Roland Juno-106 Filter

nebula · October 2, 2007, 7:52pm

You know, that Juno filter sounds really sweet, thick, lush. It will self-oscillate but in a remarkably musical way. It kinda got me thinking …

You may already know that each voice on the Juno-106 uses a custom Roland IC called the 80017A, which is essentially a combined VCF and VCA. The reason many know that is because the IC is very prone to failure, and when it goes it causes the dreaded “I lost one of every six notes” syndrome.

Roland no longer makes them, but in an effort to “save the Junos”, some people have tried to clone the long since discontinued Roland IC. If I’m to believe everything I read, one guy has actually managed to get close enough that you can’t hear a difference, using readily available parts. With the Juno-106 service manual it should be a cakewalk to figure out the pinning on an 80017A.

So … you could probably stick 8 of these filters on a board with an AOUT … filter each SID from your MB-6582, or maybe use it as the basis of a Juno-106 clone.

opkod · October 2, 2007, 7:57pm

look at this vcf/vca clone : http://www.analoguerenaissance.com/D80017/3

ilmenator · October 2, 2007, 8:32pm

At 40,- Euros each (that means per voice :o) this is ridiculously expensive. It’s not a DIY project, and although I respect this guy’s work I would not try to incorporate such a thing into an otherwise completely DIY project.

Best regards, ilmenator

nebula · October 2, 2007, 9:32pm

look at this vcf/vca clone : http://www.analoguerenaissance.com/D80017/3

That’s the one I was talking about. I inadvertently linked to his older site.

nebula · October 2, 2007, 9:41pm

At 40,- Euros each (that means per voice :o) this is ridiculously expensive. It’s not a DIY project, and although I respect this guy’s work I would not try to incorporate such a thing into an otherwise completely DIY project.

Best regards, ilmenator

Going rate for CEM3378 chips is about $10 each. People in this community use them rather than make filters using discrete components, which could theoretically be a bit cheaper. Oh and don’t forget, lots of us have ponied up $20 each for SIDs, and plan to put 8 in a box!

While it’s completely subjective, I will step up to bat and say that the filters on Roland’s classic Junos are among the sweetest sounding in synth history.

Jaicen · October 2, 2007, 10:35pm

I don’t know if it’s important, but the Juno filter chips were used in a few of Roland’s products. Essentially, it’s based on the Moog filter design, using diodes instead of a transistor ladder. Personally, I find Roland filters to be more musically useful than Moogs anyway. You can make a Jupiter or SH09 (discrete filter I know) sit much easier in a mix than something like a Minimoog.

stryd_one · October 2, 2007, 10:57pm

Going rate for CEM3378 chips is about $10 each.

Huh where?

Wilba · October 4, 2007, 4:36am

from: http://www.analoguerenaissance.com/D80017/3

Stupid Question Fee

From now on I charge 10EUR for every question that people ask me by mail that is already answered on this page. I am tired of answering the same questions over and over and over again. This keeps me from doing electronics.

Maybe it’s wrong of me to say this, but what an asshole. If you’re going to charge 40EUR and make money off something like this, stop complaining and answer your potential customer questions, regardless of whether they were answered already or not. If someone asks me a question about MB-6582 that is already answered in the wiki etc. I direct them to it. If I got 10EUR every time, I’d have enough to fly to Munich and jam with TK for a week. *whack*

Sasha · October 4, 2007, 6:38am

…If I got 10EUR every time, I’d have enough to fly to Munich and jam with TK for a week.

;D ;D ;D

I understand the guy`s frustration, but when it is commercial product it is OK to listen to most stupidest questions you can imagine. It is obvious that this guy rather do only electronics, but than he needs to let others to take care of commercial things and customer care. I hate to deal with customers myself… but as it is my job I need to be nice even I would rather tell them to go to hell. Customer is always right. I hate it, but is IS truth if you want to earn money.

DrBunsen · October 4, 2007, 9:42pm

You know, he’s got some pretty good close-ups of his boards there…

Jaicen · October 4, 2007, 9:48pm

True, but not quite good enough. There are no markings on the IC’s, and the board is double sided…

DrBunsen · October 4, 2007, 9:53pm

Surely between all of us we could reverse engineer something? I could donate a working one to the cause. It’s not like my MKS7 is doing anything right now

Ashiman · October 4, 2007, 10:28pm

LOL!

besides ppl assembling groupbuy for CEM3372 filterchip on synthdiy, the one used in Oberheim Xpander , giving state variable LPHPBP at the same time

stryd_one · October 5, 2007, 11:00am

I got a datasheet here with my stack of 3372s and it says lowpass only I know you can do tricks to turn lowpass into high, but that means no resonance. I tried to find any mention of this project, to see what the deal is, but cant

nebula · October 5, 2007, 1:06pm

the Juno filter chips were used in a few of Roland’s products.

Best I can tell: the Juno-106, the MKS-30 “Planet S”, and the MKS-7 “SuperQuartet”. Are there more?

You learn something every day: I always thought the MKS-30 was a rackmount JX-3P, since it has an identical parameter set and uses the PG-200 programmer. But the JX-3P uses discrete filters (based around IR3109) while the MKS-30 uses 80017 chips.

You can make a Jupiter or SH09 (discrete filter I know) sit much easier in a mix than something like a Minimoog.

Jupiter-6 and Super Jupiter use CEM filters.

Ashiman · October 5, 2007, 1:53pm

probably the best explanation can be obtained by searching for SSM2040 http://www.siliconbreakdown.com/msl/chip.html

oh and if you check the MOTM partlist for the discrete 2040, they use the CA3086… e(r v r s n g n r pl s) ;D

stryd_one · October 5, 2007, 2:57pm

Ahh the poles are exposed at the cap pins, cool. Thanks PB!

Jaicen · October 5, 2007, 6:04pm

The early MKS-30’s did indeed use the IR3109, also used in the venerable GR-300), and Juno 6 and 60. However when Roland re-used the architecture and indeed boards for the GR-700 and GR-77b they switched over to the same chips as used in the Juno 106, HS106 and related racks. I think it can be found in some other less well known products like the VK’s, but I can’t be sure. Those are the only ones that I know feature this chip for sure.

EDIT: Check out this page, you can see that the IR3109 is in a heap of old roland stuff:

http://analog.no/rolandcustom.htm

DrBunsen · October 5, 2007, 9:29pm

I’m searching: I’m looking to restore a Juno 106 (MKS7 - the rackmount version), rather than a general purpose filter for MB projects. Though if that becomes an option, sweet.

We’ve been discussing in Chat the possibility of reverse engineering an 80017a clone and open sourcing the design. That would give hackers the option of restoring their dead Junos at a lower price (and more labour) than the available substitutes.

I just found this most interesting image in an article on Matrixsynth.

As you can see, there are separate I/O pins for the VCF and VCA stages, for both signal and control. The VCF is a Roland IC?* called an IR3109 and the VCA is BA662*

So in essence they’re two completely separate blocks, which breaks the cloning problem down into two parts: cloning the IR3109 and the BA662. Or alternatively: making a socketed adapter board that can take SSM ICs for VCF and VCA.

There are trimpots on the mainboard which allow you to fine tune the sound so each voice matches.

I also found a block diagram for the insides of the chip here

_*_according to this they are surface mount versions called A1QH800170 and uPC1252H2 BA662

===============================================================