Two plugins claim to model the same classic compressor. One sounds like the real thing and one sounds like a digital compressor with a bit of saturation. In Episode 8 of the Knobs Podcast, Simon Moro and Terry Hart get into why analog emulations are so wildly inconsistent, and how to tell which ones are actually worth your money.
The short version: the label “modelled on” hides two completely different approaches, and only one of them makes a plugin behave like the hardware it is named after.
Not All Emulations Are Created Equal
Simon walks through his own journey with UAD, from skeptic to convert. For years he argued the DSP was an expensive dongle, since modern computers could clearly run the software, and he abandoned the platform. Coming back recently, now that the plugins run natively, he put the UAD SSL channel on a vocal in place of his usual one and immediately heard it. It performed and responded like analog: you can crank the top end hard and it takes it, the way the real console does.
Impulse Response vs Circuit Modelling
Terry explains the split. Impulse-response and algorithmic emulation measures what goes into a unit and what comes out, then approximates the difference, a bit like a Kemper guitar profile. The trouble is you are feeding it sweeps and tones, not music, and we have no good way to measure how a dynamic circuit behaves over time. Circuit modelling takes the opposite path, rebuilding the actual electronics component by component.
“It’s not telling it what an 1176 sounds like. All it’s doing is saying this is what the component does to voltage.”
That is the whole ballgame. A release circuit is literally voltage draining through a capacitor at a rate set by real resistors, not a line of code that says “15 milliseconds with this curve”. Model the components and you get the behaviour for free.
The Pretty Line Isn’t the Whole Story
You can even see this gap on screen. Compare a Pultec to an EQ clone that matches its curve exactly and the envelopes on a spectrum analyser look identical, yet they do not sound the same. There is far more happening than that pretty line: the way real components interact, the small non-linearities, the parts that never show up on a meter. Measuring a device at an instant, however precisely, is not the same as capturing how it works. Circuit modelling is not flawless either, mind you. Transformers are the one thing even the best models struggle with, because the magnetic field around them is so sensitive that a slightly different mounting screw changes the sound. But everything else can be modelled remarkably accurately.
Why Compressors Are the Real Test
This is why the guys rate circuit-modelled dynamics so highly. An EQ can be captured reasonably well by reading it at an instant, because it is a snapshot. A compressor is a time-based creature, and approximating it from the outside almost always falls short. The clearest example is the DBX-160: the Waves version sounds like a thumpy attack with some lo-fi on it, while the UAD one has something else going on that is more than the sum of its parts. Night and day.
The Self-Deprecation Trap
There is a sneaky psychological cost to using the lesser tool. When your mixes never quite have the magic, it is easy to assume you just do not have it in you. Then you swap in a genuinely better plugin, the fogginess lifts, and you realise it was never you. It is the same feeling as discovering aliasing for the first time. Good tools do not just sound better, they stop you blaming yourself for a sound the software could never deliver.
Manage Plugins Like Scarce Analog Gear
Circuit modelling is expensive to run. Because every component is being calculated, turning on more modules in a channel strip uses more CPU, and Simon promptly maxed his out and went hunting for more processing. The upside is that it pushes you to treat these plugins like a rack of real hardware: save the special compressor for the thing that matters, rather than stamping it across every channel out of habit.
The Solo That Lies
One of the most useful lessons is that solo tone and mix behaviour are different things. In a shootout of a real EMT 140 plate against the UAD and Waves versions, the real unit actually came last in solo, because it is mono and noisy. In the mix, it won by miles. Plates and gold foils rarely sound pretty on their own, yet they take up all the space in a mix without taking up any space. Judge tools in context, not soloed.
When You Don’t Need Analog at All
Not every plugin needs a circuit behind it. Tools like TrackSpacer and intelligent resonance suppression are doing something specific and genuinely new, and there is no vintage box they are pretending to be, so a clean algorithm is exactly right. The warning is aimed elsewhere: thanks to coding frameworks like JUCE, a lot of “new” plugins are the same built-in EQ, compressor and gain modules in a different skin.
“It’s really no different from you just putting up the stock EQ followed by Dyn 3.”
Building a plugin from a framework is not the same as designing a piece of gear. As Terry points out, his own Federal-inspired compressor came from poking real capacitors on a breadboard until it sounded right, not from changing a number from 10 to 30.
Final Thoughts
The practical takeaway is to look past the pretty interface and ask how a plugin was actually made, especially for anything with character or dynamics. Buy the circuit-modelled compressors, use clean algorithms where you want a job done rather than a vibe, and stop paying for twelve versions of the same thing. And be patient: these circuits can be modelled exactly, so as computers keep getting faster, real-time circuit modelling may one day turn a lot of expensive hardware into very nice doorstops. Both guys sound ready for that day.
