FabFilter

Same issue in Protools as well. So far I've tested this in Cubase on PC and Protools on Mac.

Hi Snajand,

I also checked right now in Cubase and this also happened to me.
I was curious by this, so I dug a little deeper,
At the beginning I noticed that the smaller the bandwidth, the bigger the difference between the two tracks compression was.

When I set it to 30-450 the difference between them was smaller than 30-60, and like you said, when you take it from 30-30k there's no difference.

I also noticed though,that the difference comes from a kind of "auto gain" of the side chain filter.
If you take both instances of Pro-MB, and set the band's filter to "free" mode, (instead of band), and set the filter to max bandwidth (30-30k), the bands will now again work the same.

To make a long story short - when using the side chain, Pro-MB gives auto gain to the side chain path when you lower the bandwidth.
I really doubt that this is on purpose, I'd be glad to hear from the devs.

Cheers

Update:

I checked again, when you press "audition" to the both tracks while changing the bandwidth, both have the same behavior.
This means the side chain signal doesn't change (or auto-gained like I said above), but only the way the compression behaves.

This really is strange. :\
Why would it behave differently if it gets the same signal?

Another Update:

This difference in compression only happens when you use "band" mode.

If you set the filter to "free" mode, and set it manually to 30-60, there will be no difference in compression.

To Conclude:

If you reproduce the OP issue, but set the filters to "free" mode instead of "band" mode, everything will react as it should.

As a matter of fact, if you change the non-sidechained track to "free" but leave the sidechained one on "band" it will still work as intended.

It's like there's a little difference in "free" mode and "band" mode, but only when not using side chain :}

The way you guys describe it, it sounds like in external sidechain mode it defaults to not filtering the sidechain signal unless you set it to free, no? I'll play with it when I get home later today to confirm if they haven't replied by then.

Cheers,

Well, it's not happening here, W10 64bit, Studio One 3.5. I get the same compression in both cases, they actually null perfectly. Maybe the way sidechains are handled in Protools and Cubase?

Cheers,

Sorry, they don't null perfectly, but almost. There's a small difference (probably because of latency or something) but visually and the way they sound, they are pretty much doing the same thing.

Cheers,

I may have misread the previous comments but I just tested setting my sidechained Pro-MB to free band mode. I set the range to 30Hz-60Hz to match the compression band. It still compresses more than it should compared to non-sidechained mode.

With that said, if this is indeed an issue, I'd probably suggest not to fix it in this version of Pro-MB. Or maybe having an alternate version because any old projects that use the current version would sound wrong with a fixed version.

Hi everyone,

In internal triggering mode, Pro-MB splits the incoming signal into bands and uses the signal for the band to trigger on (if not set to free triggering). In external (sidechain) mode, we use the free triggering system, and set the low and high frequencies equal to the band's low and high frequencies. However, the free triggering system also contains an automatic gain compensator: if the band width is small, the gain is boosted automatically so the signal will have a reasonable level to trigger on. This explains why you get different results in this case.

As SNajand said, we can't change this behavior now because we don't want existing sessions to sound differently.

Cheers,

Thank you very much for your reply, Frederik!
The clear response is very much appreciated.

To give you more context on how I was trying to use the Pro-MB:
I had a Pro-MB on my master bus and wanted to replicate the same sound on multiple separate stems. So I copied the Pro-MB with the same settings but in Ext mode on all my stem busses and fed each of the instances the same full mix from the source stems.

For the time being I've manually adjusted the threshold of each band (I'm using 2 small bands 30hz-60hz and 60hz-120hz) to get it to sound as close as possible to what I was getting before hand.

Does the automatic gain boost trigger after a specific minimum band size is attained and is the boost gradual depending on the size of the band? In other words does it boost more and more as the band gets smaller?

Also, do the attack and release parameters' behavior change dependant on how the signal hits the threshold or based on how loud the source signal is (in this case the boosted source)? Given that the attack and release are in percentages.
I haven't yet been able to replicate the sound perfectly so that it nulls but maybe it's due to the content dependent attack and release, which could change based on the threshold changes needed to match the source input signal.

Thanks!

Thanks for the reply Frederik !
It is indeed a good idea to leave it this way for this version.

As for Snajand - it is strange the the "free" mode did not fix it for you, are you absolutely sure that all of the parameters are set exactly the same?
For me "free" mode made a full Null when I made a test like yours, and it could help you with your situation.

Yroth - Hm it is strange. I'll need to experiment more. Unfortunately I won't be able until a little while from now.

Just some extra findings below:

I've done more experiments and it seems like side chained audio that goes into a Pro-MB in Ext mode may also have different crossover curves.
I've tried to phase cancel what comes out in "audition" mode and to get the same signal (with small bands), even while adjusting in "free mode", seems impossible.

Thats a bit strange I must say, becuase it reacts differently for me.
Are you using the latest version of Pro-MB?

I am. I'm not really sure what's happening. Either way, now that I know how it's reacting on my system, I can compensate adequately and avoid using it in some ways that may become cumbersome down the line.