MIDIbox CV V2 Concept

[ilmenator]

Oh, I see - I didn't realize that your comment was exclusively CV V2 related! :smile:

[TK.]

As promised I tested the bipolar (RS422-like) MC3487/MC3486 based line driver today, and it works! :smile:

 

 

At least with a 2m cable (ribbon, not twisted pair), but 5m should be feasible as well (has to be confirmed by somebody else)

 

With a scope I can clearly see the effect of the long transmission line, how the bipolar transmission and especially the termination resistors help, etc. - I've a good feeling! ;-)

 

Schematic: http://www.ucapps.de/mbhp/line_drivers.pdf

 

This way you can transmit the signals for the AOUT* module and DOUT (gate/trigger/switches) over long distances.

 

Best Regards, Thorsten.

 

 

[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...

 

[sneakthief]

latigid on - the DOUT board can now be moved alongside the AOUT board, so there's no longer a need for 8 gates over that cable.

 

They both can use their own pair of those inexpensive line drivers (MC3487/MC3486) - so one pair for the AOUT and one for the DOUT.

 

TK - We don't need RC2 from J8/9 or J19 for the DOUT or AOUT on with the MIDIbox CV 2, right?

 

 

 

Here's a quick proposal (revised to theoretically prevent crosstalk):

 

AIN

1.         1

2.         2

3.         3

4.         4

5.         5

6.         6

7.         7

8.         8

9.         GND (analog inputs)

10.        Clock Input

11.        GND (clock input)

 

DOUT via MC3487 #1

12.        Vs

13.        Vd

14.        SO

15.        SO#

16.        SC

17.        SC#

18.        RC1

19.        RC1#

 

AOUT via MC3487 #2

20.        SO

21.        SO#

22.        SC

23.        SC#

24.        RC1

25.        RC1#

 

 

PS. I corrected my Analog-input post and pinout to match TK's corrections above and below

Edited February 2, 2014 by sneakthief

[TK.]

Clocks and Start/Stop are assigned to DOUT pins as well, you only need to add the clock input.

 

I would also use a separate (especially shielded!) cable for the analog inputs, otherwise there will be a lot of crosstalk from the digital signals.

 

Best Regards, Thorsten.

[latigid on]

I was considering a dual use board, one for MBseq and one for MBCV. This is for us old farts who are still on STM32F1 cores!

 

So we could get away with a 15 pin connector for the digital serial signals and a 9 pin for AINs.

 

 

 

[sneakthief]

TK - I rearranged the ribbon cable pinout so that there are now several wires separating the analog and digital signals.

 

This obviously isn't as good as shielding, but my guess is that crosstalk will be negligable since modular synths use 10Vpp which will then be attenuated 3:1 (0-3.3V) before hitting the ADC.

 

I really can't see digital interference hitting the analog lines enough to actually make any noticeable difference to the inputs of 12-bit ADC's.

 

Want to bet? :D

[latigid on]

Quick and naive question: are the serial clocks independent; could they be shared too?

 

[sneakthief]

No - judging by the pins going to each of those serial buses, the AOUT and DOUT appear to be completely separate.

Edited February 2, 2014 by sneakthief

[TK.]

Quick and naive question: are the serial clocks independent; could they be shared too?

 

They are independent and can't be shared.

 

Best Regards, Thorsten.

[latigid on]

Hmm, I had a try at putting 32 sockets on a 4.05" board (basically the max. height for a Eurorack PCB):

 

8 gate

8 CV

8 AIN

7 Clock

1 Sync in

 

and it's pretty tight even at 2.2" (for a 12 HP panel), especially with another multipin connector and mounting holes. 

 

I suggest a DB-25 (or whatever connector) is mounted on a PCB with all of the interfacing circuitry (line receivers, analogue input protection circuits, pass-through for AOUT and DOUT). All of the other sockets could be panel mounted and wired to MTA 100-style connectors or similar.

 

I know the trend is towards super shallow modules with no panel wiring but this would give a better return on investment for PCB space. Also it is customisable and more flexible for other modular formats (5U, Serge etc.) or even standalone MBCV enclosures.

 

 

 

 

 

[eptheca]

The STM32F4 version of the MIDIbox CV 2 uploaded successfully and MIOS is happy:

 

Operating System: MIOS32
Board: MBHP_CORE_STM32F4
Core Family: STM32F4xx
Chip ID: 0x10016413
Serial: #32002C000147333139303639
Flash Memory Size: 1048576 bytes
RAM Size: 196608 bytes
MIDIboxCV V2.000
© 2014 T.Klose

 

Also, for the MIOS32 toolchain ( http://www.midibox.org/dokuwiki/doku.php?id=windows_mios32_toolchain_core ), I modified the ENV variables to correspond to the STM32F4:

 

set MIOS32_GCC_PREFIX=arm-none-eabi
set MIOS32_FAMILY=STM32F4xx
set MIOS32_PROCESSOR=STM32F407VG
set MIOS32_BOARD=MBHP_CORE_STM32F4
set MIOS32_LCD=universal

 

Should I post this update here? http://midibox.org/forums/forum/13-midibox-documentation-project/

 

I just added the variables to the wiki, http://www.midibox.org/dokuwiki/doku.php?id=windows_mios32_toolchain_core&#additional_environment_variables

Hope that's the appropriate way =|

 

My build failed, until I found this in your post

now I have CV V2 up'n runnin as well

trying to follow "in the footsteps of giants" the best I can

 

Cheers, Halvor 

[sneakthief]

eptheca - make sure to use the latest SVN. The AOUT was changed a couple of days ago.

 

I'm still waiting for a couple of parts to come so I can finish all the measurements. The minimum feasible inside height of the case is 40mm because the chips under the LRE8x2 board need to clear the STM32F+carrier board.

 

Here's my work-in-progress:

 

(TK - yes I know it should be MIDIbox instead of MIDIBOX. This font doesn't have lower case and I haven't found an appropriate replacement yet)

 

I was inspired by FFW's SeqV4 - I'll probably use clear-grey windows for my CLCD, OLEDs and encoder-ring cutouts:

 

 

I'm going to glue the circular cutouts on top of these buttons (which are a few mm too short):

 

Edited February 5, 2014 by sneakthief

[eptheca]

Thanks for the heads up

 

Looks great

especially those tinted windows :)

His SeqV4 is such a "beast" 

 

I'm waiting for some parts as well, Chinese New Year?, 

[sneakthief]

[Altitude]

You should consider just making holes for the LED rings in the front panel over the window approach.  Just use the front panel as an aligning feature.  You can get pretty accurate results using the FP as a template, it even works with flat top LEDs

Edited February 6, 2014 by Altitude

[sneakthief]

I received my LRE8x2 with the LEDs on already soldered in and too short to reach the front panel. I'll let someone else make the individual holes if they want that for their LEDs.and feel confident enough to line up and solder 256 of them.

 

In any case, for those who want to go the route you suggested, I'll try to make a SVG with the LED-holes lined up on top but I won't be laser-cutting a test version of that one.

[eptheca]

I have ordered the OLED's, just have to wait for the Chinese finish celebrating their 15 days of New Year holiday, entering the year of The Horse? (a good year for China White?)

 

On their web site, the yellow line is at the top

on yours and TK's they are at the bottom

is it optional? software or hardware or a different OLED?

[sneakthief]

AFAIK all the mixed-colour OLED's have a yellow line in the same place.

 

When you interface the SSD1306 with a MIDIbox, the display is technically upside-down.

 

This is how you connect them to the STM32F4:

 

http://www.ucapps.de/mbhp/mbhp_lcd_ssd1306_alt_port__stm32f4.pdf

 

 

[TK.]

is it optional? software or hardware or a different OLED?

 

This is a software option, in the bootloader configuration you could mirror the Y axis if you want (but I prefer to have the yellow line at the bottom)

 

Best Regards, Thorsten

[novski]

Hi

I just folowed this thred and noticed some similar usecases from what im trying to establish.

What i need is a 8 OLED and 8 Encoder with LED rings on one board. So i read the CV2 describtion and coldn't find the reason why CV2 just needs 4 OLED and 1 LED Display. Wats the reason?

Best regards

Novski

[TK.]

MBCV supports only 4 OLEDs because the scopes consume a lot of CPU time. Adding more scopes would increase the latency for display updates so that they are not usable anymore.

 

Another reason is, that the OLEDs are connected differently compared to the normal approach.

E.g. on the STM32F4 based core, they are connected at J10B which has only 8 data lines, which are fully allocated (CS/SCLK/DATA/RESET/CS1..4)

 

In general, MBCV is not comparable to a MBNG based MIDI controller, therefore the design constraints might confuse you ;-)

 

Best Regards, Thorsten.

[novski]

Oh! thats a shame... (schade lässt sich nicht direkt übersetzten und shame ist für Deutsche missverständlich aber ich meine damit "schade"...) 

So we have to do almost the same for MB_NG again. :sad: Will that change with the support of STM32F4 for MB_NG somehow? 

best regards

novski 

[TK.]

You wanted to say "What a pity" ;)

 

I haven't fully understood what you want to achieve. A MIDI Controller with multiple OLEDs and Encoders+LED Rings can already be created with MIDIbox NG regardless of the core.

 

MIDIbox CV is not a MIDI controller, and the OLEDs are used for a different purpose. They don't show messages (like MIDIbox NG would do), but show oscilloscopes for the internally generated waveforms.

 

Best Regards, Thorsten.

[novski]

Oh! I din't realize that there is a totaly different usecase...

Best regards