latigid on

Great job TK! What about this for the cable pinning?: Unfortunately this leaves no room for clock signals (DIN sync/start). Shall we move to a larger cable like DC-37? http://en.wikipedia.org/wiki/D-subminiature Otherwise a higher density version might be an option, the problem then becomes finding a source for the more esoteric connectors...

Yep, a fair amount of work... http://processors.wiki.ti.com/index.php/CC3000_Serial_Port_Interface_(SPI)

Hey, They also sell capacitive TS overlays, just check through the menu on the left hand side. As far as I know these will also work with multi-touch gestures. As the interface is USB I guess all of the scanning hardware is compacted into the frame. Best of luck and let us know how you go! Many Greets, Andy Ah, maybe they're infrared??

I was thinking of a non-tablet solution for a MIDIbox touchpad and came across these: http://www.cyclotouch.com.au/product_info.php?products_id=2468 (you can also check out the other kits). My thought was to design a 16x16+1 LED array PCB, then overlay a transparent panel with the touch matrix interfacing to a Core module running the BLM code. In the end, a commercial tablet already has an integrated processor, WiFi and a long-life battery for wireless connectivity.

New topic: How about skipping ethernet altogether and going directly to WiFi? Adafruit offer a cheap breakout module running over SPI: http://www.adafruit.com/products/1469 Is there a free SPI buss for this purpose? I notice the STM32F4 board does not yet have an ethernet connector.

Even when that signal is buffered? JBdiver reports successful signal transfer to his breakout module using only the 541 buffer of LPC17

Let's just think about this a bit more: 8 gates, but up to 64 with DOUT/DIO_MATRIX. DIN SYNC is 1 start/stop but up to 7 clock divisions/multiples. AOUT is really 6 signals (Vs, Vd, SO(SI), SC, RC1 and RC2) but Vs is common to the Core and doesn't need a separate ground. 1 sync input to control the master clock/ forwarded to the additional clock outs. What about analogue inputs? These would be pretty cool to interface from a modular system back to the Core. Diode clamps to protect the input pins? We're getting into quite a few signal lines now, too many for a DB-15! But for a basic usecase: 8 gates 5 AOUT 1 clock start 7 clock out 1 sync in 8 AIN 1 signal ground 31 pins ---> DB-36/Centronics? With only one clock out you'd be back to DB-25.

Okay, it makes sense for all the gates to be on J5A/J5B (different to the other Cores). Right, I see the MIDI IO on J11E. For MBCV V2 I guess J5A/B will be used as analogue inputs then? >>> Maybe I will create a board for STM32F1 and LPC17 and we can sort out another one later for F4. Best regards, Andy

TK, do you have thoughts on pinouts for the "J5C" connector on STM32F4? I will put together a DB-15 breakout/level shifter board for Gates 1-8, DIN Start and Clock, and MIDI OUT 3 (no room on the DB-15). Will it be more like STM32F1 or LPC17? Thanks, Andy

The back of the board is a better place to put them, although you only really need a header for J1 (/J2), the rest could be wired directly. Try sliding off the plastic base (or cutting it) and push each pin out individually. Divide and conquer is the motto for multi-pin desoldering.

Getting closer: Data lines are okay?

Okay, I will not use my version in this way as the connections will be "sammich" style to another board. But I agree, polarised connectors make good sense. Unfortunately that makes the board bigger, but I needed to give proper clearance to the mounting holes anyway. So I moved the trimmers around and that leaves room for another connector: J2 for serial output. Can I please check pins with you because the AOUT_NG does not seem to match 1:1 with the newer Cores' J19? AOUT : Vs Vd CS SI SC AOUT_LC : Vs Vd SO SC RC AOUT_NG : Vs Vd SO RC MD (?, from wiki picture) : Vs Vd FS DIN SCLK (connections to DAC on schematic) AINSER64: Vs Vd DI SC RC1 Vs Vd DO SC RC2 _____| |_____ CORE(32): Vs Vd SO SC RC1 Vs Vd SI SC RC2 _____| |_____ So, are these connections correct ? CORE(32): Vs Vd SO SC RC1 Vs Vd SI SC RC2 ______| |______ AOUT_EURO J1: Vs Vd DIN SCLK RC1 J3 chip select Vs Vd nc SCLK RC2 RC1/RC2 => FS ______| |______ AOUT_EURO J2: Vs Vd DOUT SCLK RC1 Vs Vd nc SCLK RC2 ______| |______ Sorry for the confusion, there is essentially one ground (both top and bottom layers) but GND and AGND are connected only at one point on the board. There is a "bridge" for Vd and Vref to travel on. My question was: should this bridge be on both top and bottom layers? Thanks! I'll keep chipping away at it.

PCB layouts without and with GND plane: Routing was a bit tight for this format, but I think it works okay. I added a chip select line and put the DOUT signal on the same connector. Hopefully this is okay if somebody wants to chain more than one module. I isolated AGND and GND, and both top and bottom planes are connected. Is this a good idea? On the AOUT_NG PCB only the top layer is connected.

Yes, I've made one

Très cool! Tu as utilisé un Core 8 ou 32 bit?

Routing everything now, a good holiday project! Question: will any application make use of J2 (serial out) for chaining multiple modules? It seems unlikely so I would prefer to remove this connector. In this case, would it be better to terminate the digital out in some way (tied high or low)?

No, I have an offset wired from the +12 V buss, divided through a trimmer. The offset is mixed through an op amp summer, so load changes shouldn't be too much of an issue. This also uses a more stable inverting amplifier that needs to be inverted again for correct signal phase/polarity. The perfect place to introduce an offset voltage!

And another one, I can't see why this couldn't be compatible with MBCV v2 either:

A lot to read, been away a few days! For the breakout board, please remember this: