Phase issues with an AUX send bus

I was thinking today, why not do parallel compression by just sending certain drums to an AUX bus with a compressor on it? The main signals still route to the Master bus, and the AUX bus goes there, too. Just blend them as desired, right?

Wrong, at least in my current state of Sonar (latest), which I'm hoping doesn't have problems. Phase issues galore result. I put Waves C6, Boost, Waves C4 and some other plugins in the AUX bus and all did the same. In most instances, the phase issue then remained with the plugin removed, as if the whole bus channel was irreversibly kicked into a latency mistake.

Shouldn't plugin latency compensation enjoy victory over any and all routings I choose?

Remember, this is when I do this with AUX sends, versus routing the entire drum track to the bus. When I do the latter, it is fine, but I obviously have no clean signal remaining...

Hope this can be explained, versus it being my system on the blink.

The experiment seems to be nothing more commonplace in structure than and AUX send bus setup, yet when I use verbs or other effects on the bus there are no phase issues. Do the dynamics plugins pose a challenge, or is my latency comp on the blink, plain and simple?

Thanks!

If you like having all of your drums routed to a separate drum bus (like I do), you can also do what you want by adding a separate compression bus then use SENDS on the drum tracks you want to compress. Route the output of the compression bus to your drum bus to retain control overall.
 
Dan
 

Anderton
Try toggling the PDC button while SONAR is stopped. 

Thank you, Craig.
 
Here's what I've got:
 
I setup an AUX bus for receiving ANY channels (not necessarily drums) and, prior to placing an effect in the bus, engage an AUX Send in a track.  The meters for the AUX now start showing the sent signal, and that is additive in the final mix and is fine in terms of phase.  When I add an effect such as a verb or any other commonly-used aux effect, things seem fine.
 
When I add a compressor (like Waves L6 and even the Cake limiter Boost11) the phase issues comes in immediately.
 
NOW (and important), when I deselect the effect itself in the bus, the phase cancellation stops.  BUT, if I deselect the entire bus effect master button, the phase issue returns, but with slightly different timing (sound is a bit different, implying different lag).
 
It's almost as if certain plugins are too complex for Sonar to get the delay compensation right, as if they are too intensive, processor-wise.  Then the bit about toggling the master aux effects button and having the phase issue return in modified fashion (while the effect is still in the aux box, but off) is entirely puzzling.
 
Samplitude Pro X does the same phase issue in the same scenario when the plugin is on, but cancels the problem upon ANY disabling action, whether the plugin itself or the entire plugin master in the AUX channel.
 
Puzzling stuff.  Hope it's not just my system. Reinstalled Sonar just to make sure.  No change.
 
Thanks for the interest and response, all!

It seems like pdc is falling down on the job. As a workaround / test I would try adding the same effects you have on the aux buss to the dry buss and set them so they don't process the sound.
 
Rocky

bluzdog
It seems like pdc is falling down on the job. As a workaround / test I would try adding the same effects you have on the aux buss to the dry buss and set them so they don't process the sound.
 
Rocky

 
Yes, and I think it is just with complex plugins.  If you guys don't have Waves multibands, try the Sonitus Multiband. It does the same thing.  Izotope Dynamics works.  Sonitus Compressor works.  I guess it's just the heavy stuff that flogs it.  That's interesting and a bit disappointing, because, if I'm correct, this means that parametric compressors, which can really be tweaked to make drums romp, may not be usable on my system for parallel compression.
 
I'll be curious to hear what others find.  You can use any single audio track for input.  Just create a bus, put a multiband compressor in there (even Boost11 -- a simple limiter -- did it), and see what you hear.
 
Thank you, community!

I tested with Perfect Space and the new L-Phase Multiband. At Medium or High Precision settings, the L-Phase Multiband induces so much PDC that it will manifest as a significant timing sync error - not just a phase error - if PDC isn't working. But it worked fine.
 
The fact that you're only getting a phase error makes me wonder if the Waves plugins are misreporting PDC by a few samples or doing something else to the signal that manifests as a phase error when used in parallel.
 
The only time I ever heard a phase error due to PDC was with Addictive Drums sending to parallel compression because SONAR wasn't properly compensating the 64 samples of PDC that AD's own internal FX induce. This has since been fixed.

I tested with Sonitus Multiband. I think any oddity you hear with that plugin is due to the way it alters the sound even when it's nominally set to have no effect. I tested in another project I have that measures PDC, and Sonitus Multiband doesn't induce PDC unless the Limiter is engaged; then it introduces 64 samples of delay. But when using it as a parallel compressor, it causes the same muffling/phasing of the signal, whether the Limiter is engaged or not. My guess is that this is because it's not a Linear Phase processor.
 
If you A/B the Sonitus Compressor output with the dry signal, it sounds pretty much 'unmolested', but the two cannot be made to null even if you add a sample delay (e.g. Channel Tools) to try to compensate for faulty PDC. No matter what you do, there is always some incomplete nulling at some frequencies.
 
It makes sense that parallel compression requires using a linear phase compressor to avoid this. I don't know if those Waves plugs are linear phase...?
 
EDIT:  I confirmed this hypothesis by setting the L-Phase MB precision to non-linear. It has an effect similar to Sonitus MB even when it's not doing anything - i.e. bands are added but with no parameters altered. Invert the Aux track output and set the LP MB precision to Low/Med/High, and it nulls perfectly with the dry signal.

Glad to help. It was a learning experience for me, too. Makes me wonder how many 'golden ears' are out there right now rockin' to the phase error of parallel compression using non-linear plugins. 

I've never really used parallel compression 'in anger'; only played around with it. My mixing efforts seldom reach that level of complexity. I try not to beat myself up about it. 
 
You can start by Googling 'parallel compression site:forum.cakewalk.com' to find discussions on the forum. Some of the hits i got from this reference external articles/tutorials on the subject.

bitflipper
Using multi-band compressors in parallel is asking for trouble. I wouldn't recommend it, for the very reasons you cite. Each of the nine crossover filters will introduce phase shifts, even with all controls zeroed. AFAIK there is no way to compensate for crossover distortion via PDC.
 
Parallel compression on drum busses is a common effect, but it'll always be broadband compression. If you're using a multi-band in order to avoid issues with the compressor being overly-sensitive to low frequencies, use a compressor with a sidechain filter instead, since filtering the sidechain will not result in phase shifts.
 

Bit, what are you referring to as parallel compression in this example?  Do you mean two multi-bands in tandem?  I was not doing that.  My aux bus just acts like it can't handle some of the multi-band plugins, although the new LP MB and many other compressors work.  I'm curious what the core functional culprit really is.  If it is not the center timing of the signal, maybe indeed it is complex phase issues (timing, essentially) across the multibands which a-phase with the original signal in the same output bus.  Puzzling.  Maybe you meant putting a multi-band-processed signal in parallel with its source.  That would make sense.
 
Thanks for prior and any future dialog.  I'm the type that likes to get down to the phenomenon in play on this kind of stuff, but I don't have the detailed savvy to call this one!