[sdiy] Deep thoughts on Old drum machines: the "VCA"

[Mikko Helin]

Interesting story from the TR-808 makers here:
https://rc-808.com/episodes-of-the-mid-o-series/

On Wed, Nov 24, 2021 at 8:19 PM Richie Burnett
<rburnett at richieburnett.co.uk> wrote:
>
> Hi Benjamin and all,
>
> I've not looked at the Minipops schematics in much depth, but as someone who's spent considerable time delving into the details of other analogue drum machines like the TR-808, I would agree with everything you wrote. All of your reasoning is spot on.
>
> If you're just putting white noise through the "VCA" to have its frequency spectrum shaped downstream by a filter, the "VCA" doesn't need to have low distortion. White noise still has a flat spectrum even if it's sliced at some level by a hard-limiting comparator! ...and that is pretty much what happens in Roland's "representative swing type VCA."  Input signals above about 0.7v turn on the transistor pulling the collector voltage to ground, whilst input voltages below 0.7v leave the transistor turned off, and the collector gets pulled up and follows the envelope level.
>
> Even single tonal signals (single oscillators) can be put through quite a crappy "VCA" if you're prepared to accept the harmonic distortion and shape the spectrum as required afterwards. But individual oscillators need to have their own "representative swing VCA" to prevent intermodulation distortion! You can see this in practice with the TR-808 cowbell circuit that is composed of two differently pitched oscillators.  Each oscillator is fed through it's own single-transistor VCA, before summing them together and then being filtered to get the intended spectral formants.
>
> If they had attempted to mix the two different frequency oscillators together first and then put them through one single-transistor VCA the output signal would be polluted with a rich spectrum of "sum" and "difference" intermodulation products. The resulting dense discordant din is not what is required for the cowbell sound at all.
>
> However, a dense discordant metallic sounding din of seemingly unrelated frequencies is exactly what is required as the starting point for "convincing" cymbal sounds!  So in contrast to the cowbell circuitry, the Hi-hat and Cymbal circuits mix six differently pitched oscillators together and then shove the mix through one single representative swing type VCA. They suck up the intermodulation distortion and use it to create a rich spectrum of discrete spectral lines with only a handful of actual oscillators. This metallic din is then aggressively high-pass filtered to remove most of the  fundamental energy from the original oscillators!  Whilst perhaps not the most convincing Cymbal emultion by todays standards it is still a neat implementation using just a few cents worth of components!  (And rightly or wrongly a sound that is still held in high regard and sought after today.)
>
> Another thing worth mentioning here is control feed-through.  Something that the designers of real VCAs go to some length to minimise.  The single-transistor "Swing type VCA" circuit at best approximates a one-quadrant multiplier, so the audio input has to be biased, and the output therefore exhibits excessive feedthrough of the DC envelope signal to the output!  This CV feedthrough would normally result in a deep thudding transient every time the VCA envelope triggers.  However, if the "VCA" is followed by a bandpass filter as is often the case for spectral shaping of noise in drum sounds, then the gain falls away to zero at DC.
>
> Some trace of the DC transient (differentiated) still exists in the final output, but it's spectrum tends to follow the spectrum of the noise being gated by the VCA, so it get's masked to some extent.  At most it sounds like an explosive click at the start of a snare drum sound or hi-hat sound as you have commented.  And this can easily be claimed to be an attempt at modelling the stick impact on the drum skin or metal surface at the start of the sound! ;-)
>
> It really is amazing what those clever Japanese engineers managed to do with a handful of transistors, resistors and capacitors way back when these machines were designed.  They might not be the most realistic sounds by todays standards, but they have an enduring musicallity and tend to sit easily in a mix without much effort required to avoid clashing with other sounds.
>
> -Richie,
>
>
>
> ---- Benjamin Tremblay via Synth-diy wrote ----
>
> >I learned about drum machine technology in a random order, and here I am at the age of 54 studying the disco-era machines. I’m looking at the Keio/Univox snare and cymbal sound generators, and I’m stunned. They don’t even try to be a “VCA”; they just bias the amplifier to pass sound, and cut off the bias to make it stop.
> >This seems to be common in most transistorized drum machines from the 60s and 70s. Same for the PAiA drum sounds, and the drum machine in the old Engineer’s Notebook. I guess the philosophy is, “it’s just noise, what does it matter if it’s grossly distorted?” But that means a cymbal or brush sound must have be built lite this: NOISE—> GATE —> [Resonant High-pass Filter] because they often have a sweet and mysterious decay, which must be smoothed by the filter.
> >
> >The Minipops 5 I have been working on is as old as I am, but in much better condition than me. Both the Minipops 5 and the Univox JR-5 have this weird, obnoxious snapping snare drum sound, and it’s growing on me. It sounds like an IC is exploding on the board. I’m going to try to hack these tone generators and expose some knobs, but a slow decay using the bias-driven gate circuit will probably sound… different.
> >
> >Roland’s “Swing VCA” is just barely a real VCA: it’s like half of a current-controlled differential amplifier. It still distorts the sound a lot if you want to have a decent dynamic range, but at least it tries. Nowadays folks build drum tone generators using real VCA chips or OTAs, and that seems gluttonous, or glutinous.
> >
> >
> >_______________________________________________
> >Synth-diy mailing list
> >Synth-diy at synth-diy.org
> >http://synth-diy.org/mailman/listinfo/synth-diy
> >Selling or trading? Use marketplace at synth-diy.org
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> Selling or trading? Use marketplace at synth-diy.org