How to add a linear FM input to Doepfer's quadrature oscillator and get the weird waveshapes of Buchla's 258 VCO.
The A-143-9 quadrature oscillator is a favourite for FM. Its sine wave is pure and the DC-offset is minimal. This makes it the perfect choice as the modulator in dynamic FM patches. It's a shame then that it itself doesn't have a linear FM input. Fortunately, the fix is simple and the result sounds great, especially in quadrature 'stereo':
Most analogue VCOs are linear at heart. To get the response we need for musical pitches, CVs need to be translated. If we bypass the part of the circuit that does this - the exponential converter - we should be able to modulate the VCO in a linear fashion.
This depends on the actual circuit design but, happily, this 'dodge' is possible with the A-143-9. I asked Dieter Doepfer and he identified pin 6 of the top quad opamp (TL084) as the target (see A-126 connection PDF for board layout). A resistor determines the modulation depth. Dieter suggested 100K but I ended up using 68K to allow for some over-modulation.
There is one caveat with this simple modification, but it's a limitation that can also be used to musical effect. Pin 6 is also connected to -12V via R8 (270K). This provides the current needed for the oscillator to work. If the modulation voltage 'robs' the VCO of this current - i.e. at 0V - the oscillator will stop. This is good and and bad: at low audio-rates, over-modulation could sound choppy. On the other hand, we gain the ability to gate the VCO/ LFO on or off, similar to a track-and-hold.
The reason FM-ing the A-143-9 sounds so good is that there's no waveform converter messing things up. Even with the best schemes, the sines on triangle- or saw-core VCOs show their heritage by being either brassy or buzzy. So, a sine-core makes sense if clean FM is a priority. Question is, what do you do if you also want a saw or square?
The latter is easily generated with a comparator. For the saw, we can take inspiration from the Buchla 258's 'funny waveshapes'. We can use a triangle-to-saw converter circuit or a patch:
Patch the sine (0) output of the A-143-9 to a mixer. Patch a multiple of the sine to another mixer (e.g. mh01). Set the channel to 50%. Mix in a 2.5V offset and patch the result to the signal input of a polarizer or DC-coupled ring modulator like the A-133. Patch the cosine (90) to a comparator (A-167, Sport Modulator etc.). Set its threshold to 0V. Patch the comparator's output to the CV input of the A-133 and send the result to the first mixer. Now balance the amount of original sine to get the shape you want. Altering the comparator threshold will yield other shapes. Here's how it can sound when modulated:
It's not thru-zero, but the quality of the A-143-9's sine-on-sine linear FM is clean and stable. Given the right C:M ratio it can cover some of the sounds generated by wave-folding. With a little more circuitry to provide a permanent saw/ square output, a pair of these quadrature VCOs could form the basis of a budget timbral VCO.