I built a Buchla 259 Complex Wave Generator module for someone else. They sent me a complete kit of parts and I assembled and tested the module. Many of the components are sourced through Mouser but specialized parts, the panel, knobs, and PCBs have specific sourcing and I do not know the details. This module is in the Buchla format with V1.0 PCBs.
Buchla 259 Complex Wave Generator description from the Buchla Synthesizer User Guide, November 16, 1981. This document doesn't describe the autotune switches which disable the FM input for tuning. I would be interested in what autotune is supposed to do. The clocks of IC11 and IC15 are connected to ground so these parts do nothing.
I made reference designators from the PCB images. The components cover the silk screen legends once populated which makes it hard to debug if anything is wrong.
PCB 1 Front Reference Designators (includes trimmer identification)
PCB 1 Rear Reference Designators
PCB 2 Reference Designators (includes trimmer identification)
There is a single build thread and no modifications listed for the V1.0 PCB.
259 V1.0 Complex Wave Generator build tips
My Modifications and tips
Front Panel build
Build time for PCB1 and the panel was 3.75 hours. This is PCB1 ready for the final front panel installation (the image is a composite of two separate photos).
I use color coordinated wire for the banana jacks although there is only the single blue jack.
The tabs on the 9mm Alpha potentiometers contact the PCB so I cut the top and bottom tabs off of Pot4, Pot11 and Pot13 since there are runs directly beneath the tabs. I do not like relying on solder mask for insulation.
The square pins do not fully seat into the headers with the height of the standoffs. I insert the square pins fully into the female header and assemble the PCBs together. The pins are flush with the PCB and fully inserted into the header so the alignment is correct in all directions and then soldered in place.
Build time for PCB2 was 7.5 hours. The resistors take a long time to install since there are so many of them and they are spread all over the PCB. I use Kester .020" diameter #24-6337-6401 331 water soluble core solder so take time every hour to wash the PCB. The trimmers, vactrols, polystyrene capacitors, and power cable are added after washing. I epoxied the vactrols onto their respective transistor array and used SIP socket posts for R*.
There are two 2x15 headers on the rear of the PCB.
There are a number of calibration adjustments but the front panel trimmer nomenclature is a bit misleading. For the principal oscillator the first adjustment to make is the Low Linearity. In fact this is low frequency symmetry. The trimmer adjusts the current mirror so the rising and falling slopes of the core triangle are equal. Without this, at lower frequencies the period of the triangle will shift and the first octave or so will be out of tune. This image shows an out of symmetry principle oscillator. The rising slope of the triangle is controlled by the expo pair while the falling slope is controlled by the current mirror. Adjust the low linearity so the slopes are the same. The rising slope will not change so simply adjust the falling slope until the pint is just before the second division.
Now that you have calibrated the symmetry, I adjust the offset for 27 Hz with the coarse control at minimum and the fine control at mid. Then I adjust the frequency to 30.0 Hz, connect a 6.0V source to the keyboard input, and adjust trim for a 960 Hz output. I typically monitor the square wave output as it usually provides more stable measurements. Then I check the accuracy at 1.2V, 2.4V, 3.6V, and 4.8V. The high linearity trim sets the scale of the coarse control. It isn't very accurate but I typically set the coarse control to 440 Hz and then adjust the high linearity to this frequency. The frequencies listed are not very accurate but it is nice to have 440 Hz somewhat accurate.
The modulation oscillator adjustments work the same except on some 259 modules the low frequency waveforms can be quite distorted and I have Modifications below to address.
Calibration and full checkout of all inputs and outputs took about 2.5 hours. I calibrated the modulation oscillator at the center of the frequency control to 440 Hz (high) and 4 Hz (low). In high range the minimum frequency was 32 Hz and the maximum frequency was 8 KHz. The waveforms are well formed. This image is a composite of three scope images showing all three waveforms at frequency control mid setting.
In low range the waveforms are a bit more distorted but generally pretty good. This image is a composite of three scope images showing all three waveforms at the frequency control mid setting.
As you go to the very low frequencies the waveforms distort quite badly. At the low end of the frequency control the oscillation stops. This image is a composite of three scope images showing all three waveforms at the frequency control min setting. I made modifications to significantly improve the waveforms. Scope images of the improvement are in the Modifications section near the bottom of this page.
This image shows the difference in levels and offset between the modulation output and the CV output. The modulation output is a +/-2V output while the CV output is a 0 to 10V output.
I calibrated the principle oscillator at the center of the frequency control to 440 Hz. The minimum frequency was 29 Hz and the maximum frequency was 9.8 KHz. The sine wave has a bit of flattening. I made modifications to improve the sine shape. Scope images of the improvement are in the Modifications section near the bottom of this page.
At the maximum frequency the square wave shows some leading edge rounding.
This image shows the amplitude modulation.
This image shows the frequency modulation.
This image shows the timbre modulation.
This image shows the principal oscillator with all modulation modes on (phase, amplitude, frequency, and timbre). It produces quite a complex waveform.
This video shows the principle oscillator phase lock capability to an external oscillator.
I did some modifications to improve the flatted top of the sine wave (left image). I changed R110 to 787R (1K paralleled with R110)which rounded the top nicely (right image). The sine trim TR1 adjusts the gain a bit more with this mod so I could adjust the sine amplitude to match the square amplitude. This is a much better sine wave.
This image shows the principal oscillator sine and square outputs.
The modulation oscillator waveform distortion on the low range was quite significant. IC4C switches in a different bias point for the expo converter on low. I made some modifications to improve the match of the charge and discharge current sources but they negatively affected the waveforms on the high range. I decided instead to use IC4C to switch in a higher value integration capacitor which works very well. This image is a composite photo of the modulation oscillator lowest frequency before and after triangle wave (scope horizontal and vertical settings are different between the two and the top waveform is AC coupled). The resulting symmetry is very good..
Here are three scope images of the modulation oscillator output and the signal output at the lowest frequency of 1/3 Hz. The signal signal output is capacitively coupled so there is droop in the waveforms. However, all are much improved with a near perfect duty cycle.
The forums mention a "bleed" of the modulation oscillator in the principle oscillator with all the modulation switches off. This module had a slight "bleed" that I could hear it as a timbre change in the square wave output. I could not hear it in the sine wave or final output. I investigated the principle oscillator circuit and found that the square output would vary in width depending on whether the modulation oscillator output was high or low. I took a scope image of the principle oscillator output with the modulation output high and inverted and added it to an image with the modulation oscillator low so similar colors become gray and unique colors stand out. You can see the difference in width circled in green and is about a 100 µS delta.
I decoupling some of the logic ICs (IC6, IC8, IC9, IC10 and IC11) in the modulation oscillator with a 0.1 µF capacitor across the IC power pins. This helped minimize the "bleed" so it was barely perceptible but I could still see some pulse modulation on the scope.
I investigated further and found the +15V shifting about 20mV depending on the polarity of the modulation oscillator output. This shift is independent whether the +5V is internal or external. This module appears to be very sensitive to +15V variations so I changed to a dedicated power supply with short cables and could no longer discern any "bleed". My recommendation is to make sure you have a well regulated power supply with short cabling for this module. I didn't analyze the power distribution on the PCB but suspect some modifications in the distribution and isolation of power routing would certainly improve performance.
259 Second Build
I built a second 259 for another customer. This customer supplied SMT tempcos which I also epoxied to Q1 and Q5 and wired them to the PCB.
I took some time to further investigate the bleed between the oscillators. The architecture is such that the AM modulation is always enabled but the wet path is broken with an analog switch when turned off. I surmised that perhaps there was still AM modulation that might be impacting the power supplies that was causing the bleed. I looked at moving the analog switch from the wet signal path to the modulation index path but on this build I could not see or hear any bleed between the oscillators with all the modulation switches turned off. This build used LF412 op-amps which are not my favorite, no additional decoupling capacitors, and a Power One HDBB-105W-A with a 12" power cable and the +5 jumper set to external. The HDBB-105W-A is a large supply with 1.5A of +/-15V and 12A of +5V so it is well regulated and this seems to confirm my earlier experience that well regulated supplies are the key to minimizing bleed.
259 Third Build
The low frequency modulation oscillator waveforms looked very symmetrical on this module so I did not do the modifications. However, I could not tune it to 2V/Octave with the stock resistors.
The Principle Oscillator square output showed some oscillations on the rising edge. You can barely see them in this scope image outlined in red.
Increasing the sweep speed to look at the rising edge shows the oscillations.
U25 is a comparator to generate the square wave output from the triangle. It has no hysteresis which is simply bad design practice for a comparator. I lifted the triangle input, pin 2, and added a 10K series resistor. The other end of the 10K can connect to the anode of D9. Then I added a 1M feedback resistor from pin 7 to pin 2. The resulting square wave output is now clean.
I have had to make a number of tweaks to various 259s that come to me for repair.
Sometimes I need to lower R107 to dial in a nice sine shape. I have gone as low as 6K8.
Sometimes I need to lower R217 to dial in the AM modulation. I have gone as low as 6K8.
The AM Index way over modulates the signal. The "dry" signal is subtracted from a 100% AM modulated "wet" signal to generate the AM modulation. A vactrol-based VCO controls the amount of "wet" signal. The variation in vactrol performance will basically ensure that it is never correct. On many, 100% modulation occurs as early as a quarter turn. After this, the "wet" signal simply increases in amplitude thus generating a large distorted signal. I have adjusted R229, sometimes with two series resistors, to dial in 100% AM modulation at full CW rotation. Values I have used are 1K, 910R+62R, 2K4+330R. This composite scope image shows the Final output with original resistor values on the left and modified on the right. These scope images show how over modulated the final output is where the ramp actually creates a triangle-shaped modulation and the triangle doubles the MO modulation.
Both V1 and the vintage version contain parts used for the auto-tune circuit that isn't fully implemented. There is a CD4040 counter IC in both oscillators (MO is IC11, PO is IC15) which are powered by 15V. The ICs driving the CD4040 and the ICs that the CD4040 drives are powered by the 5V reference (not +5V). As such the CD4040 output voltage is limited by the clamp diodes in the CD4030 which just burns excess current. I don't own a 259 so I have never experimented with just removing the auto-tune parts that do nothing. I simply lifted pin 12 and added a 49K9 series resistor to limit the current on both parts. In addition, pin 11 of the CD4040 is reset and is driven by parts operating on 5V. For 15V, Vil is < 4V and Vih is > 11V which isn't being met with a 5 volt part. The CD4040 reset is in the high state is in the undetermined voltage which can also increase power dissipation. Fortunately the reset input is always low.
V2 Vintage Version
V2 reverts to copies of the original two PCB design. This image from the original reference diagram details the interconnect cable dimensions.
This photo shows the interconnect cables which allow the PCBs to lay flat for debug.
This portion of the reference diagram shows the locations of the three trimmers, the power supply connections, and the pin 1 (red stripe) designator for the cables.
This original design uses 1K tempcos R22 and R122, both marked with stars. R122 is associated with exponential pair Q5 and R22 is associated with exponential pair Q1. These tempco resistors compensate for the thermal component inherent in the exponential pair transistors. For best thermal performance the tempco and expo pair should be at the exact same temperature. It is common design practice to have them physically touch with some thermal grease between them but this vintage PCB layout simply has them in close proximity.