[sdiy] Filter strangeness.

[Harry Bissell Jr]

Woah you got some real weird stuff there :^P

inline
> 
> Back again to pick some brains  : )
> 
> Well, I've been losing lots of sleep (good thing)
> and making good progress
> (very good thing) on that optical synth thingy, ever
> since I got the power
> supply working.  I've even added on to the power
> supply, so I've got +9, +5,
> and -5V now. I made a really sweet signal generator
> as OSC 1, which makes
> square and triangle, both of which are pretty darn
> pure from 20Hz to
> probably 25Khz.
> 
> Tonight I've built a supposed low pass filter, based
> on a 4th order
> sallen-key filter.  Some of the inspiration can be
> found here;
> http://www.crossovers.com/figure_2.html
> I was focusing on the second schematic on the page.
> I am using R2 as my
> cutoff control (maybe my problem?)
> 
> It seems to do it's filter thing, and under certain
> circumstances it sounds
> very cool and acid.  BUT something is not right. 
> For one thing the output
> of the filter is half or even less of the amplitude
> of the oscillator

If the filter is tuned below the freq of the
oscillator
this is correct operation... If it is tuned REALLY
low the osc will cut off completely.  Probably you
want
to make up the gain after the filter, not before
(usually an oscillator has a pretty high voltage
signal
anyway)

 Also the filter seems to change how the
> oscillator acts more than I
> expected, like I can change the frequency on the OSC
> with the variable
> resistor on the filter, especially when the OSC is
> running at very low
> frequencies.  Sometimes it even seems to makes the
> pitch variation function
> in "steps" instead of a smooth transition...
> weird...

One:  You could be sucking down the power supply and
that would make them interact. Also, the filter could
have some HF oscillation that you CAN'T hear, but that
is getting into the oscillator, probably changing the
point the compatator trips at.

Look for voltage change with a meter... the
oscillation
can only be found with a scope (usually)

> 
> It seems to be something like a ground problem,
> because I can raise the
> amplitude by touching the variable resistor.  I do
> love the sound of the
> filter though, I just hope I can work out the bugs. 
> It self oscillates
> really nice, and I can stand or even appreciate the
> quirky interaction
> between the two circuits, but the low amplitude is a
> bit bother some.  I
> will be having an amp in series with it, but it
> seems like I could improve
> the S/N ratio first and get a better result,
> especially since the OSC is
> very quiet.
> 
> Do you think I need a buffer between the two?

Yes, you might... depends on the oscillator design
and the input impedance of the filter (essentially the
values of R1-4 in the first figure)   The two opamp
version will probably work MUCH better for your use...


  They
> are running at different
> voltages (OSC +-5V, filter +9V).  Or would you look
> more at my component
> choices, not shown of course so I just went with my
> instict.  I know that th
> e input of the filter should have grount sent
> parallel with the signal, but
> since they are on seperate voltages these seems to
> not work quite right.

You should have a series cap between the filter and
oscillator, for sure.  Or run the filter on the +/-5V
(much better)

> I'm supposing that it could also be the single
> supply filter dealing with
> the dual polarity OSC.  Or so you think it could be
> the wrong value input
> resistor.
> 
> If you wanted a photo resistor to vary cutoff, how
> would you setup the input
> stage?

>From the second figure... vary R1, R2, R5, R6 and keep
them all the same value. Play with R3, R4, R7, R8 for
resonance.  Keep R3 = R7 and R4 = R8
> 
> Also what would you do to add variable resonance to
> the circuit?  I was
> experimenting and it seemed like a small value pot
> from the ouput of opamp 1
> to the inverting input of opamp 2 did something like
> this, but it wasn't
> quite what I am looking for I think.
> 
> All in all a good project and a learning experience,
> as I obviously need to
> learn before I go at my modular.  : )

You might want to look at some commercial schematics
to get practical designs... the Korg MS-20 and Delta
have similar S-K designs like what you did. Also check
out the Steiner-Parker filter

H^) harry

> 
> Thanks in advance for the help!
> Pat
> 
>