Sunday, May 29, 2011

Buchla 258B Remake Project

This page will be a work in progress just like the build itself. I anticipate at least one more revision of the PCB.

Please note that Don Buchla has not replied to my emails regarding this project and sharing the new PCB artwork publicly or in any way, so at this time, it is not for anyone except myself; I hope that these notes are helpful to any interested in such things. To the destructive people reading this, please go and do something positive/constructive instead. HA! Okay love you, bye bye.

I have had a gent create new PCB artwork/layout from the Buchla 258B dual oscillator schematics. I've once owned this version of vintage Buchla module and prefer it over the later "C" version as the "C" is a bit too stable.

Added to the original design:

-L-1's 3PDT switch for selecting square or saw for the Waveshape section (added to the sine at the main output). Works just fine; can also simply install jumpers to hardwire one of the two types. Very clever, very fun. My preference is always sawtooth and I would have built all of my oscillators as such, but with the simplicity of a switch, I'll certainly now be using square waves to see what I can see. I envision Buchla Dub music (grin). Kudos to L-1!

L-1's drawing:

-Triangle output buffer. The triangle is a great addition to this build. Can't wait to try some exponential FM with them. It is however separate from the main output so it currently cannot switch between sine/tri. I had hoped to be able to switch between the sine and this triangle, in the main output signal path and not have it as a separate output, thus avoiding repatching etc.

Keeping with the original:

-Using 15uF Kemet tantalum capacitors for decoupling the positive rail, only. The original has no negative rail decoupling.

-Kemet 1uF tantalum capacitors for the exponential FM path.

-Carbon Composition resistors. The schematic states that all 2.2, 4.7, and 1.0 multipliers should be carbon composition resistors, 10%. This includes the (sole 1/2W) 2.2R and 10R power couplers, the 2K2 and 470Rs in the audio path at the Saw/Square selector, the 470K and 1K 1% in the CV section, a couple of 10Ks, and the 10M and 4.7M on the front panel CV and Fine pots, respectively.

The 1K 1% and IIRC a 100K 1% are a problem; I've just received 100 quantity 1K 5% and only two were just at 2%, and all were above the nominial 1K value, not above and below. No joy there, but the same effect allows a pair of 470K(-ish) resistors to serve as a 1K 1%. Two 47K should provide the 100K 1%.

I note that early Moog oscillators have CCs littered about seemingly randomly, comparing module aside module. This might have simply been an issue of supply at the time they were building. The parts bin may have just had both types of the same value. The pic of the Buchla 258C circuit board in my recent "Vintage Parts in the J3RK Build" post on this blog shows a number of CCs, but neither 258C schemo I have states anything about certain values being part of the design (or not).

Swapping out the 470Rs in the sawtooth circuit gave the circuit, to me, a slightly better feel. I don't know what change in sound will occur, but I note that guitar amp and high-end tube EQ and compressor builders swear by carbon comp resistors. Me, I do use my sense of "feel" in selecting music gear. I have made some good desicions, if anyone is interested. Having been screaming about Buchla gear for perhaps two decades now... It certainly goes far further than that, but that's a topic for another blog. I'm just sticking to Don's delineation.

Why be anal to any degree regarding such things as obsolete, badly-behaving tantalum capacitors? Again, I've owned the original and who knows what combination of items contributes to the's not for me to say, but I am interested in further knowledge if it shows up.


The uA726 and 2N3802 are becoming Unobtanium Siliconis, and the 2N4339/2N4341 etc. are as well. I note success in the Verbos/J3RK PCB project in applying J201 for the 4339 and MPF102 for the 4341; will one day build one of these dual oscillators using them.

Pads/Switching for the uA726 and TEMPCO resistor

It shouldn't hurt a uA726 to have the heater leads (positive and negative 15V) run out to a DPDT switch, and another switch between a regular 2K resistor and an equivalent TEMPCO, for those rare times you actually want an oscillator to not move. Combinations of these should be interesting...

Control Voltage Response/Scaling

I'm told, haven't confirmed, that the original is 2V/Octave. My first version and etchings are this design.

Linear FM

Listening to the J3RK board, I find that having eschewed the 258C's linear FM has been a mistake. It's not like linear FM on other oscillators, and I should have known this from the "Buchla" name alone. Will determine the correct connection to add this to the "B" design.

Stronger Exponential FM

While checking the 106 mixer outputs for inverting/non-inverting behaviours, I bypassed the standard FM in and merely clipped an amplified oscillator output to the positive modulation input of another. The modulator was sawtooth, the carrier sine. It produced, when the pich of the sine was swept, an effect slightly similar to sweeping the clock rate of certain digital noise sources, most commonly heard in the 1980s video game Tempest when leaving one level and approaching the next, the legendary Atari POKEY integrated circuit. I am very happy to have "stumbled" upon this and will likely add a switch between "normal" and "depth" exponential FM, probably adding a 1R or as required to avoid shorting to ground at zero index.

More Waveforms

The J3RK build offers a pure square wave ripe for buffering, if one is interested in having another "perfect" waveform in their system, as opposed to the mutant square available at the main output. I'm interested in the partial saw wave as depicted in the Verbos schematic. And having it and the triangle on a switch to interrupt the sine at the main output. Perhaps even the pure sine waveform found at the sine shaper FET, but all would have to be selectable to the main output, not as individual outputs.