edit: please see this link: http://line6.com/support/thread/88099

...

I don't know if many of you have experienced this or maybe you have experienced it but just didn't realize it, but it seems there is a NECESSITY to have a variable delay to be added to the Mixer Channels.  This delay should be very small - between 0 and 40ms, and it should be adjustable in very fine increments, such as 0.1 ms.  This would be useful to synchronize Channels A and B to avoid comb filtering.

OR create a new phase-adjustment Delay effect.  This effect would always have a mix of 100% and a feedback of 0% but a variable delay time with min and max and increments specified above.  These could be added to one channel to adjust the signal accordingly.  It should consume very little DSP, and be very easy to implement.

There are three noticeable cases where comb filtering between channels is currently unavoidable:

1) Setting Input 1: Guitar, Input 2: Guitar/Same - To be honest, this should be corrected before the signals enter the signal chain.  Using these options and placing a mono-summing effect before the path split would still cause comb filtering and a phasing sound, even with the additional delay feature on the Mixer.  However, until that is fixed, the Mixer solution would remedy this problem for patches without mono-summing effects in front of the path split.

2) Using different cab/mic settings between Channel A and Channel B -  This is to be expected - different cabs and mic positions will result in minute differences in timing from when the sound waves hit the microphone.  Thus, they have different timing signatures and when mixed together are unlikely to be in-phase, producing comb filtering.  Some combinations are more in-phase than others, but it's likely that they're all out-of-sync, even if to a small degree.

3) Using different/more effects in one channel than the other - this is particularly evident with the EQ effects.  And this makes sense - I would imagine the algorithm to perform EQ adjustment needs a buffered sample size large enough to detect the lowest frequency available to be adjusted before the signal can be processed and output.  If we assume this is 60 Hz, that's 16.6667 ms.  That's a rather large delay in the signal and will result in comb filtering.

As a workaround, I no longer use Input 2: Guitar/Same whatsoever.  I only use cab/mic combinations for dual amp patches that sound relatively in-phase without adjustments.  In cases where I need to attempt to synchronize the two channels, I use neutral EQ effects.  They all seem to introduce slightly different delay signatures to the signal, so with enough experimentation I can find one that syncs the two channels as close as possible, but this process seriously lacks precision.  It only gets so far, never 100% there, and in many cases simply cannot get the job done.

If you feel as I do, please submit a request:

http://line6.com/company/contact/productfeedback/

Edit: Some have mentioned possibly making this adjustment automatic in a feature request.  I agree.  Some effects even change their delay value, making it impossible to be manually synced.

the pitch glide is the craziest example.  when you turn it on it's at a certain number of samples, but as you use it, the exact delay changes.  so if it's at unison and there's 20 sample delay, then you move it up an octave and back to unison, now the delay might be like 30 samples.

the mid-focus eq also seems to have slight variation of its delay as you move the LP freq and Q.

i think the best case fix would be a system setting that can be set either to auto or manual or mixed.  auto would behave as you describe - the two channels are automatically synced.  manual is straight sample-based delays adjustible on either channel as i originally described.  mixed would automatically update the values, but you could set it to be always at a certain amount out of sync.

The way auto would work is it would sum the total delay in samples for each channel in real time depending on the effect, amp, and cab/mic choices.  Whichever channel had the least latency would have latency added to match the other channel.  This would occur in real-time, so that toggling effects on/off, or changing parameters would not cause the channels to become out of sync.

i think all modes would be important because when mixing certain cab/mics, some people may find the comb filter effect to be pleasing, reducing some harsh high-end.  So manual/mixed would be as valuable as auto to these people.  and some people may even want to toggle effects on/off to engage/disengage the comb filter, so they'd prefer manual over mixed.