latigid on

Are you using the Darlington drivers? Are the shorted pads connected to ground or any power rail? The schematic might also help: http://jeromebo.free.fr/Wiki/Schematic.pdf For me, it's quite unusual that only one LED of a matrix is failing; usually it's a whole row or column. As a last resort you could cut the traces and wire manually, considering that this may break more connections in the matrix.

Did you try replacing the LED?

Perhaps try to swap around the shift registers to see if the problem follows. There shouldn't be a short between the LED pins except when the matrix is being driven. Also check the associated resistors.

Just the buttons and encoders? Sure thing. But the illuminated encoders and displays need the big brother. I know there's Arduino code for WS2812s, it just depends on what you want to do with it. Who knows, maybe there'll be a new F7 Core design that could be made much smaller.

Yes and yes. Or even 3 for 48 encoders -- the board width will likely be 150mm, so right at the limit for a 19" rack, but the best "case" scenario will be a clear or smoky acrylic to let the OLEDs shine though.

I'm currently working on a layout using 4*4 illuminated encoders, see some of my other threads. If possible I will integrate the shift registers and displays on the same board (half the PCB cost). I'm just trialling an idea for encoder acceleration at the moment.

Serial chain from J4B, pin 6 (SC) I think.

Something else to keep an eye out for: https://www.kickstarter.com/projects/1304489933/opa-multitimbral-fm-synthesizer-shield-for-arduino/description

Confirmed on my end, I additionally tried to change the port settings to MIDI IN 3 (F1 Core, no note limits) without success. Hopefully it's a simple enough software fix.

If this only includes the silicone and FSR sheets that's still very expensive. If you don't need velocity sensing you could think about using adafruit (e.g. BLM) or Sparkfun (can be WS2812 enabled) pads which are much cheaper. Something I don't think you've considered yet is the sensor design. You need interleaved PCB traces (ENIG plating is best) to detect different resistances as velocity changes. Unless you can clone the AKAI PCB exactly you'll need lots of trial and error to get it right. Core STM32F4 has 8 ADCs but they're apparently not that good. There's a board called AINSER which runs over SPI (I think) and might do a better job, but transfers are slower. If money and time are no problem, go for it of course. But if you want something that works for cheap I'd consider an existing solution e.g. Novation Launchpad Pro (8*8, illuminated, velocity/pressure sensitive) for <$300.

Good for DIY, but for the price I would buy a pre-built unit (e.g. MPD218 = $100) and just send the MIDI out.

Jumping in headfirst! I think I confused KiCAD with Diptrace, the latter has limits on size/number of pins unless you pay. All PCB software has very steep learning curves. My advice is to start with simple boards first, and if that works out go for larger ones. It's very likely that you will make an error (I know this from experience!) and it could waste some money if these are big PCBs. Some very basic things about design: Scanning in vector objects to use as layers is not recommended. Boards are a complex stack up of copper, mask and silkscreen layers, and holes/milling. When you connect a net in the schematic editor, this creates an "airwire" for you to join in the board editor. This guides you on how to layout the PCB. PCB fabricators accept gerber files, which are the different layers in an vector format, standardised. Few take layout files. I know the gerber generation in EAGLE is pretty good. Start small, dream big!

My MBFM knowledge is limited to a sammichFM, but I'm sure the routing is also depicted graphically on the CLCD.* As I said already, this is not very intuitive and would be great if there was a more graphical depiction on OLEDs of all waves/routing/levels etc. This is more a limitation of having one encoder and 4 buttons as a control surface, but I think even with the full v1 FM CS that TK. designed it's still tricky. *It might be a good idea to first build basic Core/FM modules and see how the MIOS8 software works. It's great that you're very motivated, and you will need that for a project like this. Expect many hours/months spent and probably one or more buggy PCBs the first time that you order. If you haven't done PCB design before it's a steep learning curve. If you don't have a professional version of EAGLE you'll have a lot of trouble designing a PCB this size. I'm not sure of a PCB design package that is free and allows for large boards. You see straight away the benefit of 32-bit: only one Core. With two you need to work in CAN bus, MIDI implementation etc. In my opinion, it only makes sense to use the 8-bit Core for existing applications, or maybe when you need a good ADC. No one is going to stop you, and of course any new work is definitely encouraged, but from the start the technology is on the verge of obsolescence. We have to look at history here: there are a lot of people who dream very big, only to run out of time, patience or money along the way. It's not to say that you'll be the same, only that if you make things easier for yourself they will have a much greater chance of success. There's no problem (in my view) in keeping the discussion running, it's really a good thing to see projects and ideas flying about.

The waveforms are already depicted on the CLCD, it's just a bit ugly. :) TK. would be the one to ask, but I think the PICs are interchangeable in the Core, so just use the recommended one. How deep are you going? I think all of the routines are available in MIOS and could be reused (up to you however). I think the PCBs were designed in KiCAD, and the layouts aren't publicly available. Did you decide to follow the MB-6582 form exactly? Are you using a PacTec case? I think you need an admin to create a wiki account for you.

YMF262 manual: http://map.grauw.nl/resources/sound/yamaha_ymf262.pdf

It's a commendable idea and I wish you the best of luck! There are a few points to consider though. When Wilba first designed the MB-6582 (as far as I know) companies such as Ponoko/Formulor didn't exist or weren't well established. Your options for cases were limited to alu front panels from Schaeffer/FPE and a 40cm deep rack case -- quite impractical. Now there are good alternatives to make plastic cases, see for example the Mutable Instruments DIY stuff, sammichSID/FM, Sonic Potions LXR, Yocto etc. etc. Even complete folded metal cases (BLM, TubeOhm Ambika) are possible with minimal cost and effort. For me, it's important to design to maximise playability and longevity, which might otherwise be wasted on a device which sits half-finished, broken or unwanted in the corner. The best way forward is (I think) to design all of the parts from the beginning, so how the front panel works with the case, the connectors, heat management etc. I would propose sandwich-type connectors over the SIL ribbons used in the MB-6582. Much more durable and you can take the thing apart if needed. You can still use an angled case, just the IO will need panel mount or another PCB. Think about modular blocks, like OPL3 and Core8 sub-boards that plug into a CS. It's cheaper to buy multiple copies of PCBs rather than large ones and can make troubleshooting easier. Shift registers could be placed on the CS PCB (e.g. SEQ) or like a backpack as I did for my MBCV concept. Something that a 32-bit Core might be useful for is display options. Part of the mystery of FM synthesis is how all of the operators/modulators interact to create your sound. If there were graphical representations of waveforms/levels/routing I think that would be a great contribution. E.g. look at the SSD1306 0.96" OLEDs in use in MBCV/Programma with SPI drivers already well integrated into MIOS32.

Ah, I see the MF_NG module now, with the strobe line pulsing from RD4/pin 27. This makes perfect sense.

First off, did you upload the correct firmware using the Wilba variant?

Looks nice! Is there a current method of using touch sensors for the faders on a 32-bit Core? We're missing the J14 strobe line present on the old PIC Core. Seems to have been ported in software at least.

They're very useful and cost 1 cent for 1000 :). If you ever go for the full SEQ BLM that's 1/3rd gone!

Hello again :) No problem to use the switch as an input for a DIN. Just connect one side to ground and the other to the appropriate pin on the module. There's no need for diodes unless you're configuring a switch matrix.

According to adafruit, there are at least two LEDs with different names that use the same controller (or an equivalent). I think they should work; you should choose the 5mm LED as 8mm is too large. At some point I will sell the PCBs to suit (don't buy Sparkfun PCBs unless you want to use common RGB LEDs), price should be about 8 EUR. If you read through TK.'s code example on svnmios32, he states that driving the LEDs takes up quite a bit of memory but not too much CPU overhead. There is a limit in the chain length in order to guarantee data transmission. I'm pretty sure that the controllers hold the colour value and new data is only required for updates. The advantage of shift registers is the parallel data transmission which allows for a simple serial stream to rapidly clock multiple lines. Disadvantages are the complicated PCB and software matrices, but it does work (BLM 16*16+X). Bon weekend,

If you can verify that the hex is written, it should be okay. I think the point of selecting the device (manually) is that the burner knows what programming voltage to use.

Did you find the build guide page? http://www.midibox.org/dokuwiki/doku.php?id=wilba_mb_6582_base_pcb_construction_guide You want to follow "Power Option D" http://www.midibox.org/dokuwiki/lib/exe/fetch.php?media=mb-6582:mb-6582_base_pcb.pdf This is the PCB layout, so you should either jumper pins on the power switch directly to J4 or add bridges e.g. to bypass the 9V regulator. Hope that helps,

DOUT can also provide a type of PWM, but it only works well when the LEDs are addressed individually (i.e. not in a matrix). Which means quite a lot of DOUT pins. A better method is to use WS2812 LEDs which respond to a serial data chain. See the end of that thread for working layouts with 4*4 Sparkfun button pads and illuminated encoders.