About
R2R is a special kind of resistor dividers setup that uses resistors of R and 2*R value. It has N inputs, which generally expect logic signals, and one output, from which we can take the unweighted mix of these signals. The MSB (most significant bit) input has the highest 'volume', and it gets divided by two until LSB (least significant bit) input. Basically, this is a part of how old DIY digital-to-analog converters worked, the ones you could connect to the PC to get vast entire 8 bits of signal resolution, which was still better than nothing. In my case, though, the ladder only has 4 inputs, so we get 4-bit resolution and maximum of 16 states. But also the circuit has no digital/logical components, so it can be used with any kind of signals, and will gladly serve as a general purpose unequal volume inputs mixer as well. After the R2R stage, there's a fixed slew limiter with a selectable set of capacitors, mainly aimed to antialias the resulting signal after the R2R mixup (make it smoother). This way, one can obtain a decent sawtooth by dividing some tone by 2 four times and just plugging the divided outputs, first division to LSB, and so on, until the last division to MSB. It can also process DC signals (not AC coupled) and serve as a general-purpose fixed slew limiter, or an antialiasing filter for bucket-brigade delays and similar devices.
Schematic
This one's a beginner friendly project there. Very useful, and very easy to build. Took me 2 hours to make the module from scratch, so a day of making is more than enough for a beginning SDIYer! Either way, first, we see the 2R2 ladder itself. 4 inputs, all get summed up, and buffered with half a TL072. The resulting mixup is passed through a fixed slew limiter: charging and discharging the capacitor through a variable resistance, that controls the charge time, and buffering the smoothed out voltage on the cap-pot node to the output potentiometer. On the photo, you can see a DIY vactrol instead of the fixed 1M resistor: i did this cause i only have 50K resistors, but the best would be to just use a potentiometer here. Or put a vactrol, and have CV over it! That was my original plan, but i ditched it for reasons. The bigger the pot, the bigger the slew rate headroom - so if you have a 2.2M one on hand, you may want to use it instead. I noticed that pots more than 5 megs act weird in such circuits, though, so it might be a good idea to stay inside these margins, though.
Media
Playing around by stacking the three suboctaves from the Subharmonic Divider in different orders, and trying out different antialias rates and capacitor settings. Sometimes it does indeed sound like a promised sawtooth, sometimes it merges into a more interesting and uncommon waveshape. Stacking octaves through an R2R is a good idea!
Antialiasing a Doepfer A-188-2 BBD, notorious for its clock noise and output crunch. First we listen to the BBD running at different delay rates without antialiasing, then i turn the antialias knob until the garbage is smoothed out well enough. The clock whine is still audable sometimes, but the antialiasing filter definitely does a great job bringing it to OK levels.
Clocking Shift Core Generator's bottom gate outputs at a slow rate while running the core at a very high speed, and using the four purely random gate outputs as four logic inputs to the R2R. The output is a neat 16-state random sequence, with an embedded glissando control! It controls the A-110-4 QVCO's pitch.
Pictures
