Multidisplay missing data. Not working properly *solved

[landracer]

I've been trying to finish off my multidisply, but I am having issues. I am not an experienced coder or debugger, so I need help figuring out my current issues. I feel I have been as thorough as I can possibly be with my current knowledge set. I would love to see this circuit working! I want to run this in my other race car.

So I have 4 different computers setup to compile this multidisply code inside eclipse. 2 different KDE 14.04 systems, your VM provided for download, and a KDE 12.04 system. All ubuntu variants with the necessary dependencies met and latest SVN version of MD code. All againist Arduino 1.0.1 libs. arduinoDuemilanoveCore trunk ver923. And MD ver874.

SO the binary multidisplay.elf size will vary on 2 systems given no changes at all, this includes your VM
laptop 14.04 KDE with Luna 4.4.1
multidisplay.elf 82,396k
Xeon Workstation KDE 14.04 Eclipse 3.8
multidisplay.elf same 82,396k
Your VM Eclipse indigo 3.7.2
multidisply.elf 132,555k
AMD FX system KDE 14.04 Eclipse indigo 3.72
multidisply.elf 84,317k

I am attaching the build logs so you can tell me which ones are correct, or none of them at all... 2 of these computer systems during AVR upload and verifiy, upload a different set of bits. The .hex that seem right are both 37,548k. The other 2 .hex's in question are 37,754k and 37,548K. A few hundred K difference between systems no change in the code downloaded from SVN. I need to know what is right and wrong. At one point before I started over I was get 75,000k AVR uploaded HEX files. Werid, yeah, I am still a bit confused. But I believe the 37,548k AVR HEX is the closest I've come to the correct HEX file upload via AVR.


Ok, the other thing I did NOT do was get myself a seeedmega... Why? because I had 2 Mega2560's and a ADK board sitting around I figured why not add some header extenders and some wires. I'd love to make this all come together.


https://plus.google.com/116091853105558 ... eEKNEyFyar Link to MD Google+ post for more pics.

Ok to sum it all up MD02/03 boards are not talking to the Arduino.
IF I had to change code to make it work, I didn't because I know nothing.... Seems nothing is communicating over SPI like its suppose to. Using pin 26 clk pin 12 MISO and 57/A4 MOSI. SDA and SCL pins for the PFC8574 are connected to Arduino pins 20 SDA 21 SCL.
I might have purchased the wrong PFC8574AN I2c chip and or MCP3208ci/p, because there seems to be a slight difference in part numbers. The ones I've used are listed below.
2x MCP3208 CI/p
http://ww1.microchip.com/downloads/en/D ... 21298e.pdf
and one PCF8574AN
http://www.ti.com/lit/ds/symlink/pcf8574a.pdf

Not sure what direction to go, but MD doesn't work... Well unless I want to log voltage... That seems to work inside MUI. Please help!!!

Thanks a million in advance!!!!!!!!!!!!!!!

[landracer]

Other logs attached. It would only let me have 5 attachments. These are the build logs that seem responsive and poll data in MUI, just no data is received. I have a potentiometer attached on the boost terminals... Just no response in MUI.

Any help or insight would be greatly appreciated. Again, thanks a million!!

[erikiiofph7]

Actually the onboard boost input is not activated with the default settings. In the trunk/eclipse/multidisplay you have the file MultidisplayDefines.h, and on line 31-34 you'll find this:

Code: Select all

//choose your engines ecu
//only one of them should be defined!
//#define VR6_MOTRONIC
#define DIGIFANT

When DIGIFANT is the one that is defined MD will not use the onboard boost input, but ask the ECU over the K-line for the boost. The VR6 settings use the onboard boost input though, so if you uncomment the VR6_MOTRONIC and the comment DIGIFANT lines like this you'll have a better chance to get some response:

Code: Select all

//choose your engines ecu
//only one of them should be defined!
#define VR6_MOTRONIC
//#define DIGIFANT

[bofh]

I might have purchased the wrong PFC8574AN I2c chip and or MCP3208ci/p, because there seems to be a slight difference in part numbers. The ones I've used are listed below.
2x MCP3208 CI/p
http://ww1.microchip.com/downloads/en/D ... 21298e.pdf
and one PCF8574AN
http://www.ti.com/lit/ds/symlink/pcf8574a.pdf

Not sure what direction to go, but MD doesn't work... Well unless I want to log voltage... That seems to work inside MUI. Please help!!!

Thanks a million in advance!!!!!!!!!!!!!!!

its essential to use exactly the specified pcf8754P.
I think the wrong pcf8754 causes the problem.
mcp3208 ci/p are the right ones

[bofh]

So I have 4 different computers setup to compile this multidisply code inside eclipse. 2 different KDE 14.04 systems, your VM provided for download, and a KDE 12.04 system. All ubuntu variants with the necessary dependencies met and latest SVN version of MD code. All againist Arduino 1.0.1 libs. arduinoDuemilanoveCore trunk ver923. And MD ver874.

SO the binary multidisplay.elf size will vary on 2 systems given no changes at all, this includes your VM

Are these machines running different gcc-avr versions? If yes it's not unusual that the binary size differs slightly.

[bofh]

I had a quick look into the PCF8574 datasheet : the "A" versions have a different fixed portion of the slave address (I2C bus).
In MultidisplayDefines.h: please change the line

Code: Select all

#define EXPANDER2 B0100000      // Address with three address pins grounded. (from the PCF8574)

to

Code: Select all

#define EXPANDER2 B0111000      // Address with three address pins grounded. (from the PCF8574)

and test again.

[landracer]

....

[bofh]

PCF8574T is a replacement for PCF8754 according to digikey. But its an SOI16 SMD device; a little adapter pcb from SOIC16 to DIP16 has to be used (until we make a new md02 pcb release).
The PCF8574AN DIP16 is still available. This would be de easiest fix. Would be great if you could verify that its working with the small code change.

MD02 LED D2 is now on. The data is all over the place up down up down, crazy. the EGTs are a claim zero and change with slight touching, seems like that might work if I test later with an EGT. None of this was responding like this before the i2c address change.

do you get plausible temperature values? the temp-sensor is located on the md03 board; if you get right temperature values (°C in current code) the connection between md01 und md02 should be working.

[landracer]

Well that was fun....

I changed the few things I screwed with back to original(SPI ports) and changed the i2c address. It all seems to be working now. Nothing erratic and smooth transitions with the potentiometer.

So to confirm; (line236 in my current multidisplayDefines.h code)

#define EXPANDER2 B0100000
to
#define EXPANDER2 B0111000

the B0111000 is the base 0x38 address, for the PCF8574A chips. I found this a good brief read on the PCF8574/PCF8574AN.
http://www.ti.com/lit/ml/scyb031/scyb031.pdf

Worked just fine Thanks for the help! Hopefully this will help anyone else using the PCF8574AN in the DIL16 IC6 socket on the MD02 board.(This is most likely considering the PCF8574AN is a drop-in replacement, readily available.)

And to answer some of mine own questions, my final AVR upload build size was 36,096k flashed. The multidisplay.elf sized was 80,807k. Not that any of that really matters. Just an FYI for anyone curious. I was not able to compile the code properly inside the multidisplay VM. But, I am ok with the code working in my native OS. No complaints with that, and where I'd actually prefer the code...

Now I can start playing with this thing. Hopefully MD will make it into the race car(or something like it).

You guys are awesome! Thanks for such an inspiring and extremely fun Arduino project to complete!!

Now to get this QT app running inside android. Where can I find a how-to for MUI running android??

[bofh]

great news! I'll write the PCF8574AN alternative solution into the wiki.

you are able to compile the source with avr-gcc 4.8? thats interesting because I'm still using an older avr-gcc 4.3 (self compiled with scripts vom avrfreaks) because I have issues with the stock avr-gcc (4.8) on my arch linux.

you just have to give your bt device a name wich starts with "mdv2" and 115200 baud and pair it on your android device. then install the apps apk (not yet in play store; current beta can be found here: http://mycorrado.de/public/multidisplay ... signed.apk ). the app should insist of installing "ministro" service from playstore which den downloads the required qt libs onto the device.
Beware that on the current md01-pcb (052) are rx/tx signals mixed up. its easy to fix with a cable connection to the seeeduino mega.