ROUTING view: Unity Gain with Splitters?

[TonalDynamics]

Hey gang,

Not sure if anyone loves the routing matrix within each channel as much as I do, but it's one of my favorite and most productive features when I want to do some serious sound design with complicated signal chains.

When you use a splitter though, you reduce the output of each split by a certain amount. This interferes with gainstaging in certain setups and is thus undesirable.

Anyone know the actual convention/formula for this?

Trying to remember I'm thinking it was -1.75dB of gain per split when I tested once, but I'm not sure.

Help appreciated!

 

[SwitchBack]

If I remember correctly, when you insert a splitter then the level in each leg stays the same as the level before splitting, like splitting a line output onto multiple line inputs. This means that when the paths join again the sum of those signals needs to be attenuated to bring the level after the split down to the level before the split. This allows you to move say a compressor from before the split to one of its legs or from a leg to after the split and still get the same gain reduction from it.

The attenuation will depend on the number of legs: -6dB for 2 legs, -9.5dB for 3, -12dB for 4, -14dB for 5. These attenuations assume 100% 'phase' correlation between the signals from the legs, as would be the case with straight-through legs. But processing in the legs can throw correlation off, meaning that the splitter's fixed output attenuation will be too much and you have to gain things up a bit.

 

[Dave71]

It's a bit more complicated than that. In general you can say that the sum of each split will make no difference. A channel split will split into 2 signals which each are 6 dB lower in level compared to the full signal in order to achieve this. When you use more splits this will of course add up. Each path in a 5 way split is -25.6 dB lower in level in order to maintain a zero dB difference when having all channels on. I did not find documentation on this, I simply used a sinne wave to check this. And it was apparent that the splitter does not cause a volume drop in itself, but to keep overall levels equal it does lower the volume in each splitter path. With the above being the reason. A splitter instantiated on a channel should not cause an increase in level. Now, you can of course add a mixtool to each path to compensate if needed, but that can bring new issues of course. You can imagine that if there was no compensation of level, it would become very bad very fast in complex routings. I understand that in your use case it is not desired to have this level drop but for a correct working of the plugin it is absolutely necessary

 

[SwitchBack]

I'm pretty sure I got it right. Easy enough to test: Set up a channel with the tone generator, a splitter and the level meter. You can move the level meter from after to before to inside the splitter and it will show the same level everywhere. Then place it after the splitter and mute all legs of the splitter but one and you'll get the attenuations I mentioned before.

 

[Dave71]

I'm pretty sure I got it right. Easy enough to test: Set up a channel with the tone generator, a splitter and the level meter. You can move the level meter from after to before to inside the splitter and it will show the same level everywhere. Then place it after the splitter and mute all legs of the splitter but one and you'll get the attenuations I mentioned before.

I did the testing just now. The combined gain of splitter paths is the same as before, which means that each path has a reduced level, based on the number of paths. I tried with several configurations. The numbers I mentioned are from the tests I just did.

 

[SwitchBack]

Hmm, so did I, confirming what I wrote: Tone Generator at 0dB output level and a 5-way splitter gives 0dB signal level in each leg and 0dB signal level after the splitter. Mute 4 legs and the level in the 5th is still 0dB, but the level after the splitter drops to -14dB. So the splitter's gain reduction is on its output, not in the legs.

Oh, and the formula/convention is 20*log(1/2) for a 2-way split, up to 20*log(1/5) for a 5-way split

 

[Dave71]

Hmm, so did I, confirming what I wrote: Tone Generator at 0dB output level and a 5-way splitter gives 0dB signal level in each leg and 0dB signal level after the splitter. Mute 4 legs and the level in the 5th is still 0dB, but the level after the splitter drops to -14dB. So the splitter's gain reduction is on its output, not in the legs.

Oh, and the formula/convention is 20*log(1/2) for a 2-way split, up to 20*log(1/5) for a 5-way split

Ah, I missed that part, apologies! I only measured at the output, not the separate paths themselves. More tests to do to see where exactly it happens, but if you are correct there is much more to the splitter internally than I thought! Thanks for the insight!

 

[TonalDynamics]

A splitter instantiated on a channel should not cause an increase in level. Now, you can of course add a mixtool to each path to compensate if needed, but that can bring new issues of course. You can imagine that if there was no compensation of level, it would become very bad very fast in complex routings. I understand that in your use case it is not desired to have this level drop but for a correct working of the plugin it is absolutely necessary

Absolutely; some guitar splitters don't provide unity outputs as well.

Also in the limited testing I did, (correct me if I'm whacko on this point), but the attenuation for each leg in the split is done post-insert, i.e. the point just before the signals are summed together again?

Incredible info guys, much appreciated!

 

[SwitchBack]

Absolutely; some guitar splitters don't provide unity outputs as well.

Also in the limited testing I did, (correct me if I'm whacko on this point), but the attenuation for each leg in the split is done post-insert, i.e. the point just before the signals are summed together again?

Incredible info guys, much appreciated!

From a technical point of view there is no difference between attenuating every leg by x dB right before the summing or only attenuating the summed signal by x dB right after the summing. Since there's only one attenuation level for all I like to think it's the latter. Easier to picture too: It's like a mixer with all channel faders at 0dB and the main fader (which you don't get to see in the splitter) dropped by x dB.

 

[TonalDynamics]

From a technical point of view there is no difference between attenuating every leg by x dB right before the summing or only attenuating the summed signal by x dB right after the summing. Since there's only one attenuation level for all I like to think it's the latter. Easier to picture too: It's like a mixer with all channel faders at 0dB and the main fader (which you don't get to see in the splitter) dropped by x dB.

Right, but I'm assuming there's 10 plugins inserted on that leg, so is the attenuation PRE or POST those inserts is what I'm pondering.

I actually tested this a while back and forgot the results of the test, lmao.

Could've sworn it was POST inserts, but now I'm totally unsure...

 

[SwitchBack]

All plugins go before the little faders on the splitter's legs, the fixed attenuation definitely comes after these little faders, so POST.

Really, a splitter works exactly like a tiny mixing console with a stuck master fader. All channels/legs get the same line input signal and level. EQ, dynamics and whatnot comes next, then the channel faders, which sum into the stuck master fader. For 2 channels/legs the master fader is stuck at -6dB, for 3 channels at -9.5dB, for 4 channels at -12dB and for 5 channels at -14dB. Mystery solved.

 

[TonalDynamics]

All plugins go before the little faders on the splitter's legs, the fixed attenuation definitely comes after these little faders, so POST.

Really, a splitter works exactly like a tiny mixing console with a stuck master fader. All channels/legs get the same line input signal and level. EQ, dynamics and whatnot comes next, then the channel faders, which sum into the stuck master fader. For 2 channels/legs the master fader is stuck at -6dB, for 3 channels at -9.5dB, for 4 channels at -12dB and for 5 channels at -14dB. Mystery solved.

Well I've worked with several splitters that do have unity gain, but that is because they are tailored exclusively for multing guitar DIs as opposed to the console topology you've just very helpfully outlined.

It's good to have clarity here, thanks!