thoughts and questions regarding mixer output levels vs input

on a mixer, if all faders and gain trims etc are set to 0 and you have a signal on the line input, should you expect the signal to be the same level at the master output?

How does pan law come into play here? say I center the pan, should the signal be at 0db at both left and right outputs, OR should the output of the mixer be calibrated for 0db while the signal is panned? I assume which way it is calibrated could be different across various mixers and maybe there is no hard set rule?

I've seen some mixers have an extra +10db or so at the faders of the input channels, but the master channel goes to 0db with no extra "headroom" - why is that?

I am sure a lot comes down to the manufacturer. Back in the 70s when I was at Neve, most mixers had a separate Pan button. So normally you could route a channel to one or more groups. With presets and faders at 0, it was unity gain. The pan button then dropped in the pan circuit. If the pan was hard right or left, you still had unity gain. In the centre it was 3 and a bit dB down.

Channels had 10dB in hand for the occasions where you need to add some gain during a mix or recording. Group/master faders had 0dB in hand because they were principally intended to be used to effect a fade.

Cheers

Ian

Humner said:

on a mixer, if all faders and gain trims etc are set to 0 and you have a signal on the line input, should you expect the signal to be the same level at the master output?

Click to expand...

Maybe

How does pan law come into play here? say I center the pan, should the signal be at 0db at both left and right outputs, OR should the output of the mixer be calibrated for 0db while the signal is panned? I assume which way it is calibrated could be different across various mixers and maybe there is no hard set rule?

Click to expand...

Generally pans are down -3dB in L/R when panned center. 0dB when hard left or right (there are some variant pan laws -4.5dB etc).

I've seen some mixers have an extra +10db or so at the faders of the input channels, but the master channel goes to 0db with no extra "headroom" - why is that?

Click to expand...

The +10dB gain in channel strips is for convenience (so you don't have to re-trim a weak signal).

I have put +10dB in the master before, and even a -10dB trim on the master sum for a console with over 100 stems into the L/R bus.

FWIW back in the bad old days it was important to keep channel signals hot to preserve S/N. Modern lower noise electronics has made console noise management less of a problem.

Console designers worry about this stuff so the customers don't need to.

JR

Humner said:

on a mixer, if all faders and gain trims etc are set to 0 and you have a signal on the line input, should you expect the signal to be the same level at the master output?

Click to expand...

Depends on the mixer, and specifically the number of inputs to be combined. On a "large" mixer, you have to expect some attenuation at the mix bus so you don't overload it when trying to combine 48 inputs.

How does pan law come into play here? say I center the pan, should the signal be at 0db at both left and right outputs, OR should the output of the mixer be calibrated for 0db while the signal is panned? I assume which way it is calibrated could be different across various mixers and maybe there is no hard set rule?

Click to expand...

The idea is that when you pan a signal across the stereo field, its volume remains constant. To do this, you must have some attenuation of that signal to both left and right when panned in the center. Remember that assigning a signal to two channels gives a 6 dB voltage gain when you add those two channels. When you pan up the center, your left and right outputs sum acoustically so you must attenuate electronically before the outputs to compensate.

(How much to attenuate in the center, and how much to boost when panned hard left or right, is a matter of personal preference.)

I've seen some mixers have an extra +10db or so at the faders of the input channels, but the master channel goes to 0db with no extra "headroom" - why is that?

Click to expand...

For inputs, there is a common (and whether it is ideal in terms of noise performance or not is another debate) mix paradigm, certainly in live sound, where all of the input faders line up on unity. The reasons include knowing where the mix is by default, and also that fader sensitivity (motion vs output level change) is minimal around unity. (Small motion makes small changes; if the fader lives around -12 dB then the same motion makes a larger change in level.) In any case, with all of the faders lined up on unity, if you need to boost, you need a circuit that lets you do that.

As for groups, as Ian says for mixing recordings the group faders were mainly used for managing fades of large numbers of inputs. It's a lot easier to pull down a stereo pair of group faders than 30-some-odd input faders.

-a

Andy Peters said:

Humner said:

on a mixer, if all faders and gain trims etc are set to 0 and you have a signal on the line input, should you expect the signal to be the same level at the master output?

Click to expand...

Depends on the mixer, and specifically the number of inputs to be combined. On a "large" mixer, you have to expect some attenuation at the mix bus so you don't overload it when trying to combine 48 inputs.

Click to expand...

This is an interesting point and one I have not thought about in detail before. It is reasonable to consider music signals as uncorrelated so two equal level ones will be 3dB louder than on on its own. If you add two more, the result of the four will be another 3dB louder. add another four and the result will be another 3dB louder and so on. For 32 channels, the output will be about 15dB louder and for 64 channels it will be 18dB louder.

This assumes all the signals are equal in level and are all playing all the time. In practice there will be a few prominent signals and mostly about half the tracks will be laying at once. So it seems reasonable to assume it is unlikely the total bus level will be no more than 15dB higher than when just one channel is playing. To get the bus level back to 'normal' we need to lose 15 dB.

The question is where to lose it without compromising headroom or noise. If we lose it at the group/master fader, there is no noise compromise but there is a headroom compromise. Neve mixers had 26dB of internal headroom which would therefore be reduced to just 11dB if we lost 15dB of gain on the master fader.

Alternatively we could reduce the bus gain by 15dB. This avoids the headroom compromise and has no noise compromise. However, it only works for large numbers of inputs. For small numbers of inputs there will not be enough bus gain. Users will want to use their faders near 0dB which means they are likely to crank up the channel gain preset and this will compromise headroom.

To lose it at the channel faders means pulling all channels faders back by 15dB. Possible bit a PITA. There is then no headroom compromise. Noise is compromised minimally because the total bus level has dropped by 15dB but the bus noise has not. As a rule, noise from the channels exceeds bus noise by a comfortable margin so this is not really a noise compromise.

We could keep the channel faders around zero and alter the channel preset gain. Again there is no headroom compromise but there is a noise compromise because both the bus noise and the channel noise will increase.

The ideal solution would be a globally set bus gain switch that cut bus gain for large mixes.

Interestingly, most passive mixers (Neve) try to keep bus impedance constant by grounding unassigned channel sources. This not only keeps bus impedance constant but also keeps bus loss constant. If you let unassigned sources float instead, the bus loss would be proportional to the number of channels assigned - which is exactly what we want.

I have no idea how this whole question was tackled at Neve or by any other mixer manufacturer for that matter.

Cheers

Ian

For summing a modest number of stems there is little need to pad the master bus... I only did it for one console that had over 100 feeds to the L/R bus.

One manufacturer made a big deal in advertising about padding down their bus, but they made a big deal in their advertising about everything.  In practice some pad in the bus gain will reduce the significance of bus noise, as you end up hitting the bus hotter.

Looking at it backwards, how many of the exact same signals, or even similar signals will be playing in the mix at full level at the same time? Answer should be not many. The existing 20 dB of bus headroom above 0VU should be adequate to prevent saturation in typical use.

Only a problem if you over think it into being one.  8)

JR

JohnRoberts said:

For summing a modest number of stems there is little need to pad the master bus... I only did it for one console that had over 100 feeds to the L/R bus.

....

Looking at it backwards, how many of the exact same signals, or even similar signals will be playing in the mix at full level at the same time? Answer should be not many. The existing 20 dB of bus headroom above 0VU should be adequate to prevent saturation in typical use.

Only a problem if you over think it into being one.  8)

Click to expand...

I suppose that bus padding is hidden when considering that most inputs come from preamps with variable gain, so the original question, "on a mixer, if all faders and gain trims etc are set to 0 and you have a signal on the line input, should you expect the signal to be the same level at the master output?" is somewhat moot. You trim the inputs up to some level, perhaps 0 dB on a PFL meter, and then adjust faders to make a mix.

And this sort of plays into something I've always wondered about.  Those of us who've mixed on large-format Midas consoles (like the H3000) "know" that you have to run the input levels "in the red" to get it to sound "right." And by "right" I don't mean that it sounds "better," I mean that there's something coming out of the rig!

I suppose this has to do with the choice of operating levels inside the console.  The mix-guy hits the PFL button or otherwise watches the channel input level meter, and dials up a level as the musician makes a jazz noise. There is some level called 0 dB, and that's a reasonable target level. Lather, rinse, repeat for all inputs. Now on the big Midas, if you do this, you've got a quiet mix, and this is reflected not only on what comes out of the speakers, but also the output mix bus level meters. So to get anything happen, you have to crank the input levels until the meters turn christmas-tree red, easily +9 on the PFL. There's no distortion on either inputs or outputs, but the meters are screaming at you.

If the expectation is that you're really going to mix four dozen inputs to left/right and that those inputs are reasonably coherent (say, you're mixing an orchestra and you've got a dozen mics on the violins and violas), then it makes sense that you need to trim your inputs cold, that is to Midas' recommended 0 dB on the input meters, to prevent bus overload.

But for little vocals/guitars/bass/drums/percussion rock band I mix, which uses all of 24 inputs, and half are for percussion, I'm not going to overload the mix bus at all even with inputs trimmed to +9 dB on the H3000's input meters.

This is just a weirdness I've noticed from mixing on these things.

Andy Peters said:

JohnRoberts said:

For summing a modest number of stems there is little need to pad the master bus... I only did it for one console that had over 100 feeds to the L/R bus.

....

Looking at it backwards, how many of the exact same signals, or even similar signals will be playing in the mix at full level at the same time? Answer should be not many. The existing 20 dB of bus headroom above 0VU should be adequate to prevent saturation in typical use.

Only a problem if you over think it into being one.  8)

Click to expand...

I suppose that bus padding is hidden when considering that most inputs come from preamps with variable gain, so the original question, "on a mixer, if all faders and gain trims etc are set to 0 and you have a signal on the line input, should you expect the signal to be the same level at the master output?" is somewhat moot. You trim the inputs up to some level, perhaps 0 dB on a PFL meter, and then adjust faders to make a mix.

And this sort of plays into something I've always wondered about.  Those of us who've mixed on large-format Midas consoles (like the H3000) "know" that you have to run the input levels "in the red" to get it to sound "right." And by "right" I don't mean that it sounds "better," I mean that there's something coming out of the rig!

I suppose this has to do with the choice of operating levels inside the console.  The mix-guy hits the PFL button or otherwise watches the channel input level meter, and dials up a level as the musician makes a jazz noise. There is some level called 0 dB, and that's a reasonable target level. Lather, rinse, repeat for all inputs. Now on the big Midas, if you do this, you've got a quiet mix, and this is reflected not only on what comes out of the speakers, but also the output mix bus level meters. So to get anything happen, you have to crank the input levels until the meters turn christmas-tree red, easily +9 on the PFL. There's no distortion on either inputs or outputs, but the meters are screaming at you.

If the expectation is that you're really going to mix four dozen inputs to left/right and that those inputs are reasonably coherent (say, you're mixing an orchestra and you've got a dozen mics on the violins and violas), then it makes sense that you need to trim your inputs cold, that is to Midas' recommended 0 dB on the input meters, to prevent bus overload.

But for little vocals/guitars/bass/drums/percussion rock band I mix, which uses all of 24 inputs, and half are for percussion, I'm not going to overload the mix bus at all even with inputs trimmed to +9 dB on the H3000's input meters.

This is just a weirdness I've noticed from mixing on these things.

Click to expand...

Yup, back in the day different consoles had their different quirks (internal gain structure) mainly to deliver good S/N.  But this is the work that design engineers do so the customers don't have to worry about it. Modern low noise electronics, and digital mixers make this even less of end user concern.

Of academic interest I've done mixers using a number of different approaches, including one really obscure one, where the channel faders were labelled as having +10dB of gain, but were in fact only unity gain inside at full scale.  This allowed users to bang the channel inputs hot (near clipping), bang the channel fader to +10dB, cut the master fader -10dB, and output the clean full level input signal un-clipped.  This gain structure makes it pretty much impossible to clip a bus, with an unclipped input signal unless you double bus or do something silly like that. (It should be apparent from this you can get 0dB out from 0dB in using different internal gain structures as long as they net out to unity gain.)

I only used this topology on modest sized mixers targeted toward musicians mixing themselves from on stage so they could always push the channel faders without any fear of saturating internal paths. The trade off was S/N but modern low noise electronics made this negligible.

JR