Master Faders Demystified – Part 2
This is an excerpt from my column, “The Pro Tools Corner” at audioMIDI.com. Read part 1 here
Update: Pro Tools 10 introduced HDX which uses a 64bit floating point mix bus and 32bit insert chains, effectively making it sonically identical to a native system. Older HD cards will continue to use the 48 bit fixed mixer and 24-bit insert points until they are not supported in Pro Tools 11. I also added a new section on native vs. TDM insert points.
Update 2: As of Pro Tools 9 it was announced that Pro Tools Native uses a 64 bit floating point mix bus for internal summing and 32 bit float insert points, this was only revealed when the Pro Tools HD native card was released. Prior to this, it was stated that Pro Tools LE had a 32 bit floating point summing mixer, but it has come to my attention that it has been 64 bits for quite some time now (even before Pro Tools 8), although this was never revealed most likely due to marketing concerns over a perceived difference between the 48-bit TDM mixer found in the more expensive legacy HD systems.
Master Faders Demystified: Part 2
In the previous installment of the Pro Tools Corner, I discussed the importance of integrating and correctly using master faders in your mixes. While master faders are quite easy to use and understand in the Pro Tools Native, HD Native, LE and M-powered mixers, using them effectively in Pro Tools HD TDM is an entirely different story. In part 2 of master faders demystified, I will cover some of the unique signal flow considerations HD users must account for while using master faders and also give you some overall pointers for using a master fader’s inserts for mix bus processing. If you haven’t already read the first part of this article, I suggest you check it out before proceeding.
Note: Remember I am talking about the TDM mixer here, if you are using HDX or Pro Tools HD Native, this does not apply in the same way, make sure you read part 1 of this article to effectively use master faders in your workflow.
Master faders in the HD TDM 48-bit mixer
First off, if you are a Pro Tools Native, HD Native, HDX, or older Pro Tools LE or Mpowered user, the following only applies if you plan on creating mixes that translate to a Pro Tools HD TDM environment. Otherwise, I assume you are reading this section from an HD TDM user’s perspective. If you plan on only using Native, Native HD, HDX or LE/Mpowered from start to finish, you should definitely check out the first part of this article for some useful insights on setting up and using master faders in your mix.
So what’s the big deal with the HD TDM mixer anyways? In the Pro Tools HD TDM mixer, think of tracks as having a 24-bit input and a 48-bit output. That is to say that the HD mixer adds or “sums” at a 48-bit fixed precision and has tons of headroom to combine many hot (near full code) signals without clipping internally. Now just like every other DAW on the planet, the end result of the mixer’s summation has to eventually dump out to 24-bit converters, creating a unique challenge for in-the-box mixers. The master fader’s job, just like in the Pro Tools Native mixer, is to act as a final output trim, or “bridge,” between the internal higher resolution 48-bit mix bus and the lower resolution 24-bit converters. When signals inside the mixer combine to exceed the maximum level allowed by the 24-bit outputs (which they often do), the master fader can be used to trim down the end result, thus avoiding any clipping at the converter. Because internally these signals combine at a very high precision, no detail is lost nor samples clipped when signals are added together inside the mixer. The master fader acts as a sort of bit selector, allowing the mixer to exit the system with the most significant or “best” 24-bits to the converter, not clipping, nor leaving too much excess headroom.
Now this is all fine and good until you start submixing into busses or using sends. Because of the way the TDM mixer and DSP chips are structured, every time you use a send or output set to a bus, a new unique 48-bit mixer is created for that bus path. This mixer lives as a separate entity to the main mix bus summing out to your converter’s outputs. Because the mixer is a separate entity, the signal is truncated back to 24-bits when the bus is “returned” to the main mixer via an axillary track (remember, tracks have 24-bit inputs and 48-bit outputs, this included aux tracks). The problem with this structure is that you can clip a bus’ mixer without clipping the main outputs, or in other words, you could be clipping your mix without lighting up the clip indicator on your main output’s master fader.
You can see in the screen shot that the submix is clearly clipping while the master fader is not. Even reducing the level of the aux submaster does not illimate the clipping because it is occuring on “Bus 1-2” as it re-enters the mix on “Aux 1” before the aux tracks fader can attenuate it.
So what can we do? It’s quite simple actually. If you feel that your submix (via a send or bus output) is in danger of clipping, simply create a master fader set to monitor and trim that specific bus. Now if you are super picky about clipping anything, you might have a unique master fader for each bus you are using in your mix (for both sub-mixing and send/return fx processing), but I find that by carefully monitoring my sub-masters (the axillary returns accepting the bus signals) I can usually get away with only needing a master fader for the occasional drum sub-mix and my main submix bus (in the case where I am submixing all my tracks to an internal mix bus for printing to a new audio track).
You can see in the screen shot that the submix clipping has been eliminated by a new master fader set to the same bus pathway (bus 1-2).
Now as I said in last week’s article, this doesn’t happen the same way in the Native mixer because the mixer is 64-bit float through the mix bus and 32-bit float from insert to insert, meaning you can effectively use just a single master fader set to your main outputs (usually A 1-2). But if you want you mix to translate effectively to an older HD TDM system, you may want to follow some of these practices as a general habit.
HD TDM vs Native Insert pathways:
In Pro Tools HD TDM, insert pathways are 24-bits, just like bus inputs, so you can clip inserts without clipping the mix bus. How? For example, if you take a signal that is peaking at -6dBFS and you apply an EQ to it at one of the insert points and proceed to boost 12dB on any frequency, you will effectively clip the 0dBFS output by 6dBs. While most plug-ins have internal guts that process in floating point to prevent internal clipping of the algorithm, in Pro Tools HD TDM plug-ins, the signal will always be truncated back to 24-bits when it passes from one insert to the next, regardless of the way the internal guts handle overs. So if you were to preform that boost in your EQ, thus eating up all your headroom, and then brought that into a compressor and brought the signal back down to say -10dBFS, your signal would have already had the top 6dBs clipped off of it before you could bring it back down on the next insert. At this point, your master fader wouldn’t show clipping because signal would have already been clipped at the output of the EQ plug-in and subsequently brought down by the next insert. To avoid this, you MUST use an insert’s input and output trims when mixing in the TDM enviroment, even if the plug-in internally supports dual or triple precision processing.
The only caveat to this is if you are going from RTAS to RTAS insert (or AAX Native to AAX Native) in the TDM mixer. In this special case, the signal is indeed passed through to the next plug-in in the chain at 32 bit float. Try this experiment, take a sine wave at -5dBFS and run it into a TDM version of EQ-3, now do a boost of at least 10dB so you can hear the clipping distortion. Next place another EQ-3 TDM plug-in directly after in the next insert slot and pull the output trim down by at least 10dB. If both plug-ins are TDM, you will still hear the distortion created in by first insert clipping the output, even if you bring the signal back down to under 0dBFS in the second insert. Now take those two plug-ins and convert them to RTAS (or AAX native) using the icon at the top right of the plug-in (the one that says TDM). You should hear the clipping disappear, as the signal is now being passed at 32-bit float between the two plug-ins, thus allowing you to exceed the 0dBFS mark on one and recover that to below 0dBFS at the next insert point without any loss of signal.
Regardless of what happens with TDM and non TDM plug-ins in the 48-bit TDM mixer, in the Native mixer and the HDX mixer, RTAS and AAX plug-ins always pass their signals at 32-bit float, which would allow you to “carry over” any data exceeding 24 bits via the exponent into the next insert, leaving it up to you to make sure you reconcile your final signal at the master fader before hitting the D/A converter. However, I find that certain plug-ins (try Maxim for example) will truncate at 0dBFS regardless, based on the way they were coded, while other plug-ins, like vintage modeled compressors and EQs, tend to have a sweet spots based on their analog counterpart that may not sound as good as you’d like if you are pushing a huge amount of gain through it. Remember, most analog gear is set up to work best with a signal level around 0 VU, and most vintage modeled plug-ins are calibrated at between -14 and -20dBFS = 0VU, so keep that in mind when deciding how hot you want to run signals through insert chains. Bottom line, I find that gain staging still matters, even in a floating point environment.
It is my hunch that when I hear these guys saying, “I bought the new HD native card, or HDX card and my old mixes just sound so much better instantly,” that what they are hearing is either the sound of the new HD series converters (which sound a lot better than the blue faced ones) or they are all the sudden hearing there mix without insert and bus clipping for the first time. In reality, if you respect your gain stages and follow the operating rules of the system you are mixing in (be it TDM or Native), the mixes will be BIT for BIT the same. However, even tiny amounts of digital clipping happening over many insert points and busses can add up to a certain haziness that eats up fidelity, and muddles yours stereo image.
This “micro clipping” all over my mixer is why I am always super vigilante when mixing in the HD TDM environment. Because of this burden of always having to pay attention to every gain stage like a neurotic hawk, I have since moved over to doing all my mixing on using the Native mix bus, and only boot up the old TDM system when I need to do low latency tracking. With today’s super powerful computers and the fact that most of the cool new plug-ins are native only, I find that forgoing the TDM cards come mix down and mixing natively actually allows me to do bigger mixes with less trouble. Once more plug-in developers start supporting the AAX format, I might consider upgrading to the HDX system to get the best of both worlds, floating point mixer and DSP power to spare, but unfortunately I can’t just drop everything I’m working on and upgrade to an unsupported rig just yet.
Master fader inserts
In Pro Tools, master faders feature 10 post-fader inserts. Unlike regular inserts, which are pre-fader and do not take into account the channels volume control when processing a signal, a master fader’s inserts happen post trim and are, in essence, the final stop before the mix leaves the system and hit the converters. This makes master fader inserts the perfect place for master bus processing (or processing all the audio of a mix simultaneously). So on a master fader it goes Fader > Inserts > Meter.
The master fader is typically the place to insert a mastering, or “brick-wall” style limiter along with any dithering algorithm used for bit-depth truncation. Here are some tips for using a master fader’s inserts:
- Read the white papers: “Mixing in the Box” (http://akmedia.digidesign.com/support/…/Mixing_in_the_Box_26689.pdf) by Stan Cotey and “The Pro Tools 48-bit Mixer” (http://akmedia.digidesign.com/support/docs/48_Bit_Mixer_26688.pdf) by Gannon Kashiwa
- When submixing through a master bus internally, with the goal of printing your mix to a new audio track, you must set your master fader’s output to address that bus if you want its plug-ins to make it onto the new audio track. This does two things: first, it protects the new recording from clipping the new 24 bit audio file that is being created, which unless you are using Pro Tools 10’s 32-bit audio file format, will clip just like your converter. Second, it allows you to bring along any master bus plug-ins you are using into the printed submix. Otherwise you would need to first print your mix internally and then use “bounce to disk” to gain any additional processing reflected on your converter output’s master fader.
- When using dither on the master fader, it should be the last insert in your processing chain, as adding or subtracting any gain post dither will affect its placement at the least significant bit and negate its goal of reducing quantization artifacts. You should not use dither if you are going to print your mix internally and export the file from the regions list at 16-bit, as dither is automatically added to truncated files at export (currently, this cannot be disabled)
- When using “brick-wall” limiters/processors on your master fader, first make them inactive to ensure that you are not clipping the output or bus. In Pro Tools HD TDM, an insert has a 24-bit input and output (regardless of whether or not the plug-in itself does dual precision or 48-bit internal calculations) so they CAN and WILL clip. Normally plug-ins will identify input/output clipping by lighting up in red, however I have found that certain brick-wall limiters are not set up to show I/O clipping correctly. Therefore, you may unintentionally feed a signal from a clipped mix bus into a mastering limiter, set its ceiling to something lower than 0dBFS and never know that you mix is actually clipping the 24-bit inputs of your limiter.
- You may want to use the included TL Mastermeter plug-in to track when and over how many samples your mix clips at the master fader. This can help you can track down specific transients/sections that are causing trouble and attenuate them manually.
- While I might get rocks thrown at me for saying this, generally you don’t have to be hyper critical about a clip here or there. Being aware of the system and where clipping can occur is a step in the right direction. When you do decide to pay attention, you may be surprised at how much your mixes clear up.