The open source Jupiter-8-4-ALL project

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: lisa, luketeaford, Kent, Joe.

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

The open source Jupiter-8-4-ALL project

Post by dikkietrom » Sun Nov 11, 2018 9:32 pm

Through the courtesy of

- Mr Ikutaro Kakehashi, Roland
- Grenoble Institute of Technology, http://kicad-pcb.org/
- Alfons Wirtz, freerouter.net
- Jack Damery, An in depth look at the Roland Jupiter 8's VCO
- Open music labs, open source ba662
- Tiergrinder, DIY Jupiter 8
- https://jup8restoration.wordpress.com/about/

I started this project https://gitlab.com/anatal/JP8-4-ALL

The purpose is also to offer an open poly synth platform with a solid reference implementation. Current state is that the voice board schematic is drawn in kicad and the board is routed with freerouter. I am doing this for the sake of the best possible outcome and to give back.

Impression

Image
Last edited by dikkietrom on Mon Nov 19, 2018 6:15 pm, edited 1 time in total.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

Jarno
Super Deluxe Wiggler
Posts: 2759
Joined: Mon Nov 15, 2010 6:34 am
Location: Rosmalen, NL

Post by Jarno » Mon Nov 12, 2018 3:41 am

Nice work!

User avatar
StillNotWorking
Wiggling with Experience
Posts: 470
Joined: Thu Aug 22, 2013 3:29 pm
Contact:

Re: The open source Jupiter-8-4-ALL project

Post by StillNotWorking » Mon Nov 12, 2018 6:44 am

dikkietrom wrote:The purpose is also to offer an open poly synth platform with a solid reference implementation.
This sentence makes me warm and fuzzy inside. Also love the way you credits those before you working this topic. :tu:

Best of luck going forward. Only wish I had time get involved with this now.

btw, do this project have an active discussion somewhere?

User avatar
devinw1
Super Deluxe Wiggler
Posts: 1162
Joined: Tue Aug 07, 2018 11:20 am
Location: Portland, OR
Contact:

Post by devinw1 » Mon Nov 12, 2018 10:03 am

Very cool! Will be following!

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Mon Nov 12, 2018 3:21 pm

Thanks for reactions, that keeps me motivated. I will discuss here everything, it is the right place I think.

I finalized the proto system architecture roughly. It sticks to the original architecture but the digital aspect is implemented with Arduino.
Basically done drawing for the first protobuild. There will be midi to gate signals and pot panel control for basic control while testing. The pots I will put on breadboard or carton or whatever. The voice will go straight on pcb. The original has 2 voices on 1 board but I will start with one voice. One voice (2 x vco,vcf,2 x adsr,vca) has 484 components and I started of wrong by not specifying straight away the resistor types so I have to carefully check them all. Fortunately kicad makes that somewhat easy. I had mixed feelings while replicating the scheme, it was stressfull but also mesmerizing. The service manual itself is a masterpiece, I am very impressed by it.

I am not looking forward to soldering all the discrete parts. The board is now ca 400x200 with 1661 holes. Minimum order at elecrow is 5 pieces for $12 each so I don't care about optimizing space now. Assembly is another story, the bom shows 100 different parts but it will probably be more if I specified them all correctly, lets say 150. That can be assembled for $687 at 1 piece. That's not really an option, the sweet spot is at 100 pieces, then it is $43 for a fully assembled THT board. Smd $18 dollar.

Anyways first things first. I will start with an original ir3109 just to check quickly if the board works. I cannot find an IR3R01, if someone spots one please let me know. It will be great if the board proves to be correct asap, then the hunt for the right sound can begin, can't wait.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

User avatar
whoop_john
Veteran Wiggler
Posts: 662
Joined: Sun Mar 17, 2013 12:03 am
Location: London UK

Post by whoop_john » Mon Nov 12, 2018 4:03 pm

dikkietrom wrote:I will start with an original ir3109
Don't know if this is legit or not on Aliexpress:
IR3R01 click on last item.
'tis the pigeons that alight
on Nelson's hat that makes it white

Sebo
Common Wiggler
Posts: 61
Joined: Thu Oct 13, 2011 6:17 pm
Location: Argentina

Post by Sebo » Mon Nov 12, 2018 4:38 pm

Wow!!! Nice project!
Thanks for sharing :)
Sebo
----------------------------------
www.cosaquitosenglobo.com.ar

User avatar
home_listening
Veteran Wiggler
Posts: 661
Joined: Fri Dec 30, 2011 7:05 am
Location: locating
Contact:

Post by home_listening » Tue Nov 13, 2018 1:58 am

Hey, that is a very odd format... System design is the big challenge with a project like this.

I couldn't recommend using an autorouter for anything.

I poked around at cloning the Jupiter 8 a few years ago, this is where I left it. Far, far, far from finished.

I think a poly format in the eurorack rack format (Not eurorack synth format, the real one) with all trimpots on the rear and a direct out jack for each voice perhaps would be pretty neat.

Image

flts
Super Deluxe Wiggler
Posts: 2984
Joined: Tue Mar 04, 2008 7:01 am

Post by flts » Tue Nov 13, 2018 3:27 am

home_listening wrote:I think a poly format in the eurorack rack format (Not eurorack synth format, the real one) with all trimpots on the rear and a direct out jack for each voice perhaps would be pretty neat.
I think Jürgen Haible (rest in peace) did some of his stuff like that - at least one of the Moog PS clones seemed to have all of the dozens of voice cards installed in an Eurocard frame. Not sure if he used a backplane or just the frame though.

FWIW especially with a complex project like this, I agree that autorouting is a bit scary regarding how well the circuit will perform / function. Then again, I don't want to discourage anyone with a cool project like this, and if worst comes to worst and there are issues with the layout, it's always possible to reroute everything as the project is open.

User avatar
home_listening
Veteran Wiggler
Posts: 661
Joined: Fri Dec 30, 2011 7:05 am
Location: locating
Contact:

Post by home_listening » Tue Nov 13, 2018 4:48 am

Yeah, I think it was his PS3300 Clone.

I'd hate to be discouraging too, just throwing my 2c in!
flts wrote:
home_listening wrote:I think a poly format in the eurorack rack format (Not eurorack synth format, the real one) with all trimpots on the rear and a direct out jack for each voice perhaps would be pretty neat.
I think Jürgen Haible (rest in peace) did some of his stuff like that - at least one of the Moog PS clones seemed to have all of the dozens of voice cards installed in an Eurocard frame. Not sure if he used a backplane or just the frame though.

FWIW especially with a complex project like this, I agree that autorouting is a bit scary regarding how well the circuit will perform / function. Then again, I don't want to discourage anyone with a cool project like this, and if worst comes to worst and there are issues with the layout, it's always possible to reroute everything as the project is open.

User avatar
home_listening
Veteran Wiggler
Posts: 661
Joined: Fri Dec 30, 2011 7:05 am
Location: locating
Contact:

Post by home_listening » Tue Nov 13, 2018 4:52 am

Re: Autoroute

With both the Pro One clone and SEM clone I did, after dumb mistakes from transcribing the schematics were sorted out, most of the hard work was finessing the layout to eliminate crosstalk. This was particularly hard on the Pro One where the mod bus was running signals 40cm from the ADSR over to the other side where the mod matrix is then back over to the filter modulation.

flts
Super Deluxe Wiggler
Posts: 2984
Joined: Tue Mar 04, 2008 7:01 am

Post by flts » Tue Nov 13, 2018 5:33 am

home_listening wrote:With both the Pro One clone and SEM clone I did, after dumb mistakes from transcribing the schematics were sorted out, most of the hard work was finessing the layout to eliminate crosstalk. This was particularly hard on the Pro One where the mod bus was running signals 40cm from the ADSR over to the other side where the mod matrix is then back over to the filter modulation.
Yeah, I suppose in this case eg. getting the bypass caps close to where they need to be, routing the multitude of signals in a way that one avoids serious crosstalk, keeping the layout of VCO (core) tight and sane enough to keep it
as stable as possible, etc. contribute quite greatly to the end result.

But in any case, I suppose this just serves as a sort of a warning that if things won't work as intended in first prototypes (they never do), there's much to be gained by careful manual rerouting.

Very interested in seeing what becomes of this :)

User avatar
StillNotWorking
Wiggling with Experience
Posts: 470
Joined: Thu Aug 22, 2013 3:29 pm
Contact:

Post by StillNotWorking » Tue Nov 13, 2018 9:38 am

Not designing for production simple wire runs are a valid option I think. Modularize also keep troubleshooting simpler and makes for modular build process where new ideas easily can be inserted.

Last 5-7 years many sdiy projects has become obsessed with everything has to be soldered to a pcb. :despair:

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Wed Nov 14, 2018 3:48 am

whoop_john wrote:
dikkietrom wrote:I will start with an original ir3109
Don't know if this is legit or not on Aliexpress:
IR3R01 click on last item.
Someone tried that on the anafrog forum and it does not seem to be ok. The pics that tiergrinder uses arived yesterday, will use that for now with an adapter.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Wed Nov 14, 2018 4:06 am

Re autoroute

The board layout is also in the service notes, if some cares to map that I will certainly go with it. For proving the sceme is correctly implemented it is not really relevant is my exeperience, therefore I prefer progress for now.
I also use auto placement but it spreads per subsheet it seems, was surprised about that. In the rules you can specify track width and spacing, that guards somewhat against crosstalk. I will space out long running parallels, thanks for the input.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Wed Nov 14, 2018 4:28 am

StillNotWorking wrote:Not designing for production simple wire runs are a valid option I think. Modularize also keep troubleshooting simpler and makes for modular build process where new ideas easily can be inserted.

Last 5-7 years many sdiy projects has become obsessed with everything has to be soldered to a pcb. :despair:
Point 1 is spot on, subsheets can be used as components which makes it very easy to maintain. It is the only way I am able do take this on with joy.

Point 2 not so much (for me) You are in despair cause others like to work with fixed boards? I totally don't get your point. DIY is relative, I am sure you are not into DIY metal mining. There is probably a correlation between the costs of board manufacturing and your observations IMHO
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

User avatar
StillNotWorking
Wiggling with Experience
Posts: 470
Joined: Thu Aug 22, 2013 3:29 pm
Contact:

Post by StillNotWorking » Wed Nov 14, 2018 12:20 pm

dikkietrom wrote: Point 2 not so much (for me) You are in despair cause others like to work with fixed boards? I totally don't get your point. DIY is relative, I am sure you are not into DIY metal mining. There is probably a correlation between the costs of board manufacturing and your observations IMHO
Not exactly, more to the fact one lock users into panel design and components to use. Defeating some of the DIY aspect of it for others to explore. Also it often include an assembly scheme that hinder troubleshooting and mods to be done to the design.

User avatar
home_listening
Veteran Wiggler
Posts: 661
Joined: Fri Dec 30, 2011 7:05 am
Location: locating
Contact:

Post by home_listening » Wed Nov 14, 2018 5:35 pm

I personally consider every wire I strip and solder to be a failure on my part, but then again I only build my own projects with my own PCBs so I'm not locked into anyone else's design - I make the design.


StillNotWorking wrote:
dikkietrom wrote: Point 2 not so much (for me) You are in despair cause others like to work with fixed boards? I totally don't get your point. DIY is relative, I am sure you are not into DIY metal mining. There is probably a correlation between the costs of board manufacturing and your observations IMHO
Not exactly, more to the fact one lock users into panel design and components to use. Defeating some of the DIY aspect of it for others to explore. Also it often include an assembly scheme that hinder troubleshooting and mods to be done to the design.

flts
Super Deluxe Wiggler
Posts: 2984
Joined: Tue Mar 04, 2008 7:01 am

Post by flts » Thu Nov 15, 2018 9:11 am

It's good to bear in mind that even with a "panel" PCB one can practically always manually wire the panel components if needed. It's especially easy if it's a separate control PCB that sandwiches to main board with headers that have a clear logical pin mapping so that one can use eg. .100" connectors for wiring.

But not vice versa.

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Thu Nov 15, 2018 7:33 pm

StillNotWorking wrote: Not exactly, more to the fact one lock users into panel design and components to use. Defeating some of the DIY aspect of it for others to explore. Also it often include an assembly scheme that hinder troubleshooting and mods to be done to the design.
I don't think I used providers that have the restrictions you mention. In the context of my personal capabilities it is more efficient to work with a fixed proto board. I start with complete THT for the sake of flexibility. With small concepts I use wire-wrapping but that is not really an option now.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Thu Nov 15, 2018 8:15 pm

Checked all component values, found a handfull wrong ones and a missing wire. Will check it one more time and if nothing found then consider it ok. Replaced direct BA662, IR3R01 and IR3109 implementaions with original footprints. Implementation is either original or something adapted.

I have been thinking about the autorouting and placement comments, especially regarding the decoupling caps and decided to replicate the placement from the service manual for the proto board. That has to happen anyway so as might well do it now. It will add another few days.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

User avatar
dhaillant
Common Wiggler
Posts: 139
Joined: Fri Aug 30, 2013 5:25 am
Location: France
Contact:

Post by dhaillant » Fri Nov 16, 2018 12:27 pm

Interesting project!
Autorouting is, IMO, unnecessary, adds complexity, errors, and a lot of work in the end.

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Fri Nov 16, 2018 8:40 pm

It depends which autorouter you use and how you use it. My experience is that it mainly helps a lot with optimizing placement. It's not a black and white thing. Also you can use it to route just parts of the board that are no brainers. With freerouter you can manually push and shove a lot better than with kicad. It's a fine tool. I totally agree that one should not blindly autoroute a final board.

I am halfway annotating the scheme analog to the service notes scheme (no pun). The automated annotations were not so convinient for the board replication. A scheme trip takes about a full day, just one voice that is. Glad I did not already draw the controller like the voice, without the component numbers and types.

A controller per voice would be nice, then any split is possible. 1/7,3/5,4/4 or 4x2 and so on. Depends on how much sample and hold leak one dac can handle or if an affordable dac will do.

I have a 32 channel dac chip but too expensive to use 1 channel per cv. A voice needs 32 cv signals and 1 gate. If one channel can serve 32 sample and holds then it is another story, that would mean a potential jp32.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

User avatar
StillNotWorking
Wiggling with Experience
Posts: 470
Joined: Thu Aug 22, 2013 3:29 pm
Contact:

Post by StillNotWorking » Sat Nov 17, 2018 6:26 am

dikkietrom wrote:Depends on how much sample and hold leak one dac can handle or if an affordable dac will do.
Could S&H of the original hold some artefacts that would add to the analog fluctuating? At least that would simplify the search for a DAC :hihi:

dikkietrom
Common Wiggler
Posts: 160
Joined: Sun Aug 11, 2013 10:59 am

Post by dikkietrom » Sun Nov 18, 2018 1:15 am

I am somewhat surprised and confused about the dac's, I always thought memory was 12bit or 14bit but it is 8bit on the controller for pot data and 12/14bit on the interface for key cv. The 8 bit is a discrete ladder.

Btw it is not 32 sh but "only" 23 per dac throug 3x8-channel multiplexers, I have no clue about the s&h refresh rate, the z80 has 8mhz clock.

Anyways the voice scheme is annotated with original references now, the original scheme has some missing references but that can be recalled from the board layout. Found one more wrong junction in the kicad scheme. In hindsight it was not so bad that I had to go over it a few times cause I found mistakes that otherwise might have been unnoticed.
I am going to take a little break. Looking forward to replicate the component placement.
The open source Jupiter-8-4-ALL project
Desperately seeking IR3R01

Post Reply

Return to “Music Tech DIY”