10-27-2022, 01:08 PM
(This post was last modified: 10-27-2022, 04:43 PM by Ben@Harrison.)
I’m often asked “what makes the 32C equalizer special?”. Usually this is in comparison to a full-featured digital EQ that provides a full suite of features including selectable shapes, unlimited frequency ranges, spectrum displays, and other conveniences of modern digital plugins.
The shortest answer I’m able to provide is “the 32C knobs and buttons match the knobs and buttons that hugely famous people used, while mixing hugely famous records”. Not many EQs can make that claim!
Understandably though, people want to know more. Why did famous engineers use the EQ on so many records? It would be disingenuous to say that the ‘magic’ in those records was the EQ. We know those songs were crafted by the extremely talented artists and engineers, provided with extreme 80’s budgets. We were lucky enough to be chosen as the brand that they used to make those records. But … this still leaves the question: what is special about the 32 series EQ?
Functionally, the 32C EQ was designed for mixing. Rackmount equalizers were commonly available, but they were not convenient. By adding EQs directly into the console strip, you’d find the mid-eq-knob is adjacent to the next channel's mid-eq-knobs, and so you can boost one while cutting adjacent channel(s) to make room for that sound. Perhaps this is self-evident. But I like to remind people that there’s a ‘reason’ for the vertical layout.
Behaviorally, the exact combination of potentiometers (resistors) and capacitors made the EQ facilitate the mixing process. The 32C pre-dated the fully parametric EQ and there was no dedicated ‘Q’ knob. Turning the gain would affect the Q in a fairly complicated way. There were some other oddities as well: the shelving mode has a much higher gain than the bell mode. And finally the filters were notable for their ability to sweep extremely wide, to the point they would overlap.
If any 'magic' can be ascribed to the 32C EQ, it has to be this behavior of the circuit. The electronics were fairly simple and didn't involve any tubes, transformers, or opto-elements that might impart something mysterious and un-definable. You might say that the circuit was 'bog standard'; except that of course it's only bog-standard in hindsight. At the time it was a pretty big deal.
“Resistors and Capacitors”, oh my!
During the original development of the 32C equalizer for Mixbus32C, I asked our lead DSP designer: “will it impart a magical halo around all my music and make everything sound better?” His answer was something like “of course not, it’s just an EQ, it’s up to the user to find the halo (if there is such a thing) and make it louder”’
As the product manager, this left me without much ammunition for our marketing text. We knew the EQ was good, because many engineers had used our analog-modeled digital EQs, and made great records. My response was: “well, that’s not much help; how am I supposed to tell people that this is a really good analog-modeled eq?” Chris’s answer was: “you can say that we modeled every resistor, capacitor and transistor in the circuit, because that’s literally what we did, using best-in-class engineering practices.”
Modeling the analog circuit “resistors, capacitors and transistors” is a well-known process among dsp engineers and it results in a filter with exactly the same ‘poles and zeroes’ as the original analog circuit. This means it will have the perfect analog magnitude and phase response, subject to the funny limitations of digital sampling, the funniest of which is “you can’t have any frequencies above half the sample rate”
That last bit is tricky. It means that there’s a brickwall cutoff at nyquist, and this affects the feedback loops inside the digital EQ. The result is an inherent, analog-like effect called “cramping” where the eq curve is progressively squeezed as the frequency increases. It’s pretty minor compared to the drastic effect that occurs at nyquist (i.e. everything gets cut off!). But some people assign a lot of importance to cramping. There are methods to de-cramp the model: some (like oversampling or resampling) add digital artifacts like quantization-distortion or latency… we didn’t want to do that. Other methods involve throwing out the R-C model and taking a different path entirely. We chose to stay with the industry accepted analog-style modeling.
This leads to the funny situation where our “resistor and capacitor” text has generated some controversy online because the cramping is seen as a deviation from the analog circuit; when in fact it is exactly this “resistor and capacitor” statement that requires the curve to cramp (at lower sample rates)
Luckily there was a pretty easy solution which sidesteps the cramping issue entirely for most users. When Great River created their 500-series module, they chose to add a ‘shelving’ button to the high band. The original 32C console didn’t have this shelving mode, but it was a very common mod which only required bringing a switch to the front panel. We borrowed that same modification for the 32C DSP, and chose this as the default operation (matching the behavior of the low band and, coincidentally, the original Mixbus) and this means most users will never experience the cramping effect, even at normal sample-rates; and it didn't require faking the RC network math to artificially overcome a limited bandwidth. As a first-order filter, the shelf isn't affected by the nyquist limit.
The 32C has evolved a bit over the years, but it remains “the EQ that was used in a bunch of huge records.” We like to use it in our newest products (both analog and digital) because it covers a lot of the most common mixing tasks with just a few knobs. And it doesn’t hurt that there’s a great story to go with it.
There are many, many equalizers available today, in hardware and software form. It will be interesting to see if any of those designs survive as well as the 32C EQ, and spark as many licensed and derivative products 50 years after their introduction.
Best,
-Ben
The shortest answer I’m able to provide is “the 32C knobs and buttons match the knobs and buttons that hugely famous people used, while mixing hugely famous records”. Not many EQs can make that claim!
Understandably though, people want to know more. Why did famous engineers use the EQ on so many records? It would be disingenuous to say that the ‘magic’ in those records was the EQ. We know those songs were crafted by the extremely talented artists and engineers, provided with extreme 80’s budgets. We were lucky enough to be chosen as the brand that they used to make those records. But … this still leaves the question: what is special about the 32 series EQ?
Functionally, the 32C EQ was designed for mixing. Rackmount equalizers were commonly available, but they were not convenient. By adding EQs directly into the console strip, you’d find the mid-eq-knob is adjacent to the next channel's mid-eq-knobs, and so you can boost one while cutting adjacent channel(s) to make room for that sound. Perhaps this is self-evident. But I like to remind people that there’s a ‘reason’ for the vertical layout.
Behaviorally, the exact combination of potentiometers (resistors) and capacitors made the EQ facilitate the mixing process. The 32C pre-dated the fully parametric EQ and there was no dedicated ‘Q’ knob. Turning the gain would affect the Q in a fairly complicated way. There were some other oddities as well: the shelving mode has a much higher gain than the bell mode. And finally the filters were notable for their ability to sweep extremely wide, to the point they would overlap.
If any 'magic' can be ascribed to the 32C EQ, it has to be this behavior of the circuit. The electronics were fairly simple and didn't involve any tubes, transformers, or opto-elements that might impart something mysterious and un-definable. You might say that the circuit was 'bog standard'; except that of course it's only bog-standard in hindsight. At the time it was a pretty big deal.
“Resistors and Capacitors”, oh my!
During the original development of the 32C equalizer for Mixbus32C, I asked our lead DSP designer: “will it impart a magical halo around all my music and make everything sound better?” His answer was something like “of course not, it’s just an EQ, it’s up to the user to find the halo (if there is such a thing) and make it louder”’
As the product manager, this left me without much ammunition for our marketing text. We knew the EQ was good, because many engineers had used our analog-modeled digital EQs, and made great records. My response was: “well, that’s not much help; how am I supposed to tell people that this is a really good analog-modeled eq?” Chris’s answer was: “you can say that we modeled every resistor, capacitor and transistor in the circuit, because that’s literally what we did, using best-in-class engineering practices.”
Modeling the analog circuit “resistors, capacitors and transistors” is a well-known process among dsp engineers and it results in a filter with exactly the same ‘poles and zeroes’ as the original analog circuit. This means it will have the perfect analog magnitude and phase response, subject to the funny limitations of digital sampling, the funniest of which is “you can’t have any frequencies above half the sample rate”
That last bit is tricky. It means that there’s a brickwall cutoff at nyquist, and this affects the feedback loops inside the digital EQ. The result is an inherent, analog-like effect called “cramping” where the eq curve is progressively squeezed as the frequency increases. It’s pretty minor compared to the drastic effect that occurs at nyquist (i.e. everything gets cut off!). But some people assign a lot of importance to cramping. There are methods to de-cramp the model: some (like oversampling or resampling) add digital artifacts like quantization-distortion or latency… we didn’t want to do that. Other methods involve throwing out the R-C model and taking a different path entirely. We chose to stay with the industry accepted analog-style modeling.
This leads to the funny situation where our “resistor and capacitor” text has generated some controversy online because the cramping is seen as a deviation from the analog circuit; when in fact it is exactly this “resistor and capacitor” statement that requires the curve to cramp (at lower sample rates)
Luckily there was a pretty easy solution which sidesteps the cramping issue entirely for most users. When Great River created their 500-series module, they chose to add a ‘shelving’ button to the high band. The original 32C console didn’t have this shelving mode, but it was a very common mod which only required bringing a switch to the front panel. We borrowed that same modification for the 32C DSP, and chose this as the default operation (matching the behavior of the low band and, coincidentally, the original Mixbus) and this means most users will never experience the cramping effect, even at normal sample-rates; and it didn't require faking the RC network math to artificially overcome a limited bandwidth. As a first-order filter, the shelf isn't affected by the nyquist limit.
The 32C has evolved a bit over the years, but it remains “the EQ that was used in a bunch of huge records.” We like to use it in our newest products (both analog and digital) because it covers a lot of the most common mixing tasks with just a few knobs. And it doesn’t hurt that there’s a great story to go with it.
There are many, many equalizers available today, in hardware and software form. It will be interesting to see if any of those designs survive as well as the 32C EQ, and spark as many licensed and derivative products 50 years after their introduction.
Best,
-Ben
