How to Add a Sub

I’ll start off by saying that the “adding a sub” mentality is the wrong way to approach this. If you simply add a sub to your system as most people do, you’re probably better off without it. It’ll do more harm than good. It needs to be thought of as a single system. You’re essentially designing a new speaker system. If you’re adding a pair of subs to a 2-way system, it shouldn’t be thought of as 2-ways plus subs. It should be thought of as a new 3-way system.

We need to first establish why you’d want to add a sub. The most obvious and most common reason is to add more low end extension. Most small speakers don’t go very low. -6dB at 40Hz is already fairly low for small 2-ways so a sub is necessary if you want to get closer to 20Hz. When integrated properly a sub will also reduce the strain on the main speakers. This can potentially allow them to play louder. At the same time it lowers distortion by having the woofers not move as much. One interesting phenomenon that you can look up relating to this is doppler distortion. Other factors such as mechanical noise or cabinet resonances can also be improved.

The biggest issue with how most subs are added is the crossover. When not done properly it’ll cause all sorts of dips or bumps in the frequency response as well as phase and timing issues (really these issues are all related to each other). This is why so many people hate the idea of “adding a sub” yet they have no problems with larger multi-way systems. When done properly those two things are the same.

Most subs come with some sort of crossover built in and those should be avoided in most cases. The first issue is that you’re going to be running your audio through a lot of crappy parts (at least on lower end subs). The next problem is that if you’re filtering the top speakers, you’re adding a rolloff curve to something that already has a rolloff curve (unless you’re crossing over very high) which will never give you a proper crossover. In certain cases where the sub and satellite speakers are from the same manufacturer and designed to work together this may already be taken care off but you still have many problems that the manufacturer can’t predict in terms of in-room response and location of the speakers in relation to the room and in relation to themselves. Another issue is that most built in crossovers have pretty gentle slopes which means that the sub and the drivers will overlap across a wide frequency range. Not only is this likely to introduce problems from running the sub too high but it’s just increasing the frequency range where you need to get the sub and satellites to play well with each other.

In certain cases you can let the satellite speakers go down their natural rolloff and then match the sub to that. The advantage is that you don’t have to add additional filtering on the satellites. The disadvantages aside from lack of control is that in this case you won’t get any of the benefits of offloading material from the satellites to the sub. Most ported speakers benefit from a high pass filter below their natural rolloff which basically removes unnecessary woofer excursion (which doesn’t produce sound). That’s a topic for another post.

The first step in adding a sub is to pick a subwoofer to buy. Unfortunately many of the subs on the pro audio market don’t perform very well and cost a fortune. I’ve seen a number of units that start rolling off at 40Hz. To me that’s not really a sub. There are a number of units that will get you down to 20Hz for $700 or less. They come from smaller manufacturers that aren’t very well known in the pro audio world. If you’re interested in finding out more about those, reach out to me. There are also some excellent DIY “flat pack” kits where all you have to do is glue together and paint the cabinet that gets shipped to you.

The other important part of picking a sub is that it’ll mate well with the satellites. If your speakers roll off at 100Hz then you’re going to want a sub that plays extremely cleanly up to there. If your speakers go down to 40Hz then you’ll probably want a high excursion sub that gets you down to 20Hz and doesn’t necessarily have to play cleanly very high. It’s important to keep in mind that the higher a sub is playing, the more you can localize it in the room.

The next step is to position the sub in the room. If using a pair, I normally recommend placing them by the speakers. If it’s a single sub then some experimentation in placement can be beneficial. Using REW or some other measurement tool is critical. For placement, you want to focus on looking at the area where the crossover will be and below. If you’re crossing over at 40Hz then I’d focus on looking at 20Hz - 60Hz.

In general you’ll probably have the best response by having the sub close to the front wall. The goal is to get the smoothest response. Bumps aren’t too much of a problem and any sort of boundary reinforcement generally isn’t an issue either. The biggest things to avoid are dips especially around the crossover.

Once a decent spot is found then it’s time to set up the crossover. Many people will argue this but my opinion is that DSP needs to be used for this. There are many passive systems that use DSP for crossing over to the subs simply because it’s not possible with passive circuitry and even active analog doesn’t give you the flexibility needed in most cases. At the same time, more advanced forms of DSP such as FIR processing is unnecessary for this. All that’s needed is the ability to create a crossover with a slop of at least 24dB/oct (48 or 96 is preferable in most cases), the ability to time delay the subs and satellites independently, and a few bands of parametric EQ. Options include units from miniDSP, DEQX, Trinnov, dbx, and Xilica.

Once I’ve picked a crossover point (a process I won’t get into here), I start by measuring the sub(s) and satellites individually with and without the filter to see how their responses are affected. These measurements also allow you to match the volume of the speakers. The next step is to try measuring the system with them together. In certain cases it might be great right off the bat. In other cases it might just need the polarity of the subs fixed. Most likely, however, there will be some fine tuning required. The next thing to look at in REW are the phase and impulse responses. This data is used to time align the speakers. Once aligned, the phase will usually sort itself out. If after aligning and checking polarity there are still problems, it might be worth trying a new crossover point. In the case where the crossover overlaps with the natural rolloff of the sub or satellites, then EQ bands can be used to smooth that out and create the slope that you’re after.

Once the crossover has been properly set up then the last step is to add some EQ. This can be to counteract effects of the room or extend the response of a sealed sub. These are topics for posts of their own but what I’ll say is to be judicious with the amount of EQ used to counteract room issues. Around 4 bands of EQ with at most 6dB should be enough for most cases. In terms of extending a sealed sub, this essentially trades off the max SPL of the system with how low it extends. Care must be taken to not push a sub too far.

For anyone curious about my own system, my main speakers use 15” Acoustic Elegance subs which have pretty much the lowest distortion possible and an STW350 sub for my LFE which is flat down to about 18Hz. For travelling I use a 12” Ultimax sub. In my new room, each main speaker will have dual AE subs and for my LFE I’ll either add a second STW350 or switch to using the Dayton MaxX drivers.

It’s important to note that this post is by no means an exhaustive look at how to set up a sub. Anyone interested in doing this on their own should do more research in the areas I’ve covered. Shameless plug - setting up subs is one of the services I offer. This is typically included when I design rooms or set up DSP units for speakers. Whether you’re doing it on your own or having me help out, the end result should be a cohesive full range system.

My Mastering Process

Many people still find mastering to be some mysterious process and don’t understand what it’s for. When starting out it can be confusing to understand why mastering exists.

There are many different reasons and goals for mastering. Most importantly is offering a fresh perspective to a song or set of songs. The mastering engineer typically won’t be very involved throughout producing, recording, and mixing so they haven’t heard the song hundreds of times. Typically a mastering engineer’s monitoring system is as good if not better than the previous systems used in the process so they are able to make more informed decisions such as making any necessary changes to the tonal balance of a song. Then there’s the more traditional aspects of preparing a song to be released which includes getting it to the proper levels for various methods of distribution as well as sequencing an album and making sure that the songs in an album all work together. This also includes the final quality control of a song to make sure that there isn’t anything technically wrong with the files being released.

There is a lot more that could be said about the role and responsibilities of a mastering engineer but in this post I’d like to focus more on my general workflow and approach.

On some projects I will be involved from pretty much the start and may hear some material throughout the process to make any suggestions I might have but more typically I’ll only be brought onto a project at the very end once the final mix is approved or very close to being approved.

The first step after receiving files is to load them into my template. The template just includes things like routing and exporting operations prepared rather than any predefined processing and plugins. Files imported include the main mix and any alts (instrumental version, TV mix, etc.). It also includes the limited ref mix. If the project includes multiple songs then I typically work on all songs in a single session. The exception to this is if an album requires more complicated sequencing where I’ll print all of the processed songs and bring them into a new session just for sequencing.

The very first thing I do is sit back and listen through the ref mix in its entirety. This will usually be the first time I experience the song. While listening I’ll be thinking about the intentions of the song, perhaps the direction that it’s trying to go, and if there’s anything that sticks out as an issue I’ll make a mental note of that to address later.

After this first listen through I’ll have a pretty good idea of what I think might need to be done and what tools I think will work. At this point I start to play around to see what works and what doesn’t. A tool that I thought might be perfect to push the song a step further or to fix an issue might not work at all. After a few minutes I’ll settle on something that works well and then I’ll start fine tuning. At this stage it’s very much a matter of A/Bing small changes to try and hone in on the song. Of course sometimes I might end up not being happy with where I am and take a few steps backwards to try a different approach.

It’s important to mention that throughout the entire process I’m constantly A/Bing with the ref mix. This is both to make sure that the decisions I’m making are actively improving the song (taking it closer to what I envision it being) and to make sure that I don’t make things too different from what everyone else on the team has gotten used to. The intentions of the song and mix need to be preserved. Having said that, on some projects the client will want the mix changed drastically while others might want me to barely touch it. Figuring out how much room I have to play can be difficult and usually takes either talking to the client about what they’re looking for me to bring to the project or I can do what I think is best on a first pass and risk doing too much or too little which can then be refined on a second pass.

Once I’ve gotten everything refined and am happy with where the song is, it’s typically been around 45 minutes. At this point I’ll usually take a short break of 5 to 10 minutes. After this I’ll listen to the song entirely through and see if I’m happy with it. After making any necessary adjustments I’ll jump around the song and A/B with the ref mix to double check that everything is still good in relation to the ref.

If I’m working on a project with multiple songs then at this point I’ll move on to the next song. When working on albums then I also frequently jump around to compare the current song I’m working on to other songs and make sure that they work well together. Once I’ve gone through all of the songs then I’ll sequence them meaning that I determine the flow of the album. After this I’ll listen to the album all the way through and make any notes of anything I want to go back and adjust. This is often the very first time an album has been listened through entirely which is quite exciting to have that honor.

The last step in the session is for me to listen though again on headphones. This is the first QC step for me. I make sure that I didn’t miss anything on the speakers and if I had doubts about something will use this as a last opportunity to make a change. Because I get a lot less detail on headphones than my speakers, I’d say that 98% of the time I don’t make any changes following this headphone listening.

After exporting the masters at the highest resolution I bring the files in to RX for SRC and bit depth reduction. Typically the highest resolution masters are either at 96kHz or 48kHz and 24 bit. I normally use RX to generate 44.1kHz files at 24 and 16 bit.

At this step all of the masters have been created and just need a final round of QC. Usually I’ll listen through the 44.1kHz 16 bit files since any issues on the higher resolution masters would be present in these. The full mixes get listened through in their entirety. For alts I don’t always listen through entirely. If that’s the case then I’ll at least open every file in RX to compare with the full mix and make sure it all looks good. I’ll also skip through each file to make sure that something like an instrumental is actually an instrumental all the way through with no vocals accidentally left in. A 4 minute song with 4 alts would take 1 hour to listen through all of the files at all of the formats so it’s just not possible.

The only times I’ve ever had issues show up in my final QC is when I’ve used new plugins so normally there aren’t any issues here but I wouldn’t feel comfortable sending out masters that haven’t been listened through. Just because it plays fine in the DAW doesn’t mean that exported files are fine. Because of this, any time I use a new plugin (or update a plugin) I’ll either do extensive testing in testing sessions or I’ll then listen through every alt file exported until I’m confident that the plugin isn’t going to cause any issues when printing masters.

The last step is to upload the masters to Dropbox and send a link to the clients. If any revisions are necessary, then after making changes all of the exporting and QC steps are repeated.

With all of these steps, mastering a single usually takes me around an hour and a half from when I download the files to when masters are sent out. I try to make that at least an hour of listening before QC. Any revisions will add to that. The range of time for working on a single ends up being 1 to 4 hours and so my rates are essentially an average reflecting that. Projects with multiple songs typically start to move faster as I get a sense of the album.