The BBE 422A arrived on Tuesday, as promised, which is more than can be said for most things promised on a Sunday. On Wednesday we plugged it in and measured it. By the end of the day there were over 100 recordings made: tones at every level, noise, sweeps, and a great deal of knob turning. Today, we added a bunch more.
Last time I said a few of those measurements were there only to catch somebody lying. Here's how that went.
The manual's text says the middle of the Lo-Contour knob is flat. The calibration table at the back says it adds about a decibel and a half at 50 Hz. The table wins. Channel A adds 1.8 dB there and channel B adds 1.4, so the two channels don't entirely agree with each other either, but neither of them is flat.
The bypass switch should be a plain wire. It very nearly is. From 100 Hz up it's within a tenth of a decibel of doing nothing at all, and it only starts to lean on the very bottom octave, about two and a half decibels down by 20 Hz. So the unit was on, and so were we.
One thing nobody mentioned anywhere: Definition at zero isn't neutral. With the knob all the way down, the box takes about 3 dB off the top end. Turning it up gives the highs back first and then keeps going.
The same, but different
The two channels are the same circuit, but that is not the same thing as saying they behave the same. Channel B's Definition knob comes on later than channel A's. At the same click, B gives you less sparkle until you get near the top of the dial. We'd only measured B's knob at the ends and the middle, so today it got the full treatment, every few clicks, and The FortyTwo now has a separate curve for each side, the way the real one does.
Auto, and the trouble with tones
By Wednesday night The FortyTwo was making real sound, built from the measurements instead of the stand-in. Manual mode matched the hardware almost straight away. Auto took longer.
Auto is the mode where the box listens to your music and decides how much sparkle to add. All of our measurements of that listening were done with tones, and the plugin matched every tone we threw at it. Then I played the same piano through both, and the plugin was brighter. Not by much, but enough.
The problem turned out to be that a tone only asks the detector one question at a time, and none of our tones had asked it what it thinks of the very top end. Noise asks everything at once. So today we fed the box noise at a ladder of levels, and it told us: it doesn't listen up there nearly as much as we'd assumed, and it takes a few milliseconds to react instead of jumping at every peak. With that in the model, the brightness dropped to match the hardware.
There was one more thing making the plugin sound better than the hardware, and it had nothing to do with the plugin. My comparison setup was playing the hardware about 6 dB quieter than the plugin. Louder almost always sounds better.* Once the levels matched, the comparison got a great deal more honest, and a great deal harder.
Where it stands
The same piano, recorded through the box and through the plugin at the same moment, now comes out within about a third of a decibel of each other from 40 Hz to 24 kHz, on both channels, in both modes. They're close enough that I have trouble telling them apart.
It isn't finished. We haven't measured what happens when you drive the box hard, which is the part where old hardware usually gets interesting, and there are a few smaller things on the list. But the part that took the longest is done.
No sign of the question yet.
Also simmering
Also simmering on a back burner: a little tool for the Listen bus. You put on the headphones you mix with and nudge a band of noise left and right until it sits dead center in your head, then do that again at a handful of pitches. From that it builds an EQ that balances those headphones to your two ears, so something panned center actually stays there. It only changes what you hear while you're mixing, never the mix itself. It doesn't have a name yet, because I just dreamt up this tool (literally) last night while I was sleeping.
* This is the oldest trick in audio sales. It turns out it also works on the person doing the selling.