Wednesday, June 19, 2013

258J build with switchable TEMPCO resistor

I have long played with the thought of having a front-panel switch on vintage oscillators so that you could apply a TEMPCO stabilizing resistor for those very rare anomalies when you want less "analog" behavior for whatever reason. The vintage Buchla 258 dual oscillator series used carbon film resistors instead of the more accurate TEMPCO later found in the to me less-interesting Buchla oscillators. (Note that all Buchla analog oscillators used polycarbonate timing capacitors, which are similar to polystyrene, both of which are more stable than mylar or other film capacitor types. The Buchlas don't drift a whole lot, especially when used with a modern power supply; a limited tracking range is the main hindrance to their typically being useful for "melodic" music, although you can certainly set them up for that if you wish. Later versions "improved" tracking range. All of which is missing that the atonal exponential FM is the real attraction here...)

However, the 258J, based upon Mark Verbos' mutant 258C design, recommends a TEMPCO resistor. When I was asked to build a dual 258J set to mount on a Buchla-style front panel, I suggested the "Vintage/Modern" switch as a possibility. I'd have to build it and test it first to see if any foibles got in the way of making it actually happen. An SPDT would appear to work just fine, although there is an audible dip and return to pitch when the switch is thrown in either direction.

TEMPCO mounting:

Switched TEMPCO resistor

That's an SSM2220 with a pair of matched PN3565. The top of the SSM is heat-transfer greased and the TEMPCO sits across this. Note that both resistors are soldered into the pad at the SSM IC; the carbon film 2K is wired off to the switch and simply stands up off of the PCB. That's the orange wire. The TEMPCO needs to be in physical contact with the top of the SSM and the heat grease, but with its other end unsecured (not soldered into the PCB, but connected from there via wire to the switch), it simply moves around depending upon that wire's placement.

I noted a via (tiny through-hole) right in the curve of the top of the IC socket (first and last IC pins are there) and soldered in a clipped resistor lead, stood it up, and wrapped it around the heat-shrink-wrapped TEMPCO leg/wire and the carbon film resistor's wire as well, which effectively holds the TEMPCO horizontal and upon the prepared top of the SSM. Very cool, very serendipitous. The TEMPCO is the green wire and the black, the return from the DPDT switch. You can use an SPDT as well; I got DPDT in case the two resistors needed to be better isolated for any reason.

Also note that the heat-transfer grease will then get on and into everything else. :/ And that obviously, I've taken this picture before prettying up the PCB. Will remove the flux bits for delivery and likely replace this with a photo of a clean PCB.

Waveform Switch wiring:

258J Buchla clone Waveform switch wiring

As mentioned earlier in the blog, L-1 at a few forums discovered and published that the Buchla 258 oscillator could be built with all waveform resistors present, brought out to a 3PDT switch (front panel, placed in the hole originally used by the Fine Tune pot), to allow the user to select saw or square per oscillator. The vintage factory version (258A, B, and C) had one oscillator offering saw and on the other, square. This required you to have two dual oscillators to do a detuned saw pair. Now available in a single dual oscillator if you're up to doing the wiring. (A new hole must be drilled in the front panel for a Fine Tune pot if you want one.)

10V CV Modulator Range Modification

As is also mentioned earlier in this blog, Mark Verbos pointed out in his blog that early Buchla control voltage was 15V unipolar, travelling from zero to positive 15V. Half-way through producing the 200 series, Don Buchla set a new standard of 10V, which would then sweep a modulation destination across its entire range.

People building the 258J and some other Buchla clones such as direct copies of the 292 gate will run into not only that they are optimized for use in a system where the audio level is 1.2VRMS, but also that some designs may call for 15V or 10V modulation, and not all modern or even vintage systems output that much. The 258J's Waveform CV input is one such parameter which will need to be modified for use in a 10V system, if you're stocked with more late 200 or 200e modules.

Simply replace the 680K resistor at the input with a 402K or close value.

Front Panel:

Buchla 258 J3RK clone build

The switch directly above the large Frequency control selects between saw or square waveforms per oscillator. The switch above the gray banana (1V/Octave, calibrated via an added multi-turn trimpot) selects TEMPCO or carbon film resistors. A Fine tune control was added to the right of the Waveform switch using a Davies 1900 knob. The left of the set of three audio output jacks is from the triangle wave (buffer circuit added to the experimenter's "muck"). The second two are the sine to (alt) waveform per the original.

Note that I've used the Alpha 9mm panel-mount/PCB-mount pots here as they appear to be sealed versus the 16mm which are open (!). I've found that if you use a pliers to snap off the tiny orientation nub on the pot near the bushing, then place three washer rings on the bushing and then mount it to the panel, it will sit with just the right height to drop on the small Davies knobs with just a touch of space between them and the panel so they don't brush against it.
I may add more to this page at a later date.

Sunday, June 9, 2013

J3RK's 291 clone build

A Wiggler asked me to build a pair of these; I hadn't seen this board before due to having already etched several of Scott Stite's and Luka's modification to Mark Verbos' design.

I'd always had trouble with this filter and have recently found a partial culprit...a bin of parts at the local supplier, indicating .022uF which were in fact .0022. D'oh.

Using Scott's output buffer design to allow for lowpass and highpass modes in the boards "muck" experimenter's section (with the power most thoughtfully brought out next to it), I found that using a dual opamp resulted in an unexpected result: the highpass section, which is basically just a 1K resistor between one input and the output, put out 12V+ for no apparent reason, even with the input unconnected, even with the 1K resistor disconnected, using a 1458, 072, etc. No solder bridges, no mis-connections. The same on both boards/builds. I've ended up adding a single opamp IC solely for the highpass section and it works just fine. Go figure.

Both boards are now built and wired to a Buchla-style front panel. I've changed R39 and R43 to 47K (originally 68K). Both are to set the range of CV to Fc and to Width. As early Buchla systems used a 15V unipolar control range, and the later modules (281, etc.) output and accepted 10V (the 200e as well AFAIK), more range was required to allow a 10V system to fully sweep the intended control ranges. This was not a scientific substitution. It works. Depending upon the input signal, the filters will self-oscillate with bandwidth at minimum and resonance at maximum. The Width CV input allows self-oscillation at will.

I've opened the FM range through altering the limiting resistor to 220R instead of 47K. Haven't yet tested the results but am aware that the vactrols themselves limit (slew) fast signals so audio-rate change simply reduces to not much at all as the modulator frequency is increased.

The 258B schematic says that resistors with multipliers of 1, 2.2, and 4.7 should be 10% carbon composition. I've applied this standard to the build, using modern Kamaya 5% carbon composition 1/4 watt types (vintage Allen Bradley suffer from rising values beyond their tolerance range).

For example and reference, a Kamaya 470R resistor is mouser part number

As I may have mentioned earlier in this blog, I'd built one of Scott etc.'s clones and put in several carbon composition resistors and found a bit of a sonic difference. I'm not sure exactly what or why so I suggested that this build use as many as were possibly indicated. (This is nice because vintage CC often require the leads to be cleaned of dirt and age in order to solder well.)

I've complained at Muffwiggler because upon hearing the first build, it lacked the oomph, depth of frequency center range into the bass, and overall character of the 291 section I remember from the 212 Dodecamodule I once owned (and as also exhibited to an extent in the Scott etc. clones). Upon wiring up the second build, I find that (as some also found at MW), one build may sound very different from another. The second build is meaty; the first is thin-sounding. Going over it now to detect the culprit; has to be user error.

I noted that the better-sounding build did not have the two resistors modded as indicated in the MW thread, to 330R and 100R, so I've restored the other.

The MW thread indicates the use of .047 capacitors in place of the .022 (as well as the resistor change) to lower the overall Fc. I've settled upon .068 types and with resonance at maximum and bandwidth to minimum, it sweeps nicely down into the bass range, allowing the highpass to be used to fatten bass sounds and drums, without losing the upper range Fc.

I believe the original Buchla module used silver mica capacitors for the .022s. Will attempt to find .068 which aren't the size of a Buick...

I actually installed vintage Allen Bradley carbon comp resistors as indicated on the first build, and had to swap them out. The J3RK boards are great, fun to build, and durable, but as with many PCBs, cleaning out the pads of excess solder once resistors etc. have been removed can be a pain in the ass. I've since taken to simply using the drill press (#65 drillbit) to open the solder in any blocked pad(s). I'll be using this technique in the future instead of solder wick, the solder sucker, and even attempting to blow the heated solder out of the way.

I've added 1/2 watt 2R2 Allen Bradley resistors at the positive power input on each board, per the original. Photos of the original module indicate an even larger 2R2 resistor in grounding the front panel to one of the audio jacks...may attempt this once the build is finished.

The recipient wished for vintage sound so I've used a 741 at the input (per the original) and a 1458 (dual 741) for the CV section. The 074 remains; have not checked for a vintage plug-and-play substitute.

The rest of the resistors are 5% carbon film, as there are no 1% resistors listed in the schematic.

Overall, this is an excellent filter made ever sexier through bringing out the highpass and lowpass responses. Most recommended! (Photos to follow.)