[sdiy] Buchla 291 analysis?

Pete Hartman pete.hartman at gmail.com
Sat Oct 10 08:32:18 CEST 2015


Bah, what I get for half asleep posting.  That graph is before I got the
cutoff down where I want it to be.  But it does illustrate the issue of
full bandwidth shifting the frequency and being a much softer slope.

On Sat, Oct 10, 2015 at 1:29 AM, Pete Hartman <pete.hartman at gmail.com>
wrote:

> So I'm still beating my head on this.  I found some changes that have
> tamed the buzziness etc, actually mostly to do with better grounding
> practice (originally was grounded only through the panel), and a couple of
> component tweaks.  Also matching expected audio levels of Buchla gear more
> closely helps as well.
>
> The remaining problem is that the lowest setting of the cutoff knob comes
> nowhere near the "spec" value of 30Hz.  It's more like 700Hz in the
> original circuits I was working with.  I've also since gotten another
> single 291J pcb that I can play with a bit more easily, and I find that if
> I increase R22 to 4M and C2 & C5 to ~55nF, I can easily get my cutoff down
> where I expect it to be.... but it's at the expense of the resonance.  High
> bandwidth/low resonance settings get a realllly soft knee that is a few
> steps higher in cutoff frequency than when I have lower bandwidth/higher
> resonance; the peak / knee shifts quite a bit (from 150Hz at max resonance,
> to 200Hz at moderate high resonance to 400Hz at no resonance... though that
> latter is squishy because of how soft the knee is).
>
> http://elmegil.dynathome.net/~elmegil/IMG_20151009_223326015.jpg
>
> (obviously that highest resonance setting is oscillating too)
>
> I tried simulating this in iCircuit, but unfortunately it seems their op
> amp models (or something) are crap for this kind of circuitry, I've found
> that it doesn't even come close to resembling what goes on in the real
> circuit.
>
> I've tried experimenting (on my own board) with various other component
> values, but I can't seem to find anything that will let me get both the
> reasonably good resonance curves of the "stock" configuration and the
> expected bottom cutoff frequency.  When I'm more awake again tomorrow I
> think I may try simply running the CF pot from +V to a resistor going to -V
> so I can find a lower voltage than ground that will take me all the way
> down to 30Hz without mucking up the other circuit curves.  Using a resistor
> as a voltage divider with the pot to set a bottom to it.
>
>
>
>
> On Fri, Aug 28, 2015 at 12:21 PM, Walker Shurlds <walkershurlds at gmail.com>
> wrote:
>
>> Here is my work so far: https://www.flickr.com/gp/pancreas/Dk1sF8
>>
>> I was treating the FETs as ideal op-amps for my analysis anyway, so no
>> change unless there's a resistor value I missed.
>>
>> The middle block's large capacitor creates a pole that actually moves
>> woth the CV in a bizarre way, when Rx is high,  there is a low-sheld filter
>> effect ... but for less than a dB and for frequencies below 1Hz. Gonna have
>> to assume that the cap is there for other reasons only.
>>
>> Other than that, it looks like if we ignore the 470p cap, all the
>> branches are just positive feedback around the two main blocks, the second
>> of which has one moveable pole and one pole at zero.
>>
>> There's one path that lacks capacitors--from the output through the 6.8k
>> to the middle stage through the 10M||X to the third stage, so *most* of the
>> loop gain depends on that. I haven't done the math yet. All of the other
>> paths only really affect the loop gain at particular frequencies. Looks
>> like the output to 22nF to trimpot path is to boost the loop gain of
>> frequencies above some minimum only ... and the 470p was added so that
>> above some maximum frequency it isn't boosted *too* much.
>>
>> I'm completely ignoring the fact that the capacitors in the loops
>> influence the phase margin / gain margin to change the gain needed for
>> resonance anyway.
>>
>> Anyway, that's my take on it, so far.
>>
>> (Your most recent message just came in. As for your problems ... we could
>> put it in spice and see if you het the same results. I've had suspicious
>> amounts of success with making analog filters in spice reveal things.)
>>
>> -Walker
>>
>>
>>
>> On Fri, Aug 28, 2015 at 11:52 AM, Richie Burnett <
>> rburnett at richieburnett.co.uk> wrote:
>>
>>> A very odd schematic indeed!
>>>
>>> There's some strange things going on there, like connecting the outputs
>>> of
>>> op-amps 1 and 3 together with a direct capacitive path via C2 & C3. I'm
>>> not
>>> surprised that someone has written on that schematic about needing
>>> additional resistors to tame HF oscillations!!!
>>>
>>> -Richie,
>>>
>>>
>>> -----Original Message-----
>>> From: Pete Hartman
>>> Sent: Friday, August 28, 2015 4:35 PM
>>> To: Walker Shurlds
>>> Cc: Synth-Diy
>>> Subject: Re: [sdiy] Buchla 291 analysis?
>>>
>>>
>>> The Stroh schematic is here:
>>> http://home.comcast.net/~r3cogniz3r/291j.pdf
>>>
>>> Major difference that jumps out at me is the use of an op amp instead of
>>> the
>>> pair of FETs at the start of the main block and after the main block.
>>> The
>>> one after is obviously just a follower for buffering, not so sure about
>>> the
>>> role of the FETs/op amp at the start of the section.
>>>
>>> Thanks
>>>
>>> Pete
>>>
>>>
>>>
>>>
>>> On Fri, Aug 28, 2015 at 10:12 AM, Walker Shurlds <
>>> walkershurlds at gmail.com>
>>> wrote:
>>>
>>> I think I changed my mind: the first op-amp doesn't seem to have a pole
>>> that
>>> actually moves with the control voltage, so probably not a SVF.
>>>
>>>
>>>
>>>
>>>
>>> On Fri, Aug 28, 2015 at 9:47 AM, Walker Shurlds <walkershurlds at gmail.com>
>>>
>>> wrote:
>>>
>>>
>>>
>>> I'm looking at this one:
>>> http://www.synthfool.com/docs/Buchla/Buchla_200/291/Buchla_2910.jpg Is
>>> it
>>> different than the one you're looking at?
>>>
>>> It looks like SVF to me, but you have to remember than the "states" (I'm
>>> about 90% sure I'm really abusing the control systems terminology here,
>>> but
>>> bear with me) include all three possible outputs, all three possible
>>> inputs,
>>> every possible branch...
>>>
>>> So to me it looks like *both* the input and output are in different
>>> nodes
>>> than in the common tow-thomas and KHN topologies. And there seem to be
>>> extra
>>> branches added for stability. And C6 is making everything unnecessarily
>>> complicated, analysis-wise. Really tempting for me to analyze it instead
>>> of
>>> listening to my assembly coding professor lecture for the next hour.
>>>
>>> Hope that helps,
>>> Walker
>>>
>>>
>>>
>>>
>>>
>>> On Fri, Aug 28, 2015 at 1:52 AM, Pete Hartman <pete.hartman at gmail.com>
>>> wrote:
>>>
>>>
>>>
>>> Are there any analyses anyone's aware of about how the 291 works?
>>>
>>> Looking at the Dustin Stroh schematic, it looks *sort of* like a state
>>> variable filter, but not quite, so that throws me off a bit. Maybe it's
>>> just that the schematic is oddly drawn, I think the last time I asked a
>>> question like this it was a matter of re-arranging the bits to see it :)
>>>
>>> Trying to determine if what I'm seeing with one is "normal" or not, and
>>> with
>>> Buchla circuits, I can't just fall back on "does it behave like all
>>> these
>>> other filters I have". :)
>>>
>>> Thanks
>>>
>>> Pete
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at synth-diy.org
>>> http://synth-diy.org/mailman/listinfo/synth-diy
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> No virus found in this message.
>>> Checked by AVG - www.avg.com
>>> Version: 2014.0.4830 / Virus Database: 4365/10529 - Release Date:
>>> 08/28/15
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20151010/5b052201/attachment.html>


More information about the Synth-diy mailing list