 8) are summed, then the bucket outputs are summed for the final mix. It is a trade-off between the net noise gain of every channel you add to a mix buss. This works quite well, and if you use balanced mix busses results in the lowest noise floor of most any approach.

 

[JohnRoberts]

This is too complex for a quick simple answer, but IIRC it has been well discussed here before.

#1 what- there is not unanimous agreement regarding the ideal pan law. -6 dB at center for constant voltage, -3dB for constant power, -4.5dB as a split the difference compromise (I personally used -3dB but opinions vary).

#2 how- with a single linear pot the typical topology has 10dB insertion loss which must be made up in or post summing bus. This too has been widely discussed. Optimal for bus noise would be a dual custom taper potentiometer that delivers the desired curve with minimal insertion loss. Without much agreement over a single pan law, standard off the shelf parts will not to be found.

For a very high performance DIY approach I would consider executing the pan and level with say a 13 position wafer switch. Impedances and insertion loss could be optimized. Instead of padding down the signal and boosting it back up in bus, or post bus gain stage, increase the impedance being sent to the bus to attenuate signal while simultaneously reducing the noise gan or make up gain required by the bus.

Or not...

JR

 

[SSLtech]

John's post sums it up.

Problem with choosing a pan law which is "right" (-3dB or -6dB) is that it's also 'wrong' for certain applications.

-3dB is 'right' for equal power summing, but it's 'wrong' if you hit the mono button (signal rises 3dB in the middle). -6dB is perfect for downstream mono summing, but 'wrong' (signal dips 3dB in the middle) when listening in stereo.

-4.5dB -or thereabouts- is a common compromise. 3dB in a moving pan is noticeable, but "1-point-something" dB is rather less noticeable but -as you can see- reasonable minds can differ in opinion as to what is better.

Be mindful that if the pan is left static, then all discussion is completely academic. The pan law ONLY affects the apparent signal loudness of an object as it MOVES through the panoramic range.

A dual linear pot will give you 6dB down at center. adding some curve-bending resistors can get you very close to the ideal sin/cosine 3dB down at center law, or anywhere in between, at the expense of a slight insertion loss.

John covered the essential points, I've just added this to add some of my own thoughts.

Keith