[sdiy] Does anyone have the Roland IR3109 datasheet? (looking into sh-101 filter modification)

[Oakley Sound]

The IR3109 is simply four OTAs and their buffers with a common expo 
driver. In the JP4 and JP8 Roland simply wired them as four cascaded low 
pass elements with feedback wrapped around from the fourth stage to the 
input. The JP-8 has a tap at the second stage for the -12dB output. In 
JP6 they are wired as two SVFs in cascade - and consequently sound 
rather different but give HPF and BPF responses.

As we have heard the JP-8's HPF was from a simple BA662 OTA wired as a 
single pole high pass filter and fitted before the main four pole VCF. 
Since its not actually tracking the VCOs it has little more effect than 
a single high pass element in the final output stage of the whole synth. 
Which is what they did in the Juno-6 et al.

HPF modes can be obtained from cascaded low pass elements in a 'derived' 
fashion. This is where the various taps are mixed together to form the 
common filter responses. But to get HPF you need to have the input 
signal of the filter so that it can be subtracted from the stage 
outputs. There's a nice section in Electronotes about this - can't 
remember the number off hand.

But see the Oberheim Xpander for the way they did with the CEM chip. 
Note that Oberheim used the CEM chip to also mix the two VCO inputs 
using the CEM's two internal VCAs. Since the internal connection from 
the VCAs to the first OTA stage is not made available as an output pin 
they had no direct tap to the input of the four stage filter. In high 
pass mode you need to have this so what they did was to turn the first 
stage into a buffer stage - this way the first stage tap was actually 
the input signal to the following three stages. The way they turned the 
first stage into a buffer is quite clever. Its still a low pass filter, 
but by changing the timing cap to a very low value, they made the 
cut-off so high in that stage alone as to make it behave as simple 
voltage follower. You only get a maximum of a -18dB/oct HP function 
since you only have three active integrators.

Tony

www.oakleysound.com