MUFF WIGGLER Forum Index

 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler Blog & NewsMW News   Muff Wiggler StoreMW Store 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Mutable Conversions to DIY Friendly PCBs
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2, 3, 4, 5  Next [all]
Author Mutable Conversions to DIY Friendly PCBs

abelovesfun

Hello,

With the blessing of Oliver, I am in the beginning stages of creating DIY friendly versions of mutable modules. By DIY friendly, I only mean that these PCBs will use thru-hole parts whenever possible. The goal here is not to get rich (at all - I'm no dummy) but to provide DIYers with more enjoyable build experiences to add some Mutable love to their systems (especially non-euro systems). I will be producing PCBs and MOTM Panels. Per Oliver's wishes, these will be under a different brand name, and the modules will have different names, but credit will be given.

Now for the shaking your head moment: I have only been doing DIY for two years, and though I have built many many modules for myself and occasionally for others, I have only vaguely fussed with Eagle, so this is a learning experience for me.

I am currently working on porting the Braids module. I am under the impression that the easiest method here is to, using the Braids BOM, start a new project schematic using thru-hole components, sort the PCB out (checking against the existing) and going through prototyping. Is this incorrect? Should I just be working from the existing Schematic and replacing parts on a part by part basis?

If there are any parties who would be willing to review files, I would very much appreciate it.

Thank you.


Starspawn

I think you can actually import the files for Braids, then change type of parts to thru hole in eagle and sort lines from the mess that appears then.

Also, we love you for this, I know a few have started, and some of the modules are pretty simple regarding parts.

Best of luck smile


DMR

I'm looking forward to this. But why start with Braids? Is there a through-hole equivalent of the Braids CPU? It might be easier to start with one of the AVR-based modules.


logicgate

I'm very fond of diptrace for PCB designing

I would first create a schematic file and then convert it to PCB and start arranging the components around...


pld

Good plan.

Someone recently posted pictures of a Braids with a through-hole PCB, wired pots and jacks, and dev board heart on the MI forums, maybe they'll share? No use duplicating effort.

And as was said, you should be able to open the existing file, change the packages and re-route. Simple, but not easy smile It seems a reasonable next step in DIY, taking a known-working schematic and making a new layout...


Altitude909

the trickiest part to solder is the processor which only comes in a surface mount version so without a slick way to deal with that, you're not really making this any easier


qp

abelovesfun wrote:
Should I just be working from the existing Schematic and replacing parts on a part by part basis?

Thank you.


I would swap the parts and try to keep everything grouped as in the original PCB design. You don;t know if parts are placed where they are for a reason, such as noise issues. In eagle schematic, just below the add tab on the left is a replace tab. Use this to select the new part then click the part on the schematic you want to replace it with. Of course, doing this will require a lot more board space, so keep that in mind. One of the layers on Oliviers brd files has markings for all of the eurorack size restrictions, so it should be pretty easy to resize the board. The best option would be to make a PCB just for the controls and a stacked board with all of the parts. This way you can keep it the same footprint. Thinking on it, you could put the digital stuff on the front PCB and the analog VCA's and the like on a stacked PCB. That would save you the need for a lot of headers.

I would highly suggest getting proto's made first to before taking any orders. You could either use OSHpark for prototyping, or use the fab for this. Personally I use OSH, then order all of my boards through PCBcart, but they also offer a prototype service which I will probably use next time. Helps clear out any discrepencies with their process of manufacturing and OSHparks.


lvoemachine

I've started a Moog format panel for clouds. If you do a pcb for that is be stoked.


qp

Altitude909 wrote:
the trickiest part to solder is the processor which only comes in a surface mount version so without a slick way to deal with that, you're not really making this any easier


then there's this, which is 100% true. All of the parts are easypeasy after you solder that CPU. I've got the right tools and I still managed to make a couple solder bridges. But, this is what braid and a flux pen is for.


abelovesfun

qp wrote:
abelovesfun wrote:
Should I just be working from the existing Schematic and replacing parts on a part by part basis?

Thank you.


I would swap the parts and try to keep everything grouped as in the original PCB design. You don;t know if parts are placed where they are for a reason, such as noise issues. In eagle schematic, just below the add tab on the left is a replace tab. Use this to select the new part then click the part on the schematic you want to replace it with. Of course, doing this will require a lot more board space, so keep that in mind. One of the layers on Oliviers brd files has markings for all of the eurorack size restrictions, so it should be pretty easy to resize the board. The best option would be to make a PCB just for the controls and a stacked board with all of the parts. This way you can keep it the same footprint. Thinking on it, you could put the digital stuff on the front PCB and the analog VCA's and the like on a stacked PCB. That would save you the need for a lot of headers.

I would highly suggest getting proto's made first to before taking any orders. You could either use OSHpark for prototyping, or use the fab for this. Personally I use OSH, then order all of my boards through PCBcart, but they also offer a prototype service which I will probably use next time. Helps clear out any discrepencies with their process of manufacturing and OSHparks.


Oh absolutely - no orders until I've not only proto'd via OHS (local to me, I think) but also built, tested, and documented the first couple of boards I get back for sale.


synchromesh

Since this topic is clearly in the air at the moment, I might as well mention this here... I was going to wait until I'd actually built up a prototype (and run my project past Olivier) before announcing anything (if only to prove to myself that I could pull it off), but hey.

I have been thinking along similar lines, although my approach is to keep the SMT and just make it easier to DIY. So far I have*:

* Uploaded *the Ripples design to *Upverter Community Edition (i.e. as a public, CC-BY-SA open-source hardware project).
* Split the *filters, *pots and sockets out onto separate PCBs/projects. This will make it easy to build for other formats.
* Laid out socket PCBs for the original *Erthenvar jacks and *Thonkiconn jacks.
* Laid out the pot PCB. The filters, pots & sockets PCBs should fit together behind a "standard" Ripples panel.
* Shrunk the filters PCB height to 100mm so it'll be cheaper to have made.
* Replaced all the imported Eagle library components with either generic passives or parts via Octopart. (FWIW this threw up a number of discrepancies with Olivier's BOM.)
* Selected 1206 footprints for all the passives. (Although I'm wondering whether I should just go for 0805.)
* Replaced the SMT electrolytic capacitors with through-hole parts, since I gather they're a PITA.
* Started laying out the filters PCB. I'm currently trying to decide whether it's a good idea to spread parts across both sides of the board, or switch to 0805 and try to stick to one side.

I did electrical engineering at university but I'm a software developer by trade, so I have some idea what I'm doing but little recent experience with SMT. So the whole project is a learning experience for me. If I can pull this off then I'd tackle the Braids design next using a similar approach, except for:

* I'd use 0805 footprints for the passives.
* I'd use a *STM32 Nucleo F411RE development board for the CPU. This would add more horsepower for alternative firmware, while presumably simplifying the build.

I'd like to do this using tombola's "Music Thing Modular" model of providing the Gerber files (indeed, the entire project) and instructions on how to upload them to a PCB fabrication house, but not selling boards or providing support directly. These days it's so easy and cheap to get a few boards made for something like this that doing the whole board run thing via the forums seems like overkill, unless you're selling panels and/or kits into the bargain (and I do appreciate it when people do that - thanks fonik!). Although buying small quantities of SMT parts is a bit painful so perhaps an SMT parts kit would be useful and vaguely cost-effective. But obviously that's all moot until I've actually finished.

So I've been working on this off and on for a few weeks and I hope to maintain my slow-but-steady pace to completion, but like everyone else I have many other demands on my time. So we'll see how it goes. I would be grateful for any feedback on e.g. soldering 1206 vs. 0805, components on the bottom as well as the top, or other experiences with DIY SMT.

I don't mean to steal anyone's thunder with this little announcement, but I figure it's all part of the community figuring out the best way to move forward, or something. This is fun!

* I like to inline links, but I don't think they show up properly in the default theme. I've switched back to the original "garish shades of gray-blue" theme so that links given above stand out, but YMMV.


jpj

Maybe, you can use a Mapple Mini as BRAIDS core. Check Sneak-Thief post at the BRAIDS Gerber/Export Thread
https://www.muffwiggler.com/forum/viewtopic.php?t=120940&postdays=0&po storder=asc&start=75

The only problem seems to be, that Mapple Mini has no output pin PB1 to 74HC595's Serial In (DISP_SER). There are some unused pins which can do the serial job.

Here is my eagle Mapple Mini Library
https://drive.google.com/file/d/0ByIEww3RjUnNYlNldmV2VTNIckU/view?usp= sharing



sneak-thief

The Maple Mini works but you need to cut the trace to the on-board LED and re-route it to a wire that goes to whatever host board it sits on - I wasn't able to successfully remap it for my Peaks experiment.

Also, traces going to the USB either need to be remapped elsewhere or cut and the USB circuit interferes with operation of whatever you connect to it.

Some Maple Mini clones have crappy power regulation. Look for the ones that match the schematics of the original.


Re. through-hole redesign attempts: another thing to consider is that most of the DACs used are only available in VSSOP-8 aka MSOP-8 aka holy-fuck-tiny-as-hell smd packages. Just like the STM32F103CBT6 mcu.


abelovesfun

Here is the image of what @pld mentioned.
Looks like "picard" has already made it so.
Very clever use of a separate board for digital. Oh well applause



bennelong.bicyclist

synchromesh wrote:
I did electrical engineering at university but I'm a software developer by trade, so I have some idea what I'm doing but little recent experience with SMT. So the whole project is a learning experience for me. If I can pull this off then I'd tackle the Braids design next using a similar approach, except for:

* I'd use 0805 footprints for the passives.
* I'd use a *STM32 Nucleo F411RE development board for the CPU. This would add more horsepower for alternative firmware, while presumably simplifying the build.


I think this is an excellent approach. The Nucleo boards are very cheap and licensed to use in products (and for licensing purposes, "Product" includes free, open-source designs, BTW), whereas the STM32 Discovery boards aren't licensed for use in anything that is distributed to any third parties.

A few issues to consider:
* are there any problems porting the existing Braids code to run on the STM32F4 processor? In theory, no, but some fiddling with the GPIO drivers will be needed, and a close look needs to taken at the hardware abstraction libraries for both the STM32F1 as used in Braids and the STM32F4, to make sure there aren't any differences (and if there are, accommodate them).
* the clock speed would be different, and some fiddly may be required to get the sample rate correct. Making sure the faster MPU talks correctly to the ADC, DAC and display is potentially the trickiest part - you can't assume that it will just work, because the timing may need adjustment. That's why it is best to prototype this stuff on breadboard first before spending too much time on laying out PCBs. Just replicating the schematic with alternative parts doesn't mean all the components will talk to each other at digital communication speeds without further fiddling with the hardware interface parts of the code.
* as sneak-thief points out, there is still the problem of the DAC with its tiny pins. You can obtain (or make) little carrier boards for such packages which provide break-out pins suitable for through-hole, but someone still needs to solder the DAC chip to the carrier board.
* if the aim is DIY-friendly, I would use 1206 SMT parts. I suspect that 805 parts are still too fiddly for many people. Avoiding SMT caps is a good idea - they are a pain to solder and have a plastic base that is easily destroyed with too much heat from the iron. Likewise the reset and sysboot switches - use through-hole equivalents for those. If you provide big pads, the polyfusas should be easy enough to solder as SMT

If it works, I'd be happy to provide versions of Bees-in-the-Trees that make use of the additional storage, and CPU power. I've been musing about a HyperBraids running on an STM32F4 for quite a while, but lack the skills and time to make it a reality. And I am constantly frustrated by the lack of flash storage on Braids when working on the Bees-in-the-Trees modified firmware.


bennelong.bicyclist

abelovesfun wrote:
Here is the image of what @pld mentioned.
Looks like "picard" has already made it so.
Very clever use of a separate board for digital. Oh well applause


Oh yeah, I'd forgotten about that! Yes, that Olimex board uses the same STM32F103 processor as Braids, so no hardware interface issues to be solved in the firmware. And he uses a little carrier board for the DAC as I described above.

Next step would be to approach picard about open-sourcing his adaptation of the Braids hardware - ask him to make it so. Unless your aim is to create a SuperBraids, or HyperBraids with an STM32F4 processor. That's still a worthwhile aim, IMHO.


makers




abelovesfun

I messaged Picard. I hope he isn't turned off by my email address ([email protected]).


bennelong.bicyclist

The Bit Crusher on the Picard worm-hole version of Braids it will process Data better than the real McCoy.


synchromesh

bennelong.bicyclist wrote:
I think this is an excellent approach.


That's because it was your idea in the first place, Tim! hihi

Thanks for the feedback, that's a great list of points to think about. I can imagine there would be a few bumps in the road, but I'm as interested in working on those (and learning about Olivier's C++ template-driven code generation for embedded synth algorithms) as I am in doing the layout etc.

bennelong.bicyclist wrote:
If it works, I'd be happy to provide versions of Bees-in-the-Trees that make use of the additional storage, and CPU power.


Cool! w00t


puzo

this is a truly excellent idea, very happy to hear. very very cool. one of everything please smile


synchromesh

puzo wrote:
this is a truly excellent idea, very happy to hear. very very cool. one of everything please smile


Let's not start all that again... at this stage it's still just a couple of amateurs with good intentions (plus this "picard" who looks to be way ahead of us both). We're (or certainly I am) a long way from taking orders for anything.

But thanks for your vote of confidence in the meantime! thumbs up


sduck

I'm all in favor of this, and would prefer to have as little smt as possible!


Reality Checkpoint

This sound like a very interesting project which I would be overjoyed to see coming to fruition! we're not worthy


the bad producer

I've just noticed that Rapid are now stocking the Olimax STM32 boards, probably only interesting - if at all - for those in the UK...

MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2, 3, 4, 5  Next [all]
Page 1 of 5
Powered by phpBB © phpBB Group