lm13600/700/ne5517-SVF-Problem
Magnus Danielson
cfmd at swipnet.se
Mon Aug 7 22:27:47 CEST 2000
From: modlar at gmx.net
Subject: lm13600/700/ne5517-SVF-Problem
Date: Mon, 7 Aug 2000 14:40:34 +0200 (MEST)
> Hi all !
>
> I have a general understanding problem with a special state var filter.
> It is based on a LM13600 and looks very like the tomg vcf9 (Serge?).
> The problem is, that the offset voltage of the lowpass output
> (which should be zero) shows changes of 8 to 10 Volts if there is a
> sudden change of the control voltage. After this sudden offset change
> the output offset slowly decreases to the normal zero value (2-3 secs).
> Principally, the filter works (HP,LP,BP, with and without resonance)
> but for musical purpose, these offset changes are inacceptable. Imagine
> a AR-generator (feeding voltage control of the filter ) with Attack time
> set to 0 sec ! The output will change to 8 Volt, added to the filtered
> sound -> overflow !
>
> I tried to replace the LM13600 buffers with OPAmp-buffers (071), I
> changed the LM13600 to the 13700 and the NE5517, no changes.
>
> To my knowledge, this isn't a principle behaviour of a state var filter,
> (my formant filter, who works with the 3080, does not show this
> at all).
Strangely enougth I recognice this behaviour from my Formant's 24 dB filter!
> Is it the LM ?
> Any help ? (Perhaps this is a simple beginner question)
Well, just to get the discussion up and kicking, I am just pondering about the
DC offset errors. In the ASM-1 VCF (a state-variable 2-pole filter) there are
trimmers at both the CA3080s for doing DC offset adjustment. One must recall
that directly following the OTAs in a state-variable design is the integrators
and I would not be supprised that the quick changes is so quick that the
feedback loop is unable to compensate for the offset error while the change
occurs, then we see a buildup and then it takes some time before the feedback
loop to compensate this out. Interesting experiments would be to see how the
speed sensitivity and dropout speed changes with diffrent frequency range and
Q settings.
This is just me doing a quick pondering about the problem, but hopefully it
will be enougth for some others to also ponder over it and we can have a
discussion and possibly learn about it.
Cheers,
Magnus
More information about the Synth-diy
mailing list