[sdiy] Oberheim Xpander VCF

Fri Jan 16 22:58:14 CET 2009

Hello,

The precision/matching of summing resistors is very important to obtain correct
filter shapes... Say the matching should be in the range of 0.1% or better. 1%
-Xpander schems -matching is really average to obtain the good filter curves !

Summing resistors are not the only error source here :  gain/level of each of
the filters cells must be matched  too. A very little error or difference here
can ruin the summing resistor matching effort immediatly. Equations who give
the nice curves are based on ideal filter stages...

As both kind of errors are cumulated , it s very very hard ( read almost
impossible)  to get the ideal filter curves with this summing/difference method
.

Does it mean that the circuit will not produce a nice or interesting
sound ? No :)  But just that you will not notice a big difference between
+/- similar filter modes. Do not expect extreme filter shapes too : i have never
reached the ultra HP and killer Notch filters with this circuit, even if they
are mathematicaly do able!

On my prototype, who was  mount on a MOTM 440 vcf ( sum res match <1%) I
discovered how nice this filter sound is, when used with only  1 or 2 poles :)
I have not tried yet with other VCF , but i guess that the result will be +/- in
the same range.

On 16 available functions , there are at least 5/6 who are really great. I
seriously plant to add them permanently to my 440 one day or another with a
simplified circuit...Others ones are +/- subbtle variations :)

Improvement of this circuit could be :

1) match all summing  resistors , add/or trimmers here or there .

2) Add trimmable gain buffers for each input ,cells inputs gain error can be
compensated

3) match each filter poles for the most  similar level/response. A CEM/SSM can
help, may be ...

4) the resonnance path is also very important. There is lots of things to
experiment here

5)do not expect to reach the pezrfect math curves or either to get a 'miraculous
VCF' who can do everything very well ... This filter do not exist yet , at least
no yet in the analog world and if it was the case you will probably be aware of
it existence, imho...

Cheers,

Marc B.

Selon Scott Nordlund <gsn10 at hotmail.com>:

>
> I believe the precision is necessary because certain frequency bands must be
> canceled out to get the different responses.  To achieve maximum attenuation,
> then, the levels have to be exact, though it doesn't necessarily mean that
> incomplete cancellation renders it useless.  For something like a notch
> filter, I doubt it would be very audible.
>
> ----------------------------------------
> > From: tom at electricdruid.net
> > Subject: Re: [sdiy] Oberheim Xpander VCF
> > Date: Fri, 16 Jan 2009 11:06:46 +0000
> > To: dlmanley at sonic.net
> > CC: synth-diy at dropmix.xs4all.nl
> >
> > Dave,
> >
> > This is extremely interesting stuff.
> > It also strikes me as extremely counter-intuitive, which makes me
> > very curious. How can a 1% variation in anything make a 40dB
> > difference to something else? I know it could, but it just doesn't
> > seem right! That makes me want to know more.
> >
> > It does look like the resultant filter *is* musically useful, since
> > no-one seems to dismiss the Xpander/Matrix synths for the rubbish
> > filter, or the horrible variation between voices. According to the
> > plots, this should be *severe*. So what's going on? Did Oberheim use
> > 0.1% resistors? Does it actually not matter much? Does the ear not
> > hear large variations in filter characteristics? One or two people
> > have commented already that many of the settings on the A107 sound
> > the same, although on paper they vary quite a bit.
> >
> > Regards,
> > Tom
> >
> >
> >
> > On 16 Jan 2009, at 06:35, Dave Manley wrote:
> >
> >> Digging thru the files I found some LTspice simulations of the
> >> Expander filter circuit I ran back in Feb 2007. I was all set to
> >> build a circuit that could sweep (not step as in the Expander/
> >> Extrapole) between different modes, and then ran some simulations
> >> looking for sensitivity to component variation.
> >>
> >> The link below is an AC transfer plot for the "4 pole BP" mode. In
> >> this mode three sections of the filter are summed with ratio
> >> 100:50:100. There are 27 curves in the plot representing sweeping
> >> the 3 summing resistors plus/minus one percent. Solid lines are
> >> voltage, dotted are phase. Note how the low pass half of the bandpass
> >> varies more than 40dB, with a resistor tolerance of 1%.
> >>
> >> http://www.sonic.net/~dlmanley/ExtraPole4PBP_1_PerCent.png
> >>
> >> Here's the "3 pole HP" mode, with 1% and 5% tolerance on 3 of the 4
> >> summing resistors (this mode needs 4 resistors, but LTspice only
> >> allows 3 varying components per sim).
> >>
> >> http://www.sonic.net/~dlmanley/ExtraPole3PHP_1_PerCent.png
> >> http://www.sonic.net/~dlmanley/ExtraPole3PHP_5_PerCent.png
> >>
> >> Clearly the circuit is extremely sensitive to gain error. Switches
> >> and precision resistors work, but will a VCA based circuit? How
> >> tightly controllable are any of the proposed gain circuits? DAC
> >> controlling VCA? PWM controlling VCA? Programmable
> >> Potentiometer? Vactrol?!!! I doubt 1% gain control is easily
> >> achieved. 5% gain variation yields anything from a high pass to an
> >> all pass!
> >> Again, I'm not saying the resultant filter wouldn't be musically
> >> useful, but I don't think the plots previously presented would be
> >> seen in a real world circuit. The LP will look like a low pass,
> >> but will you get a quality bandpass or high pass?
> >>
> >> I stopped considering this circuit because of these results. If
> >> someone has built it I'd like to see/hear the real world transfer
> >> functions of the different modes.
> >>
> >> -Dave
> >>
> >>
> >> Adam Schabtach wrote:
> >>> I have one of Marc's PC boards for that add-on. It's one of my
> >>> back-burner
> >>> projects; I haven't finished it because I've been waffling over
> >>> how I want
> >>> to do the panel and the mode display. I really should at least get
> >>> it to the
> >>> point at which I can test it. I used to have an Xpander and the
> >>> filter was
> >>> probably my favorite part of it. I have two MOTM-440 filters and
> >>> like them
> >>> very much, so giving one of them a number of different topologies
> >>> is very
> >>> attractive to me.
> >>>
> >>> I attempted to reduce the gain errors by hand-matching 1%
> >>> resistors. I also
> >>> realized that many of the resistance values form simple integer
> >>> ratios, so I
> >>> used resistors in parallel, chosen to create the required ratios as
> >>> accurately as possible. I don't know how much it will actually
> >>> help but it
> >>> seemed like a good idea at the time.
> >>>
> >>> --Adam
> >>>
> >>>
> >>>> -----Original Message-----
> >>>> From: synth-diy-bounces at dropmix.xs4all.nl [mailto:synth-diy-
> >>>> bounces at dropmix.xs4all.nl] On Behalf Of David Moylan
> >>>> Sent: Thursday, January 15, 2009 8:26 AM
> >>>> To: George Hearn
> >>>> Cc: synth-diy at dropmix.xs4all.nl
> >>>> Subject: Re: [sdiy] Oberheim Xpander VCF
> >>>>
> >>>> This type of versatility is available to cascaded integrator
> >>>> filters in general, by mixing outputs of the separate stages.
> >>>> Marc Bareille created an add on for the MOTM-440
> >>>>
> >>>> http://m.bareille.free.fr/modular1/vcf_multimode/extrapole.htm
> >>>>
> >>>> Although it's nice to be able to switch in discrete steps you can
> >>>> probably have some fun with a 4 way mixer as well.
> >>>>
> >>>> Dave
> >>>>
> >>>>
> >>>
> >>> _______________________________________________
> >>> Synth-diy mailing list
> >>> Synth-diy at dropmix.xs4all.nl
> >>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> >>>
> >>>
> >>>
> >>>
> >>
> >> _______________________________________________
> >> Synth-diy mailing list
> >> Synth-diy at dropmix.xs4all.nl
> >> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> >>
> >
> >
> > _______________________________________________
> > Synth-diy mailing list
> > Synth-diy at dropmix.xs4all.nl
> > http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>
> _________________________________________________________________
> Windows Live™: Keep your life in sync.
> http://windowslive.com/explore?ocid=TXT_TAGLM_WL_t1_allup_explore_012009
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>