Use of buses in SONAR/Effects Sends

What are buses used for in SONAR? In other threads, it seems like they are primarily used for organization. Like all drum parts go to one bus , the guitar to naother, all harmony vocals to one bus. Is there any audio reason for using them?
What about effect sends? I usually record audio and then insert reverb and other effects into the audio track. What is the effect of using an effect send?

The busses are for organization and effects. I apply most of my effects to the bus, not the track, simply because I can change multiple tracks at once. For example, I might have three background vocal tracks so I can control reverb by collapsing them to a buss and applying accordingly. Same with EQ. I usually stuff all the drum tracks through a single buss.

There's no magic or right/wrong here. Just options.

rallenjones,

you use effect sends to save cpu power.
Say you have 16 tracks and you want to put reverb to all tracks. Then you can insert your reverb plugin to all 16 tracks. If you do that, you have 16 instances of your reverb plugin running.
You'd better use one bus with one reverb plugin. Use track sends to send a bit of your track signal to this bus. Be sure the reverb is set to 'effect only'. With the send poth you can adjust the amount of reverb per track.

Now you could say: but I want reverb just at 5 of the 16 tracks, and I want to put very different reverbs to my tracks.
The answer is: NO, you want reverb at ALL tracks. You maybe want to use two or three reverb busses with different reverbs, but you need to put one or two of the reverbs to all tracks to give them the same room feeling.

hth.
Melf

Hi, rallenjones!

Lots to discuss here. You should get a bazillion responses, along with numerous posts advising you to search the forum - as this has been discussed elsewhere. . .

Anyway, organizational uses are one, yes. Get the drums how you like them, assign them to a bus; get the background vocals how you like them, assign them to a bus; etc., etc. Then you can more easily make adjustments like bringing the background vocals up or down as a group, etc. You got that already.

In addition, you can easily export "stems" from these busses. If you're working with someone else who is mastering or re-mixing, sometimes having stems that represent your general mix (a stereo .wav of your drums, of your bass, of your background vocals, etc. - i.e. a handful of logical submixes that in total represent your overall project) make it easy for them to correct problems, or to have a starting point for a remix project.

There's an efficiency aspect, too, that you mentioned. If you put some general FX (a few different reverbs and/or delays are most often mentioned) on busses to use for FX sends, it can be more economical for your processor. Rather than having a 8th or 1/4 note delay patched in to numerous tracks, why not use one delay as an FX send? Same thing for your general reverb or ambience FX, that you might be using on multiple tracks anyway.

Having the flexibility is the thing. There are many ways to approach it, but it's not related to sonic issues -- bussing the drums to a sub-group doesn't (shouldn't!) change their sound.

M

See, in the time it took me to post, you already had two other respondents!

Thanks guys. Very helpful. After three years of SONAR with no "perfect" projects complete, I do believe that I am now going to research all of this stuff and try to get it right from the start. I had basically been trying to use the computer as a glorified tape recorder. Now, I see that's wrong.
I think I'll spring for some Garrigus too, since he's the Garri-guru of SONAR.

Rallen,

> What are buses used for in SONAR? <

You got a lot of great answers, and I'll just add my "rule" for when you add an effect to a track and when you use a send/return bus:

Any effect that alters the sound passing through it should be added to a track. This includes EQ, phasing and flanging, compression and gating, and so forth.

Any effect that adds new content should be added to a send/return bus. This is mainly reverb and echo, and for those effects the wet/dry mix should always be set to 100% wet.

Note that I said "send/return" bus above rather than just bus. One other use of a bus is when you want to add the same processing to multiple tracks. In that case you use an output bus, not a send/return bus so, for example, all five background vocal tracks can share the same EQ. This not only saves CPU loading, as Melf explained, it's also a lot more convenient to adjust only one EQ rather than 5.

--Ethan

Excellent advice, Ethan. I inserted my first send effect in Sonar today running out of cpu on my newly acquired UAD-1. I did imagine something like your advice, but you made it very clearcut.

ORIGINAL: rallenjones

I think I'll spring for some Garrigus too, since he's the Garri-guru of SONAR.

Thanks, Rallen! I hope you enjoy the book!

Scott

--
Scott R. Garrigus - Author of Cakewalk, Sound Forge and Sound Forge 6, SONAR, SONAR 2, SONAR 3 and **Sonar 4 Power!** books. Books up to 37% off at:
http://www.garrigus.com/

Publisher of DigiFreq. Win a free copy of Native Instruments' INTAKT software loop sampler and learn cool music technology tips and techniques by getting a FREE subscription to DigiFreq... over 17,000 readers can't be wrong! Go to:
http://www.digifreq.com/digifreq/

Hope this question desn't highjack this thread but let's say I record 2 tracks of acoustic guitar simultaneously from 2 sources. I want to apply reverb to BOTH tracks so I set up an effect send bus and load the Sonitus Acoustic guitar pre set, which I beleive is set to 100% wet.

I set the output in each audio track to the effect send, and the output of the effect send to the master bus. So now when I turn on the send in each track I can adjust the amount of reverb that is applied to each track by adjusting the fader in the effect send? Correct?

Then there is 'pre' fader and 'post' fader. It seems as though the pre setting sums the outputs of each buss causing the master to clip and the post setting doesn't sum the 2 tracks? Or do they both sum the busses?

Am I close to understanding this?

So now when I turn on the send in each track I can adjust the amount of reverb that is applied to each track by adjusting the fader in the effect send? Correct?

The send volume knob on each track controls how much of that track's signal is sent *to* your effect (reverb, in this case). The fader on the FX bus controls the amount of reverb that is sent to your master output. Clearer?

Then there is 'pre' fader and 'post' fader.

The pre/post selection enables you to tap your source channel either before or after the channel's fader. If you go post, then lowering the fader also lessens the amount of signal sent *to* the FX. This has the effect of keeping your relative mix of straight and effected signal consistent. If you go pre, then lowering the channel fader does *not* lessen the amount of signal set to the FX. This has the effect of changing your relative mix of straight and effected signal -- given your example, the acoustic guitar in this latter case would appear to get quieter and also "wetter" with reverb, as you lowered the channel fader.

Hope that helps. . .

M

M,

I THINK I'm going to understand the phenom with your explanation. It helps very much.

I still don't get the summing phenom so well, or it's too obvious. When I increase the signal from any source it seems to sum those signals at the master. Is it that simple and when someone is tracking 50 tracks those tracks need to be set low enough to keep the sum of all the tracks low enough to not clip the master? Perhaps a new thread to answer that question?

I'd not rule out the possibility of EQ, compression, or any effect on a bus.
I've often EQd, or compressed a collected group of tracks that are summed to a bus,
with great results.

Steve

Humble,

I won't pretend to know all the math or the exact algorithm that is happening; but the simple answer is yes, your individual tracks are being summed into your output, or to your output bus, if that's how you're set up. (And, yes, if you're grouping like tracks to busses, then you're summing to those busses, and those busses are in turn being summed to your master out.)

Anyway, someone will show up with the precise answer, but the basic answer is yes.

M

When you sum two uncorrellated (idependent of each other) signals, with levels A and B, the summed level will be given by the Pythagorean theorem as the root of A^2 + B^2. So if the two signals are roughly the same level, say A, the new level will be root(2)*A ~ 1.4142*A. Now in decibels, factor root two gives 20 log(1.4142) ~ 3 dB. So far so good. Now how fast dow it grow?

Well, if you sum 50 uncorrelated signals, each at level A, the sum will be at level root(50)*A. Root(50) is about 7.07, so the sum is 20 log(7.07) ~ 17 dB higher. So it's not all that bad.

But even for a perfectly correlated signal, the new level will be, obviously 50*A, or about 34 dB higher. Still, not all that scary. The key is that logarithmic scale turns multiplications into summation, so it does not grow so fast. If we measured the levels in linear scale, the metering would be extremely inconvenient: it would have to span too many orders of magnitude.

Another way of understanding this is the following realization: if your signal peaks at -18 dBFS, it means that the peak is no more than a mere 1/8 of the full scale. (Because 20* log 8 ~ 18). So there's plenty of room for adding 50 others just like this one.

HTH

See, Humble? What'd I tell ya?

Hey, tonyd, you don't really record with your *peaks* only at -18dbFS, do you? Does anyone? We'd be wasting available bit depth (resolution) if we did that, no?

Dang, I already had to help my daughter with her math homework tonight. And now we're breaking out the logs!

Thanks for a better response.

M

M:
...you don't really record with your *peaks* only at -18dbFS, do you? Does anyone? We'd be wasting available bit depth (resolution) if we did that, no?

M, the good thing is that---and this brings us back to the discussion of busses---all the summing happens in the (32 bit on Sonar) floating point arithmetics. While a DA-convertible signal must be between -1 and +1, the floating point arithmetics obviously works for numbers bigger than 1. Even if the original signals peak at 0dbFS, and thus the (internal to Sonar) floating sum will be at + 17dB (way over), this is no problem because in 32 bit floating arithmethic there's over 760dB of headroom. (Floating max is about 3.4 * 10^38, and so 20 log(10^38) = 20*38 = 760).

As a result:
1) internally the signal is not clipped
2) the input tracks took full advantage of the available bit depth.

All that's needed is to lower the gain on the bus to bring the summed signal down by at least 17 dB, before it gets converted back to (16 or 24 bit) integer. This is (at least in principle, if not necessarily in practice, especially with 24 bits) better than attenuating all the inputs by 17 dB before summing them (which would, as M has pointed out, waste some precious bits).

So that's another neat aspect of bussing: the 32 bit float is employed for summing, panning, and all that.

On the other hand, there are some other good reasons, why the individual tracks should still have some headroom, just search the forum for "levels" or "metering" or something like that. In 24 bits, a few wasted bits (6dB per bit) is not such a big deal, especially if the final product is a 16 bit CD.

Very interesting. Even though the actual math goes over my head the essence of the summing process is much clearer, thanks so much for the effort in explaining the phenom.

One last point of clarification, let's say the recording levels are just a bit hot, but under 0db and not clipped. In order to keep the master from clipping, is all that is necessary is to lower the fader on the master so IT doesn't clip? Put another way, can you feed hot, non-clipping tracks into the master and adjust the levels there and still get no clipping on your final output?

If true then recording levels become a matter of preference and not a matter of absolute levels as their summing, no matter the actual levels, can be attenuated in the master bus. True?

Good information, tonyd! You actually cleared up some things for me that I've read elsewhere.

Thanks for taking the time to explain.

M

ORIGINAL: HumbleNoise

One last point of clarification, let's say the recording levels are just a bit hot, but under 0db and not clipped. In order to keep the master from clipping, is all that is necessary is to lower the fader on the master so IT doesn't clip? Put another way, can you feed hot, non-clipping tracks into the master and adjust the levels there and still get no clipping on your final output?

Yes. However, the absence of clipping in the digital signal by itself does not guarantee a distortion free sound on the output. The signal will ultimately get converted into sound, and the DA converters may (and will) clip between samples, if the signal is too hot. The DA needs to interpolate smoothly between the sample points, and even though the discrete points themselves are below 0dBFS, the interpolated points need not be. Depending on who you ask, and depending on where that digital signal goes (you may feed an analog console for further processing), depending on the quality of the CD player you target, the peaks should be at -0.1 dB or as much as -6 dB.

If true then recording levels become a matter of preference and not a matter of absolute levels as their summing, no matter the actual levels, can be attenuated in the master bus. True?

Yes, as long as all the processing is done using 32 bit floating math, the delayed attenuation by itself won't introduce any distortion. However, there may be other reasons why you may want to keep some headroom on the individual tracks. These are mostly related to the interplay between the analogue and the digital domain: wherever analogue processing is involved, you need to worry about the distortion free headroom of the analogue machinery, as well as about the intersample clipping mentioned above.

Mixbus compression is another cool thing that can be done with the busses in SONAR. There was an awsome article on this in a back-issue of TapeOp, WRT to mixing out of the box on an SSL console. But, the concept is definately applicable in SONAR and I think it's pretty interesting. I haven't spent much time trying it out myself. But, I can find out which issue if anyone's interested. Anyway, it goes like this...

Depending on what you want to do, you setup maybe two, three or four busses with different compressor plug-ins, or even different instances of the same plug-in with different settings that would be suited to different 'sonic' groupings. In other words, one buss setup for high-end, one for mids, one for low-end, and others if you want. Then assign channel sends to the desired busses as you desire.

This seems like it would allow for a lot of flexibility and control over the final mix. Plus, with an EQ on each mixbuss too, it would seem that you could end up working with a pseudo multiband compression scheme, and have the ability to pick and choose which tracks, or groups of tracks, and how much of those groups you send to each (EQ/compression) band or combination of bands.

Anyone tried anything like this?

~ Charlie

Now, that's what I call a reply! Scott, I got your book. Where is the best place to start for an experienced user who wants to get into the more internediate level stuff?

The term sum is not a good one here MIX is better. The point is that in the output bus the signal mixed is not summed as such, which implies an addition is going on. It is not. It is simply being mixed. If a true addition or summing were going on then 2 signals of -12 dB would equal -6 dB. That is not happening. The two signals are being mixed at -12 dB. With an output of -12 dB.

Best
John

So does any know in Sonar 4, what is the difference between using busses and track folders? Do they accomplish the same thing? Is one method more versatile?

John:
The point is that in the output bus the signal mixed is not summed as such, which implies an addition is going on. It is not. It is simply being mixed. If a true addition or summing were going on then 2 signals of -12 dB would equal -6 dB. That is not happening. The two signals are being mixed at -12 dB. With an output of -12 dB.

Oh no. Addition is what's going on. Each channel is represented by the (time discretized and level-quantized) sound wave values. In 24 bit precision, these are integers between 2^-23 and 2^23 and their values are, indeed, directly added. When they get bigger than 2^23, we run out of bits, so we substitute the max value 2^23, giving the wave a nice flattop.

When you listen to two acoustic instruments playing live, the pressure waves are summed up. The sound is linear. Air does it. A DAW simply imitates that.

The confusion is we do not express levels in terms of integers between zero and 2^23, we use the decibel scale, 20 log(level/2^23), between minus infinity and zero. Once you measure the levels in decibels, obviously the decibel levels are not going to additive, because this is a logarithmic scale (log (a + b) is NOT log(a) + log(b)). But the wave values, the physically meaningful quantities underneath are additive.

I thought at least part of the question was what is mixing? What does it really mean to mix? And the answer is to mix is to sum.

I guess I put this badly

Live sound is indeed summed. Audio is not in a mixer it is combined without a level change.

Example a single instrument can produce a dB at say X dB add another instrument and you get a louder sound or X + dB, that is in a live situation. In mixing we are combining the wave forms of two or more sounds without adding a level change. That is X + Y = Z. Z is no louder then the loudest sound of X or Y not the loudest sound of X and Y.

I hope I am making myself clear.

So does any know in Sonar 4, what is the difference between using busses and track folders? Do they accomplish the same thing? Is one method more versatile?

Track folders have nothing to do with buses.

Best
John

ORIGINAL: John

Track folders have nothing to do with buses.

Oh ok. I though you could apply one instance of an effect to all tracks within a folder. I guess I haven't really tried it. Thx anyway.

Example a single instrument can produce a dB at say X dB add another instrument and you get a louder sound or X + dB, that is in a live situation. In mixing we are combining the wave forms of two or more sounds without adding a level change. That is X + Y = Z. Z is no louder then the loudest sound of X or Y not the loudest sound of X and Y.

I hope I am making myself clear.

Not really. You seem to be saying that mixing sound in Sonar will not result in an increase of volume, and this is not the case.

I tried a simple test of cloning an audio track 9 times, and playing all 10 copies through a master bus. The volume of the master bus was 20dB louder than the volume of the individual tracks, which seems to jibe pretty closely with what tonyd was saying would happen with a correlated signal. Sure enough, additive signals.

BTW, that test also verified the non-clipping nature of mixing to busses. My first attempt actually ended up clipping, with +10 dB on the master bus. Applying the multiband to the overloaded bus resulted in a nice, clean, unclipped sound despite this fact.

Thanks for the info, tonyd.

Sinc, this is absolutely amazing! On one or more of the interminable threads on this forum asking how to make their mixes sound louder, I proposed my solution, which was to mix everything down to wav, reimport it to SONAR and clone the resulting wav track and then re-export it to wav.
Oh. I just showed my ignorance again, didn't I?
I must have missed something.
Sorry, just a humble hobbyist.
"Free Guitar Solos to All Who Ask".

sinc, I'm wodering why I didn't try that myself to see what the results would be. Thanks for the explanation, I'm going try a couple of experiments myself. As I look back I think what you have noted is what I was seeing, but had no clue why or how to control it.

And thank VERY much to all the posters I've learned a lot from this thread.