To keep the focus of this article more direct, only the IR3109 / 80017a / BA662 style filters will be discussed.
The development of IR3109 was in direct relation to simplify the process of building a resonant filter with discreet BA662 ICs (which had to be hand-picked, tested for matching, organized into specific groups, etc). The amount of labour it took to build the first revision of Jupiter-4 voice cards was staggering, and IR3109 helped reduce that as no matching was necessary and the four discreet ICs became one. Later IR3109 would be reduced further in an integrated surface mount package (80017a).
The million dollar question is how different are the sonic changes between: four BA662 ICs, IR3109, and the later surface mount shrunk 80017a.
Roland Jupiter-4/Promars (generally, only later units)
Roland JUNO-106 (80017a)
Roland MKS30 (80017a)
Roland MKS7 (80017a w/ IR3109)
Roland MKS80 (generally, only earlier units)
Roland Jupiter-4/Promars (generally, only earlier units)
Roland SH-09 (BA662 A)
Roland System 100m (BA662 A)
I'm not certain if there is a sonic change using a BA662 A over a BA662 B. I'm also not certain if there's a sonic change when BA662 is used for buffers in conjunction with the FET transistors, or FET transistors exclusively.
This section will be pending. I intend to begin comparing a Jupiter-4 (IR3109) with a SH-09 (BA662 A w/ Hybrid FET/BA662 buffers). And then if I can get in contact with someone who has SH-2 or BA662-based Jupiter-4 to continue doing further comparisons. Once that's accomplished I would like to compare IR3109 from a JUNO-60 with 80017a on a JUNO-106 (preferably original ICs 'just to be safe').
Due to the amount of classic Roland synthesizers that have had a bit of a resurgence, many of the discontinued ICs have been cloned or reverse engineered.
It's important to note that if you are going to replace BA662 *filter designated* ICs on an earlier Roland synthesizer, you actually have to buy a large batch of them and transistor match them just like Roland had to do in the late 70's/early 80's. Failure to do so will prevent oscillation at certain frequencies... and possibly cause problems with portamento.
Analogue Renaissance AR662 - http://www.analoguerenaissance.com/AR662-DOT/
Probably one of the most unique and ambitious of the BA662 clones, Jeroen Allaert intends this chip to be dynamically configured to 'match' any of the hand-picked ICs Roland designated with the coloured painted dots at the top. I'm going to assume these too, won't require a SMD to SIP adapter.
Coolaudio V662A - https://www.coolaudio.com/features-page.php?product=V662A
The Coolaudio BA662 clone was reverse engineered from original BA662 ICs in conjunction with AMSynths. An adapter will be required to convert the SMD package to SIP.
Rpar AS662 - https://www.alfarzpp.lv/eng/sc/AS662D.php
The AS ALFA RPAR BA662 clone appears to be similar to the VA662A, however, I'm not certain where these are actually purchased from. An adapter will be required to convert the SMD package to SIP.
OpenLabs BA662 Clone - https://synthcube.com/cart/open-music-labs-ba662-clone
The OpenLabs BA662 can have an optional trimmer soldered on to adjust it as if it were a 'matched transistor' for usage in Roland filter application. You simply need to calibrate all of the packages to the same value to one another.
Analogue Renaissance AR3109 - http://www.analoguerenaissance.com/AR3109/
Jeroen Allaert created a clone of the IR3109 to be a drop-in replacement for any synthesizers that use it. It appears it can be purchased from Syntaur (I'm not certain if there are any other distributors): https://syntaur.com/Items.php?Item=4243
Rpar AS3109 - https://www.alfarzpp.lv/eng/sc/AS662D.php
The AS ALFA RPAR AS3109 clone could very well be sonically identical to IR3109 (not a whole lot of information on their performance), I'm not certain where these are actually purchased from. The IC is offered in both a SMD and DIP package.