Man this happens once in a while. Tracks fail to respond to widget changes. Todays scenario:
 
Drummer  was adding some crash fills.
Recorded the 2 crashes using the stereo overheads.
Drummer comes into control room and sits down to listen.
Asked to have crash hits panned about 40% left and 40% right.
So - we have 3 hits in 4 different locations in the song.
1st hit is the 21".
2nd hit is the 17"
3rd hit is the 21".
 
So I insert a pan envelope on the track.
Add nodes at proper locations for each hit.
Drag envelope up and down to required percentages.
So far so good.
 
Hit PLAYBACK.
Console pan widget clearly follows pan envelope but the audio doesn't seem to want to oblige.
 
1st hit sounds pretty close to 30% left.
2nd hit sounds pretty close to 30 LEFT too.
Hmmm.
3rd hit is darned close to 30% left.
 
 
Drummer says, hey - the pan thingy's moving - what kinda crap you using here ?
Me all red faced . . . . murmur "well I've seen tracks fail to respond before". Nobody knows why they go dead like this.
 
So I drag the envelope to 100% left on hit 1, 100 right on hit 2 and back to 100% left on hit 3.
 
Hit playback.
 
No love.
Pan widget does as expected.
Audio stays in left channel.
 
Lord why am I still around.
 
 

When using a pan control on a stereo track, how much movement you hear in the phantom image depends on how well-separated the instruments are in the track: the better-separated, the less movement there'll be. The amount of perceived shift can range anywhere from full to weak to imperceptible. 
 
Because of this inherent unpredictability, I think of the pan slider and pan automation as mono-only effects and almost never apply them to stereo tracks.
 
 
If you want to pan a stereo track use the Channel Tools plugin instead. Better still, split that stereo track into two mono tracks and pan away.
 
 
 

Dave, if I really assume Sonar works  as just you have described, can you please explain how that could possibly be acceptable?
 
On a analog console, (at least every last one I have ever used since 1980) if you pan a track hard right, you will hear silence if the left. Conversely in the opposite condition.
 
On that same desk, If  I have a stereo bed on say channels 1 and 2 with 1 hard left and 2 hard right and change channel 1 to hard right, the left channel will be silent even though I am feeding a stereo bed (excepting adjacent channel crosstalk). IOW, actual panning should follow relative pan position (sound stage) position, not room mic setup conditions, mostly.  Otherwise, what good would they (pan pots) be for any simulated stereo purpose except hard left and hard right on a stereo pair (assuming that's what you had, I suppose) which should then ONLY give the maximum separation that was achieved with the room mic setup when the recording took place? Hope that makes sense.
 
Or am I all wet ?

On a analog console, (at least every last one I have ever used since 1980) if you pan a track hard right, you will hear silence if the left. Conversely in the opposite condition.

 
Yes - because it's a mono track, not stereo, and Sonar behaves exactly as it should with mono tracks and stereo tracks
 

On that same desk, If  I have a stereo bed on say channels 1 and 2 with 1 hard left and 2 hard right and change channel 1 to hard right, the left channel will be silent even though I am feeding a stereo bed (excepting adjacent channel crosstalk)

 
Yes - because you're still dealing with 2 mono tracks, even though your source material is "stereo", it's still hitting your desk as 2 different signals and you will quite rightly fade from full on to full off as you sweep from one extreme to the other.
 
Try it on a hardware stereo input - you'll hear what you're hearing in Sonar - the pan control when dealing with stereo tracks isn't really pan at all, it's a BALANCE control.
 
Like Dave says, trying this will cause very unpredictable results, and it'll be the same in any DAW - it's not a Sonar problem.

Thanks Jonesey. I believe you and Dave. . . . But  . . . I am having a hard time wrapping my head around this concept or perhaps I should say DAW phenomenon . 
 
I am also trying to picture the sort of  "stereo" hardware input you are referring to that might have a pan control.
 
As far as unpredictability, how is that possible ? Varying  signal strength to a output is varying signal strength to a output from what I remember. And how could you guys have possibly discovered your concepts using analog consoles ? 
 
So basically, you are both saying that a pan knob is not really a pan (panorama) control. Here's a idea. If it is designed to work as you say it is (in a DAW) how about if the bloody slider was just greyed out in a true stereo track so that any other nitwit like myself could recognize that it is useless in that instance.

Just to make sure that my last few brain cells didn't die off in the hot sun the other day, I looked your hypothesis up on Wiki.
Here

On all those classic consoles you've been using for the past 30 years, I'll wager that few of them even had stereo channels, or had only a few. This proliferation of stereo-everything is a fairly new phenomenon. (And it's killing the width of mixes, but that's another rant.)
 
Picture the worst-case scenario: a stereo track where an instrument is only in the left channel. When you pan the track toward the right, the volume of the instrument decreases until it's completely silent. But it never "moved" in the panorama. And you'd be wondering why it wasn't panning.
 
Now imagine a best-case scenario: a stereo track in which the instrument is equally represented in both channels. Now, when you move the pan knob the instrument's perceived location does indeed change.
 
Bottom line: the more mono-ish the stereo track is, the better it pans. The more separated the instruments, the less perceived panoramic movement you get.
 

bitflipper
On all those classic consoles you've been using for the past 30 years, I'll wager that few of them even had stereo channels, or had only a few. This proliferation of stereo-everything is a fairly new phenomenon. (And it's killing the width of mixes, but that's another rant.)
 
Picture the worst-case scenario: a stereo track where an instrument is only in the left channel. When you pan the track toward the right, the volume of the instrument decreases until it's completely silent. But it never "moved" in the panorama. And you'd be wondering why it wasn't panning.
 
Now imagine a best-case scenario: a stereo track in which the instrument is equally represented in both channels. Now, when you move the pan knob the instrument's perceived location does indeed change.
 
Bottom line: the more mono-ish the stereo track is, the better it pans. The more separated the instruments, the less perceived panoramic movement you get.
 

Ok Dave, I'll accept your explanation as to what they have engineered for a DAW. I ain't no physicist.
 But if I  am correctly understanding these explanations ,  how has the specific "panoramic" location of a instrument on a soundstage become synonymous  with stereo imaging which in my mind, has always been more a function of aural perception based on the combined effects of instrument position, room reverberation, echos and the listener's spatial position? I have always understood these to be two fundamentally different concepts though intertwined. 
 
Tomorrow I intend to spend a good bit of time with EWQLSO which we all know was recorded in true stereo with multiple mics in front, center and rear.
 
 
I intend with test number 1 to output the signal (Lynx direct) into the analog console with a pair of channels panned 8&4 o'clock which will give me the widest stereo separation possible. I believe that as I use the Kompakt or EW Play player widget to pan  say a trumpet from it's normally staged position to left (audience perception), it will pan and volume will remain somewhat consistent. Maybe 3-4 dB louder at it's recorded position in the hall but not horribly perceptible changes in volume.
 
Test 2 will be done  with the EW output printed in Sonar with a pair of convertor channels patched directly into the monitors. I will use the Sonar track  pan slider.
 
Perhaps I will measure the dB output levels at engineer's position for comparison.

Those EWQL samples were probably recorded via microphone positions much further away from the source than your drum overheads were. Consequently, they're going to be more blurred left-to-right, and probably take panning better. Try your experiments with a violin section, assuming it's conventionally positioned to the left in the panorama. Compare the difference between using the pan knob versus Channel Tools. Depending on the sample, the difference can be dramatic.

I've always thought of it this way:
The pan knob on a stereo channel decides a balance of what's going into the channel strip. Turning it to the left doesn't bring what's on the right to the left, it instead seems to go back to the input stage, increase the volume of the left signal and decrease the volume of the right signal.
 
For actual panning of each side on a stereo channel ,you want to insert plugin Cakewlak Channel Tools, which has a pan fader for each side. Like ProTools, if you've ever used that. It definitely makes much more sense to have TWO pan controls for a stereo channel, like you would on a desk.