Printing Stems: Phase issues

[alexdingley]

Hey all — 

Just finished a 6 Song EP, and thought it would be smart to print stereo stems of each mix. I mixed on my Toft ATB-16 and printed to an external stereo disk recorder... those mixes all sound great and worked just fine for mastering... they had a gentle compressor on them (Elysia Xpressor) and I was happy with the sound / they collapse to mono just fine when checked.

Then I printed all the mixes back through groups in the console, back into PT. Stereo: Drums, Keys, BG Vox, Guitars, Bass, Ld Vox (even though those last two are mono)... each tune ended up with between 5-7 Stereo Stems. I did a bunch of muting and ensured that the time-based FX were printed into each stem, so that I would have the luxury of printing instrumentals later, etc...

I decided to do some experiments with the idea of Stem-Mastering... and sadly, I found that the Unity-Sum of my stereo stem sessions (just all stem-faders sitting at unity, no FX added to any stem) was NOT collapsing to mono as well as my original mixes. each song, across the board was losing some definition in the mono check... So I basically have a bunch of useless stem sessions. I mean, it wasn't like they canceled completely out... they were just weren't as solid-sounding in mono as the original stereo mixes that I'd made.

Trying to figure out how that happened, so that I can know for the future / avoid issues.

I checked my TRS->DB25 snakes (the ones that fed from my 8-busses back to the DAW input) and none of those have any polarity inversions. I also checked my mix channels as I was printing the stems... no phase-flips were engaged. All time-based FX were plug-ins through PT...

[Magnetic Services]

Did you turn off the compressor before printing stems? It can react much differently to stems than to the whole mix.

Also, how'd'ya like that Toft?

[TapeOpLarry]

Beyond the (usually very true) L/R bus compressor point (if you are printing stems one by one via the L/R bus....)

For stem remixing everything has to be exactly the same, but you used different output and input paths to capture the stems. Therefore, why do you expect the sound to be exactly the same?
Do the outputs of the busses have the exact same circuits/transformers/servo circuits as the output of the L/R bus?
How do the input convertors and electronics of the stereo disc recorder (why not record back into Pro Tools?) compare to the Pro Tools' inputs?
Are all the Pro Tools inputs via the exact same converters? Are there several converters being used together? This is phase hell.

I've heard a number of really good engineers complain about not being able to rebuild mixes with stems. I do it all the time but I am very electronically careful. You need to be!

[Nick Sevilla]

I'm with Larry,

Stems MUST be printed through EXACTLY the same mixdown path, without bypassing anything.

I highly suggest you do those stems again. Throw out the obviously bad ones, they are useless.

Use the mixdown recorder, then import the stems back into PT afterwards.

Cheers!!!

[alexdingley]

Beyond the (usually very true) L/R bus compressor point (if you are printing stems one by one via the L/R bus....)

For stem remixing everything has to be exactly the same, but you used different output and input paths to capture the stems. Therefore, why do you expect the sound to be exactly the same?

Fair point Larry, and thank for chiming in! — my desire to print stems from my mixes did happen to come directly form the conversations you had with us during the weekend mixing workshop you hosted back in July 2016 (still use plenty of what you'd suggested, btw)... You'd talked about building a relationship with a good mastering engineer, and that providing them stems would offer some added flexibility.

So, anyway — I wasn't expecting the individual stems to sound identical because I was no longer hitting the master bus compressor that sits directly in front of my mix-down deck. I certainly expected a little tonal / dynamics difference... but I was shocked that the phase coherence would have been mucked up by pushing the mix through a bus-out, as opposed to the main outs. When I have a moment, I'll recall the mixes (obviously that will introduce a another set of differences in the mixes) and print a set of stems through the main console outs-> back into the DAW

I guess what I'm gathering here is that (and correct me if I'm wrong) you really cannot necessarily trust every different output path on a console to pass the signals "identically"... differences in the summing busses on the group paths vs the main mix path may likely introduce various issues into the signal. I'm still not sure that I understand how (barring some technical issues with the signal path I happened to choose) partial phase shifting could happen when the signal was routed out a pair of bus outs vs. a pair of main outs... ex. My "piano" stem (the piano was a soft-synth sample-based piano) lost it's presence and nearly diminished in volume dramatically when checked in mono... but the Piano doesn't cancel out, at all, when mono-checked in the regular 2-mix version. I'm eager to understand it — but I just don't quite see how that happens, at this juncture.

[alexdingley]

Did you turn off the compressor before printing stems? It can react much differently to stems than to the whole mix.

Also, how'd'ya like that Toft?

Yep, the master comp was not in the chain when pushing the stems back into the DAW. It was just the following path:

DAW Outputs (Antelope Orion32+) -> Line-Ins on the Toft -> Grouped through Toft Busses -> Toft Bus outs -> DAW Inputs (Antelope Orion32+). Thankfully I got away from my previous setup (many different I/O units from Apogee) before the mixing stage... The Apogees were nice, but I like the sound of this modern Antelope, and it runs cooler / lots of other reasons.

As for how I like the Toft?? I like it a lot — the EQ is lovely & the preamps are solid enough for my rock uses. It's not without its issues, but overall, it's been a great "project studio" board.

[Nick Sevilla]

I guess what I'm gathering here is that (and correct me if I'm wrong) you really cannot necessarily trust every different output path on a console to pass the signals "identically"... differences in the summing busses on the group paths vs the main mix path may likely introduce various issues into the signal. I'm still not sure that I understand how (barring some technical issues with the signal path I happened to choose) partial phase shifting could happen when the signal was routed out a pair of bus outs vs. a pair of main outs... ex. My "piano" stem (the piano was a soft-synth sample-based piano) lost it's presence and nearly diminished in volume dramatically when checked in mono... but the Piano doesn't cancel out, at all, when mono-checked in the regular 2-mix version. I'm eager to understand it — but I just don't quite see how that happens, at this juncture.

These differences are only really with older consoles, and ones which have not been maintained properly.

Usually, Buss outputs (commonly called Tape sends or Tape inputs), do not have as much electronics going on as the Master Buss, which will absolutely have a Fader on it, which means at least one more amplifier gain stage to drive that fader, in addition to a Insert point, which again will have another amplifier stage. Of course, each console design has different ways of dealing with these features, some may not have an amplifier for the fader, but at minimum you'd have one extra gain stage for the insert send and return, and an extra ground there which could also introduce more noise, or at the least raise the noise floor a bit compared to the group (tape sends) buss outputs.

As to the phase issue, there could be a bad wiring situation at the patchbay (I typically suspect anything hand wired first, whether a cable or patchbay or whatnot outside the console), or a phase issue with the compressor connections (pin 2 or pin 3 hot anyone? The standard is pin 2 hot, but there is still gear out there, especially vintage gear, with pin 3 hot...).

If it is not a 180 phase change on one side, and only a partial phase issue, I would definitely suspect the compressor first, before the console.

The only real way to find this out, of course, is to wire up a sine wave and a test jig with an oscilloscope capable of giving you a phase view (this is, two signals being seen from the front of them, not their sides). Any phase problems will immediately show up there. Do you have an oscilloscope handy?

Cheers

[TapeOpLarry]

Yeah, phase problems like that seem odd. I'd send some 1 kHz around and see what gives. On my RND 5088 I can point stems out of the busses and it's super close to the same mix when I recombine them. But I know these outputs are all the same op amps and transformers as the main LR outs. I'm not sure I'd trust a "lower cost" console. It'd be interesting to examine the circuitry....

I thought your name was familiar! Those workshops are great, I need to schedule another. Busy summer!