32C Channel Plugin

[madmaxmiller]

All right, day 3:

Subject:
Can I replicate a setting in Mixbus32C channel simply by setting the same dial to the same amount in AVA32C channel strip?

Possible outcomes:
1) If yes and yes (same dial+same amount)--> the two are most probably identical.
2) If yes and no --> this could reveal a gain bug suspected by others, you can still achieve the same result with the same ease but slightly different dial parameters (e.g. x dB more or less in the frequency band gain)
3) no + yes don't exist, really (amount is meaningless)
4) If no and no that means they are not the same at all (but still unique each, proven at lease for the AVA32C in my former findings and empirically found for the Mixbus32C in thousands of mixes all over the world)

So we are ready to go, here are the three signals:
Patchbay:
Signal feeds into a control channel in Ardour ("SIGNAL"), into a Mixbus32C utility bus ("Bus1") and into the AVA32C plugin.
The outputs feed into the Ardour channels "MB32Cv7" and "AVA32Cv1" which contain RTAs for signal check after the respective MB/AVA channels (or input check into the Ardour channels respectively).
The two channels feed into Ardour Master bus which does the summing and contains the RTA for the result.
All RTAs are derived from one instance ("SIGNAL") via drag & drop to ensure the exactly same settings for an optically correct result.

   

Left: original signal from TAP Pink Noise/Fractal generator
Middle: direct output from the Mixbus channel
Right: direct output from the AVA channel

   

Again, I isolate the EQ here only to eliminate complexity and stick to the original question: are the EQs the same and are they the same as any stock EQ? The results for AVA versus the Reaper stock EQ are in my other two postings, this is between Mixbus and AVA. Let's begin:

1) as always, a dry test first: both EQs OFF, one signal flipped, signals should be exactly the same, therefore it should "null"

SURPRISE!
- both signals come in at -6 dB, the straight sum shows logically -3 dB, all good
- BUT: suddenly there's a dip at 63 Hz! Where does that come from? Can only be a phase shift, right? And since this didn't happen when comparing the AVA with the ReaEQ I suspect the Mixbus32C channel being the culprit

Whatever I did, activating/deactivating one or both EQs, no matter which, the result stayed the same. This ends my experiment for today, because there's no point flipping phase, as we see.

The same area around 63 Hz seems overrepresented when 1 channel was flipped (sounds logical) while the rest appears unchanged - completely as if I would use pink noise from two different generators.

This again seems to indicate a statistical but random phase shift which accidentally comes close to 180 degrees around 63 Hz. Would it be a fixed amount phase shift we would see comb filtering in the sum. And this must happen *before* it hits any channel DSP.

sum straight

   

sum with 1 channel flipped

   

This of course gave me an idea: if this phase shift sits somewhere in the input stage, the the outputs of two channels in Mixbus32C should be exactly the same and the "damage" is done already and evenly. Let's check this:

Top: sum straight with 2 MB32C channels
Bottom: sum with 1 channel flipped IN ARDOUR, NOT in Mixbus! I also flipped around (haha) in Mixbus and both Mixbus and Ardour at the same time - as long as the sum was 180 degrees it nulled.

   

CONCLUSION:
Mixbus 32C and AVA 32C EQs cannot be compared directly and mathematically via null test as Mixbus changes the signal in the input stage. This somehow contradicts Ben's mentioning of a bug in the AVA plugin which's output in fact seems to be more phase coherent to the input. Also, as shown in my previous experiments, I couldn't find a gain error.

This phase shift seems to be statistical. Next step would be testing the Mixbus32C channel with sweeps and record the RTA to find out more.

I'll write Ben an email, Harrison may be interested, however that takes nothing away from Mixbus because if this is a systematic bug in the input stage it will occur the same way in all channels and if it still sounds good all is good. Or is this the

Secret Sauce ™ ? 

MMM

[allank]

This is a great post..

Quite timely as I only saw Dan Worrall's video yesterday.

He really should be putting a disclaimer that he has a conflict of interest given that he does videos for FabFilter on their channel.
Pretty horrific considering he is an educator in real life, he should be aware of these things.

As for his feedback.. He did have a couple of good points..

People are looking for a plugin that does colour the sound and in context of today's advertising language it's fair to expect people to expect that from the 32c channel strip plugin..

I would love a version of the plugin that added a "drive circuit" that emulated the distortion qualities of a 32c desk when pushed. (especially built natively into the 32c).. I think that would be an amazing selling point. Something similar to the tape saturation dials in Mixbus, that can be added in to taste without changing the level of the signal.

I also agree with his belief that a compressor plugin that had the same tonal quality but added attack, release, and threshold would be a fantastic little EQ.

That being said, in the way I use MB for mastering I love the AVA 32c plugin, because I can insert into buss.. That really just has to do with how I have arranged my workflow.

But .. he really should have made a disclaimer that he does work for Fabfilter.

[sunrat]

People are looking for a plugin that does colour the sound and in context of today's advertising language it's fair to expect people to expect that from the 32c channel strip plugin..

Did Harrison ever claim the plugin would colour the sound? I think they say it emulates the EQ characteristics of the 32C desk but not that it adds colour or saturation.
This appears to be the fallacious premise upon which Worrall's video and the whole subsequent shitstorm of discussion is based.

[Nameerf]

I don't agree with the statement that people use Mixbus for colouration - I use it for the analogue-like workflow/UI - the ease of use having dynamics, EQ and busses all set up etc. AND as an added bonus - it SOUNDS better than my mixes on other DAWS - whether this is a matter of "colouration" I'm no it sure - it could be the mixing decisions I make based on the aforementioned workflow? It could be something magical in the summing? It could be colouration - it doesn't really matter to me WHY. But I certainly didn't choose Mixbus FOR the colouration and I wouldn't feel duped or ripped off if it was scientifically proven there was no colouration in a particular Harrison plugin!

[madmaxmiller]

People are looking for a plugin that does colour the sound and in context of today's advertising language it's fair to expect people to expect that from the 32c channel strip plugin..

Did Harrison ever claim the plugin would colour the sound? I think they say it emulates the EQ characteristics of the 32C desk but not that it adds colour or saturation.
This appears to be the fallacious premise upon which Worrall's video and the whole subsequent shitstorm of discussion is based.

Gary has made that clear in his statement. The goal in the analogue world was to keep it as clean as possible. The fact that we are used to "dirt" is because the technolgy couldn't provide the clarity then and overcome the tape noise floor at the same time.
If you go into a symphony concert in a good hall you will indulge in clear sound, you don't want overloaded instruments except as an effect.
This "analogue" fashion has all become a bit hyped and people set analogue = distorted = good. It's now expected that an "analogue sounding" software has a distortion unit, no matter the context. They'll put "colour" onto everything.
It's like with spices: a little bit of the right spice here and there goes a long way, but putting your favourite spice generously onto everything is yuk.

My personal opinion is that it's easy to dial in a good tone with the Harrison EQs: Mixbus32C, AVA32C and also Mixbus. It's also easy to achieve the appropriate channel compression without separately controlling every parameter - for extreme control which you may need on a small number of channels there are plugins. I really do love the Input/Output section with the big meter in the AVA plugin

MMM

[Jostein]

I also agree with his belief that a compressor plugin that had the same tonal quality but added attack, release, and threshold would be a fantastic little EQ.

This is not a rant against you @allank, I'm just commenting on the point above as it's central in the video:

My biggest issue with this video is that he's missing what the parts of the plugin are about. The one knob (actually three) action of the compressors is a selling point. He also judged the leveler by listening to drums, which is ridiculous. 

And about the EQ, one of the points is the dynamic Q when you gain up/down a frequency, making a sharper Q. This EQ has a completely different philosophy when it comes to how one find and correct frequencies. And IMO, the hi-pass filter is great, it's so easy to get a clear mix in the bottom because of it.

IMO, when I judge by the video contents, he has no idea about why the compressors, filters, and the 32C EQ are designed the way they are and he just wants them to be like the stuff he's used to. The way the compressors and the 32C EQ are supposed to be used can not be compared to the other stuff he prefers in that video.  When he is unsure what kind of channel strip the plugin is for, he clearly shows that he does not know what he's reviewing.

[Clint]

While I do not own this plugin, I am glad that it is offered-
As I use Mixbus 32c for mixing now since version 5, I have often wondered what
I would have to do if I got into a situation where a band wanted to remix or something and wanted
to have a starting point in a DAW other than Mixbus.  
With this, I could say hand off a mix to my all Cubase friend and slap this plug on each channel and set things like they were in my 32c session and he could go from there.

Analog to me means “analogy of” and a screen picture of knobs and such with the functionality of the console (or the DAW analog of the hardware console) is an analogy.

Now if I had to simulate the tape drive saturation present in the busses and the master, I’d be in a little trouble, but hey.

[madmaxmiller]

Final post in this series of experiments:

To eliminate the stochastic phase shifting (or whatever it is, I have emailed Harrison about it) in the input stage of Mixbus32C I fed both signals into Mixbus: the original signal of the noise generator and the signal from the AVA32C channel in Carla and I summed them in Mixbus32C's master. Both again at -6 dB which makes a sum of -3 dB unflipped. Flipping one channel nulled as expected.

Experiment: set a frequency gain in AVA and try to match it in MB32C using ONLY the SAME two DIALS.

I don't have time for screenshots etc today, but the findings are:
- it's rarely completely nulling
- it's possible to get very close using only the same two dials
- using more than the two doesn't improve the result, it makes it worse (more difference)
- with both mids at a positive/negative gain and high and low at shelves or bells with +/- gains it was always possible to closely match both curves with a rest noise (difference) between -50 and -40 db
- the frequency/gain readings in both EQs were slightly different, so you can't set one to certain values and replicate the result in the other by setting these same values

Conclusion:
It was always admitted by Harrison that the numeric values are not exact (can't be). Debates at this topic are in the forum.
It was always possible to match settings closely (difference for the respective frequencies in the -50 to -40 dB area) using only the same corresponding dials. Perhaps even better, I did these tests very quickly.
Therefore the characteristics of both EQs are the same.
So I believe it can be said that the modelling is the same. The fact that it not completely nulled in all settings (although in some, down to -120 db) could very well point to the gain bug and the noise floor bug Ben mentioned, we have to wait.

Anyway, I hope some of you enjoyed the nerdy stuff, see you next time.

Cheers,
MMM

[tedma]

Final post in this series of experiments:

To eliminate the stochastic phase shifting (or whatever it is, I have emailed Harrison about it) in the input stage of Mixbus32C I fed both signals into Mixbus: the original signal of the noise generator and the signal from the AVA32C channel in Carla and I summed them in Mixbus32C's master. Both again at -6 dB which makes a sum of -3 dB unflipped. Flipping one channel nulled as expected.

Experiment: set a frequency gain in AVA and try to match it in MB32C using ONLY the SAME two DIALS.

I don't have time for screenshots etc today, but the findings are:
- it's rarely completely nulling
- it's possible to get very close using only the same two dials
- using more than the two doesn't improve the result, it makes it worse (more difference)
- with both mids at a positive/negative gain and high and low at shelves or bells with +/- gains it was always possible to closely match both curves with a rest noise (difference) between -50 and -40 db
- the frequency/gain readings in both EQs were slightly different, so you can't set one to certain values and replicate the result in the other by setting these same values

Conclusion:
It was always admitted by Harrison that the numeric values are not exact (can't be). Debates at this topic are in the forum.
It was always possible to match settings closely (difference for the respective frequencies in the -50 to -40 dB area) using only the same corresponding dials. Perhaps even better, I did these tests very quickly.
Therefore the characteristics of both EQs are the same.
So I believe it can be said that the modelling is the same. The fact that it not completely nulled in all settings (although in some, down to -120 db) could very well point to the gain bug and the noise floor bug Ben mentioned, we have to wait.

Anyway, I hope some of you enjoyed the nerdy stuff, see you next time.

Cheers,
MMM

Thanks for your work on this. I actually did enjoy reading your posts. Best
Ted

[Overmann]

I think the amount of negativity surrounding this plugin is a genuine problem. I've been a mixbus user for years now and whenever I work in a different DAW I'm tearing my hair out. Mostly due to the workflow, but I also love how easy it is to get a descent sounding mix without adding a bunch of plugins to every track.

Having said that; When Harrison release a plugin that is said to be a painstakingly modelled (like-new) 32 console channel strip while sounding indistinguishable from a fully digital plugin that is a problem. I understand Ben and Gary's arguments that these consoles were made to be as clean and transparent as possible. They know a lot more about these things then I ever will and based on my correspondance with Ben in the past I have no reason to doubt that he is both sincere and truthful in his defence of this product. Ben is a standup guy! But I've downloaded this demo, just out of curiosity, and I really think they made a mistake releasing this plugin.

It doesn't SOUND analog. It adds no harmonics, distortion or character to sound. Not in the filters, EQ or the compressor-modes. It's suuuper clean. Like a digital channel strip. And while I DO trust the people at harrison I don't believe that a brand new 32C would sound THIS sterile. Also, not being able to push into it and get some of that juice is a missed opportunity. The Honey3 channel strip from Acoustica does this and it really makes a difference. The way the 32c channel strip plugin from Harrison is set up, and sounds, make me question if the code to fascilitate this sort of thing is even there. As I would expect it to be if every component truly was modelled. If this was a true model of a 32C channel strip then it would have more character then a digital plugin, surely?

If harrison was to market this as a good sounding digital plugin with a Harrison-like layout then I wouldn't have a problem with it, but as it stand I think they should revisit this one and give it a proper overhaul, and a free upgrade to existing costumers. Add gain controls, harmonic distorition and some real character to it. That is what people expect this to be, and what the marketing implies. But in reality it is a collection of SUPER clean sounding EQ, compressor and filters that says "Harrison" in look only. Indistinguishable from a stock digital EQ.

EDIT: Please take care to note that I am not saying that this is a bad sounding product, or that it is not a good go-to plugin for getting a basic track sitting well. It could very well be excellent for that, but that is not the selling point of this plugin. The selling point is built on the legendary sound of the Harrison Series 32 console. In order to not fail as a product this plugin needs to at least make an effort to sound like that console, and to give users a way to harness the sound of it. To impart that sound onto the tracks. And this plugin REALLY doesn't do that. It adds no discernable character, and has no way for the user to push into the hardware. That is to add or decrease the amount of analogueness. Those are the problems.