Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
32C Channel Plugin
#1
Ben asked me to chime with a new thread in the General Discussion area where we can continue the conversation.

From the 32C Channel press releases and our website.

"Predating the modern parametric EQ, the 32Series equalizer had to serve many purposes and remain easy to use.  The middle 2 bands use a proportional-Q design:  small gain changes provide for gentle classical-style tone controls; but cut a little harder and the filters become sharper, allowing you to hunt down and then notch-out troublesome frequencies. The high and low bands are a shelving design with a huge gain range, allowing you to add dramatic punch and shimmer to any source. If you needed more surgical work, however, you could switch the high and low bands to operate as additional proportional-Q 'bell' filters. The effort put into voicing this EQ was unmatched at the time, and its signature remains on some of the biggest records ever made.

The 32C Channel Plugin provides a complex emulation of the original Harrison 32C EQ. Every resistor, capacitor, and transistor
is included in the model.  When combined with the 32C Filter section, a huge range of tonal possibilities are available with just few knob turns."

Now as we have said publicly in various places...

We do not add distortion or noise to the EQ.

Instead of being focused on artifacts found in old gear, we focused on the original work that went into voicing the EQ at the time. As mentioned above the 32C EQ predated "parametric EQs" and it was very difficult to make that EQ do a lot of work with just the two knobs per band. The feel of the knobs, the depth of the proportional -Q action, the speed of the filters relative to the knobs etc. etc. etc. was a serious trial by error. Those refined circuit designs (yes resistors, capacitors, transistors etc.) are what we included in the mathematical model.

Also note, that the distortion and noise measurements of those original consoles (WHEN THEY WERE NEW) were extremely low and virtually unaffected even with the EQ in circuit. The entire point back when there were ONLY analog consoles was to make them as clean, with the lowest possible noise and distortion, lowest possible cross-talk, and as phase coherent as possible. Harrison was not alone. Every analog console manufacturer was after the same thing/s. Distortion, noise, cross-talk, phase issues etc. etc. were all the enemy. We spent many man years tweaking those console circuit designs for optimum performance.

All of that... is what we built into the mathematical model of the Mixbus32C EQ and now the 32C Channel EQ. As with the original console EQ design, this was no easy task. It caused a lot of headaches and issues here because the mathematical model was built on the knob resistor values to get the tapers and behaviors correct rather than using off-the-shelf Bi-quad math.

In our collective opinions, there is nothing wrong with our marketing and we stand behind what we make and offer just as we have done for the last 45+ years.

We think the 32C Channel is a excellent plugin and it stands on it's own as a great mixing channel strip, and a good value as well.


Thanks,
Gary
Reply
#2
Thanks for the insight Gary.
Things I like about the 32C are: how fast I get the sound I'm after, the Q factor, bump and the Leveller which is just so handy and easy to setup.
Macmini 8,1 | OS X 13.6.3 | 3 GHz i5 32G | Scarlett 18i20 | Mixbus 10 | PT_2024.3.1 .....  Macmini 9,1 | OS X 14.4.1 | M1 2020 | Mixbus 10 | Resolve 18.6.5
Reply
#3
Thanks Gary. I especially liked


Quote:The entire point back when there were ONLY analog consoles was to make them as clean, with the lowest possible noise and distortion, lowest possible cross-talk, and as phase coherent as possible.



That can't be told often enough.

Cheers, MMM
Linux throughout!
Main PC: XEON, 64GB DDR4, 1x SATA SSD, 1x NVME, MOTU UltraLite AVB
OS: Debian11 with KX atm

Mixbus 32C, Hydrogen, Jack... and Behringer synths
Reply
#4
(09-14-2021, 02:29 AM)bachstudies Wrote: If you can null 32C EQ against a stock digital EQ there is absolutely no analog magic going on whatsoever. And it's not complicated DSP that couldn't be achieved five years ago (as categorically stated in the website).

Dear Mr "bachstudies",

you obviously rely on random Youtube videos without any questioning, let alone own experiments. But your (and that Youtuber Dan's) claims of "nulling" are simply wrong.
Before you call me a "fanboi" - let me present you with my own findings.


1) Test arrangement

1.1) Used devices
To eliminate any possible influence of the Mixbus-DAW I have not used it throughout the experiment, instead I used a neutral plugin host for noise generator and 32C channel strip and used Reaper only as a ReaEQ host. So I really match the two EQs and the EQs only.

- Carla plugin host runs TAP Pink Fractal Noise as a sound source
- Carla plugin host runs Harrison AVA 32C channel strip as one of the test candidates
- Reaper v6.15 runs 1 channel with ReaEQ and direct output (not through Master)
- Ardour 6.5 serves as mixing/phase flipping device for the actual null test
- 4x Robin Gareus 1/3 Octave Spectrum Display for visual analysis

1.2) Connections
I chose the shortest possible connections to eliminate any possible influence through internal buses, master buses etc. I also didn't place the RTAs directly into the signal channels to avoid any possible influence of these plugins. I'm measuring the raw signals as they come from the signal generator and the respective EQs.

- noise out --> 32C channel in --> 32C out --> Ardour ch1 (AVA32C) in
- noise out --> Reaper ch1 with ReaEQ in --> Reaper ch1 direct out --> Ardour ch2 (ReaEQ) in
- noise out --> Ardour ch3 (SIGNAL) RTA in (RTA for original signal)
- 32C channel plugin out --> Ardour ch4 (Harri RTA) --> RTA post 32C EQ
- Reaper channel (with ReaEQ) direct out --> Ardour ch5 (ReaEQ RTA) --> RTA post ReaEQ
- Ardour channels 1 and 2 --> Ardour MASTER for summing, all other (RTA) channels have no master connection
- Ardour Master has an RTA for the sum signal

Screenshot 1: test arrangement
   

1.3) Checking the test arrangement
- bypass all plugins
- run noise source
- check meters

We see: noise comes in and out at -6.3dB evenly on all channels except the master channel which logically shows -3.3dB as a summing result. All RTAs show identical pictures. Flipping one phase results in a perfect null as it should be with everything EQ bypassed.

Screenshot 2: signals
   

Screenshot 3: NULL at bypass
   

We are ready to go!


2) The test

2.1) both EQs switched on but all dials on neutral. In the AVA32C plugin ONLY the EQ is used, all other stages like HP/LP and compressors stay bypassed - the result should be null.
- meters appear unimpressed
- phase flip nulls perfectly - as expected
in neutral positions nothing should happen

2.2) second check of the test arrangement: If I change 1 EQ while in phase flip there should be a difference
- checked alternating with both (only one engaged at a time) and it works

Screenshot 4: only 1 EQ altered
   


2.3) trying to dial in equivalent settings
- lower mid 1kHz +3db --doesn't null, lowest result was when moving the bandwidth slider to 2.72 oct at the ReaEQ
- higher mid 3kHz -3db - lowest result at 2.82 oct bandwidth in ReaEQ but never null
- both above combined and Q-tweaked - never nulled
- changing channel levels in ardour always resulted in higher differences, as expected, look at the meters!

Screenshot 5-7: both EQs set to same values and ReaEQ's Q set to lowest difference
   
(sorry the other two screenshots are in the next post because of forum restrictions)

3) Conclusion: I have diligently built my test equipment and method to eliminate all possible flaws. Please correct me if you find error.
The tests showed that even with very moderate settings it is not possible to achieve a NULL between the 32C channel EQ and ReaEQ, as claimed by other parties. The EQs differ in frequency characteristics and I expect them to differ even more with more advanced settings. In all tested cases the differences were not only visible in the RTAs but also very well audible, around -57 dB. This was tested with pink noise, a real musical signal will show more difference in character.
Therefore I suggest that either the test arrangements or the methods or both in the mentioned Youtube videos were flawed if not manipulated.

All used tools or similar are freely available and everyone is invited to try themselves.

Happy Mixbusing!

MMM
Linux throughout!
Main PC: XEON, 64GB DDR4, 1x SATA SSD, 1x NVME, MOTU UltraLite AVB
OS: Debian11 with KX atm

Mixbus 32C, Hydrogen, Jack... and Behringer synths
Reply
#5
The missing two sreenshots from my former post:
   
   
Linux throughout!
Main PC: XEON, 64GB DDR4, 1x SATA SSD, 1x NVME, MOTU UltraLite AVB
OS: Debian11 with KX atm

Mixbus 32C, Hydrogen, Jack... and Behringer synths
Reply
#6
(09-14-2021, 08:33 AM)bachstudies Wrote: I think you missed the point made in Dan Worrall's video: The reason they don't null is because the gain settings as input in the 32C EQ are not accurate. When you add say +3dB of gain, you are actually getting more than that.

I'm sorry, that is a 'bug'.

The original console had tickmarks around the knob that were labeled up to 12dB, and the labels were for the 'bell' mode. The gain range for 'shelf' and 'bell' modes are different, but I don't see the value changing when I switch to 'shelf' mode. The Shelving mode provides up to 18dB of gain but the display only shows 12.8, the same as it does in Bell mode.

The default value for the bands is 'shelving', leading to the confusing situation where the console tickmarks (and in this case, the displayed text) don't match what you are hearing.

It is not uncommon to have a few bugs in a 1.0 software release. Luckily this looks like a 'display' bug and shouldn't have any effect on save/recall.

We will investigate further and I'll report our findings here and in the AVA forum.

-Ben at Harrison
Reply
#7
P.S. I've also noticed that the 'noise floor' of this plugin is completely wrong, compared to our other AVA plugins and Mixbus. This suggests that it might have been released without a gain+dither stage. The difference in sound will be vanishingly small but we will likely update the plugin to match our other AVA's and Mixbus32C.

Thanks for bringing these issues to our attention. Software development is tricky and we appreciate feedback from our customers to get it right!!!

Best,
-Ben
Reply
#8
(09-14-2021, 08:49 AM)Ben@Harrison Wrote: P.S.  I've also noticed that the 'noise floor' of this plugin is completely wrong, compared to our other AVA plugins and Mixbus.  This suggests that it might have been released without a gain+dither stage.  The difference in sound will be vanishingly small but we will likely update the plugin to match our other AVA's and Mixbus32C.

Thanks for bringing these issues to our attention.  Software development is tricky and we appreciate feedback from our customers to get it right!!!

Best,
-Ben

Haha I'm happy to repeat this experiment in Mixbus 32C. I expect no different results though.
MMM
Linux throughout!
Main PC: XEON, 64GB DDR4, 1x SATA SSD, 1x NVME, MOTU UltraLite AVB
OS: Debian11 with KX atm

Mixbus 32C, Hydrogen, Jack... and Behringer synths
Reply
#9
Side note: some bickering posts have been temporarily hidden. If I missed something important then I can un-hide it. -Ben
Reply
#10
What if you drive it a little harder. So put in a slightly stronger signal…

But that's just one component of the original console. I guess no one thinks that it alone can get 32C sound. That's why we have the Mixbus 32C.

Personally, I rely on Harrison's statement. It's just a channel plugin. Other candies can be found in the Mixbus 32C DAW. But definitely quality stuff.
Small recordingstudio in Finland countryside. Mixbus 10 Pro, AvLinux AVL-MXe 23.2, Rme UFX+, Rme 802, Adam A77X, Genelec 8020c, Genelec 7050b, Yamaha HS7



Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)