259 Sine?

phdinfunk · Joined: Jun 04, 2008 Posts: 110 Location: Taiwan

I'm testing the main oscillator. The f.m. REALLY works, and with THAT trannys everything is quite stable.

I found one mistake on original schematics. Instead CD4555 in the main oscillator, it should be CD4556 (with inverted outputs).

However, Sine output is not sinus, it is triangle. And from schematic I cannot see how it should be sinus output over the whole frequency range. What shape should it be?

--JP

JovianPyx · Posted: Sat Dec 29, 2012 6:58 pm Post subject:

I really wish I knew what a "259" was...

Please point us to the schematic and we can help Smile

elmegil · Posted: Sat Dec 29, 2012 7:15 pm Post subject:

A buchla 259... "Complex Waveform Generator"

There are 4 separate schematics on this page:

http://rubidium.dyndns.org/~magnus/synths/companies/buchla/

JovianPyx · Posted: Sat Dec 29, 2012 7:29 pm Post subject:

argh... Buchla Smile

Thanks elmegil...

ummm....

I gotta turn 'em all sideways so I don't get a bigger headache reading 'em...

gotta find where the sine is supposed to come out... holy crap...

elmegil · Posted: Sat Dec 29, 2012 8:08 pm Post subject:

On the "principle" oscillator, it's not too bad, it's the third output down on the right (after you rotate it of course).

Not that I'm taking on a buchla analysis without one of my own on hand Very Happy

JovianPyx · Posted: Sun Dec 30, 2012 2:11 pm Post subject:

The tri to sine shaper uses the Middlebrook & Richer JFET-based method. See this for some details: http://www.timstinchcombe.co.uk/synth/tri_sine/tri_sine.html (about 2/3 of the way down the page)

There are differences between Tim's site's drawing and the Buchla schematic, note that Tim's site indicates the two resistors connected to the JFET gate are called out as approximately 1 meg where the Buchla schematic shows 33K. This could be due to the exact JFET that Buchla chose. Tim's site indicates that the circuit uses the JFET's nonlinear nature and that somewhere near both triangle peaks, the JFET should be driven into cutoff and saturation. This means (to me) that the problem is likely related to the triangle amplitude not being great enough (but my assumption is that all of the components are the exact types and values called out in the Buchla schematic). If substitutions, such as the JFET type were made, then other components may require different values to get the JFET to operate in a nonlinear manner. Note that a JFET has a linear region which is why I suspect the triangle amplitude. If the amplitude of the tri is too low, it will cause the JFET to simply amplify the triangle (or copy it) and it would then remain unchanged.

I don't have personal experience with this circuit, so I could be talkin' outa my butt.

Tim Stinchcombe · Posted: Tue Jan 01, 2013 3:11 pm Post subject:

To get a low distortion sine out of the Middlebrook & Richer circuit requires precise trimming of the tri wave amplitude input to it. Buchla seems to have foregone doing this exactly, and instead settles for the JFET bit outputting something approximating a sine, which I presume is more or less guaranteed to be a 'rounded triangle', given the tri wave amplitude input and the choice of resistor values and the 'selection' of the JFET itself (again, presumably this is an Idss range of 0.7mA to 1.2mA as appears on the other similar 258/259 schemos). A fraction of the triangle is then subtracted out from this to get rid of the remaining peaks in the waveform.

I guess it kind of goes without saying that if you've not used a 2N4339 JFET you'll need something with similar parameters. If the circuit isn't working you'll have to trace back through the circuit with your scope: the JFET source should be a small-amplitude (+/-0.5V ?) sort-of-sine mentioned above, and this should appear at both inputs to the OA (it's negative feedback); check that the tri wave amplitude is near that expected (...?!), and also that the attenuated version at R151 is correct (and more '...?!' here, as it may require a little arithmetic!).

Tim

(And I wish I could find a decent amount of time to update that webpage and make a decent job of it!)

phdinfunk · Joined: Jun 04, 2008 Posts: 110 Location: Taiwan

I have been away for a few days, but that was incredibly thorough. I'll run through it backwards. I'm also going to try setting up a trim going into the j-fet....

As it stands, you're right, it's basically a rounded triangle.

Thank you both for a good lead on the problem!