[sdiy] Roland Filters / IR3R05

Oakley Sound oakleylist at btinternet.com
Fri May 1 09:51:22 CEST 2009

The IR3R05 is described in some detail in JX-10 and MKS-70 service 
notes. However, it is described incorrectly in that it says it consists 
of two bandpass and two lowpass stages cascaded alternatively. This is a 
simple error on their part because what is actually happening is that it 
is two state variable filters cascaded, ie. two distinct -12db/oct low 
pass filters cascaded.

The BP function is simply the output available from the first active 
element in the SVF topology. This can be verified by looking at the 
voltage across the capacitors to see that you do have both BP and LP 
responses available which is what you would expect from a SVF.

In this the 3R05 is somewhat similar, in core topology at least, to the 
SSM2045. However, I am unsure of the actual feedback circuitry in the 
Roland part - whether it follows the 2045 or does something different I 
don't know.

The history of the 3R05 must, however, stem from the cascaded SVF 
filters in the JP-6 and the early MKS-80. These both use a IR3109 wired 
in a different way from the usual four stage OTA cascade we have seen in 
the Juno-6/60 and JP-8. I guess from the JP-6 circuitry we can deduce 
the feedback circuitry used in the 3R05. But it should be noted that the 
JP-6 and JX-8P's filters do not sound identical.

Note too the different capacitor values in the fourth stage. Probably 
due to a change in the transconductance of the final stage so as to give 
it a lower effective output impedance.

Personally, I find the 3R05 implementation less pleasant than the 
standard cascade and I have been wanting to play about with the cap 
values to see if I could improve it. Its interesting that the SSM2045 
does vary the cap values from what simple theory would dictate to give 
them an 'open rich sound' as their datasheet says.




Servicing, manufacture & modification of electronic musical instruments

More information about the Synth-diy mailing list