gm5/core trouble

[rosch]

hey guys, seems like i've really messed things up yesterday:

i wanted to test the crystalfontz 4x20 LCD that was so cheap at divineo.com (CFAH2004A-GGB-JP)

i wired it like usual for 4bit mode and i'm sure there's no short or twisted cable. i've checked both.

then i edited my setup.asm file again for 4 lines display and changed nothing else, compiled the hex and uploaded that via gm5x5x5, which worked.

the status bar stuff appeared, albeit in the two lowest lines (below the parameters)

when i tried to adjust contrast and brightness there was suddenly noise in the audio that could be taken away again by lowering the brightness. turning it up again brought back the noise.

then the brightness and contrast started to re-adjust themselves, then some symbols appeared at random spots (increasing number of them). first arrows pointing to the right and later i believe this guy: |

at some point the sidV2 app crashed and didn't reboot.

i repowered it once or twice then it came up again.

i tried to upload the firmware again. result: some weird massive flood of data on the midi in (mios2.0) that didn't even stop after switching off the core. when i powered it up again mios couldn't connect to the core anymore, though playing with the mios midi keyboard worked fine.

then i tried the next gm5 midi input, same data tsunami, after that the in is dead.

now all ins seem to be out of order. i've tried changing a 6N138, but no change.

so i guess i have somehow killed the gm5 chip

or is there a chance that there's not that much drama?

also seemslike there's some problem related to my core in the first place.

but the gm5x5x5 ins also don't work with other synths....

edit:

i'm talking about the sidV2 firmware, latest revison (.37), and it has worked fine before.

one thing to mention, which makes me think i could have had some problem with the gm5x5x5 before, is that eg i've never been able to connect the sid to the sidV2 editor. that was before the crash yesterday.

Edited May 30, 2011 by rosch

[JRock]

I had problems with my GM5x5x5 because I didn't have the chip soldered properly. I'm sure you checked that though...

[rosch]

i wouldn't know how to check that, i mean in an empiric way. of course i checked for bridges etc and was afraid i could have overheated it. but it worked all the time, and since yesterday i have this problem. it started when i was adjusting the lcd brightness. could it be that the lcd itself (maybe incompatibility of this device?) caused the problem? or maybe something happened on the core when using the trimmer, i don't know.

let's say it worked all the time for usual midi play, and also for midibox firmware uploads. i had some trouble connecting to the core with the sid editor, but it could also be i've done something wrong on the software side, settings maybe.

i'll have a close look at the functioon of the gm5x5x5 today and try to find out if and what's happening on the core.

[Hawkeye]

maybe try this? (eliminate possibilities)

- test the sidv2 with another midi-interface

- test the gm5 with any other midi synth

- swap back to the old lcd

greets,

peter

[rosch]

i don't have another interface at hand. and very unfortunately i can't find my other 2 gm5 chips anymore, while i have two empty gm5x5x5 boards.

i've tested it with the Shruthi-1, normal midi play works fine, via Klee sequencer for example. also the brand new Shruthi-1 editor (windows) works in terms of sending CC-messages for synthesis parameter, but i can't send a patch to the synth.

midi in of the gm5 also doesn't work from Shruthi.

sending firmware to that synth worked, with elektron C6 application.

i've changed the old 2x20 immediately and tried again. there were no random symbols anymore, and no noise, although the pic18F4685 still carries the 4-lines modification in the firmware, as i haven't been able to upload the old file, at least the normal way. i can try that again without handshake, maybe that works.

i'll also compare the core with another one, with regard to voltages. i guess there's something wrong, as i do not think a (maybe false or insufficient) lcd declaration in the firmware should have effect on midi out, or even interfaces, idk

thanks for the input! any ideas highly appreciated!

[Janis1279]

edit:

i'm talking about the sidV2 firmware, latest revison (.37), and it has worked fine before.

Hi Roger ,

May be I am wrong. My midibox source files are on the another PC .

I remember. When I connected 4x20 LCD to the sidV2 with setup_6581.asm (setup_6581.hex) I had problems with the displaying all 4 lines. Then I changed this part :

;; LCD line -> cursor offsets

#define DEFAULT_LCD_LINE_Y0 0x00

#define DEFAULT_LCD_LINE_Y1 0x40

#define DEFAULT_LCD_LINE_Y2 0x14

#define DEFAULT_LCD_LINE_Y3 0x54

with the similar part from the setup_6582.asm file

;; LCD line -> cursor offsets

#define DEFAULT_LCD_LINE_Y0 0x40

#define DEFAULT_LCD_LINE_Y1 0x14

#define DEFAULT_LCD_LINE_Y2 0x00

#define DEFAULT_LCD_LINE_Y3 0x54

It helps for my setup.

Regards,

Janis

[TK.]

thanks for the input! any ideas highly appreciated!

I would propose following SysEx checks on the GM5 module:

1) bridge J5:I1 to J5:O1 with a jumper or a small wire. This will create a loopback at TTL level

Open MIOS Studio, select the first IN/OUT port - a feedback loop should be detected: http://www.ucapps.de/mios_studio/mios_studio_feedback.png

Now test SysEx transfers by typing some arbitrary strings into the MIOS terminal input window as shown here: http://www.ucapps.de/mios_studio/mios_studio_feedback_sysex.png

Does this work?

Yes:

2) do the same test, but with a loopback via the MIDI IN/OUT socket (remove the jumper at J5)

Does this work?

Yes:

3) turn-off your MBSID and connect the MIDI cables.

In MIOS Studio: select the .hex file and press the START button.

You will get a notification that no response from the core has been received (of course, it isn't powered...)

Now turn on your MBSID - the upload should be automatically started via 1st level bootloader.

This method is a fallback solution which should always work, even if the application is crashing.

Best Regards, Thorsten.

[rosch]

thank you, TK!

ok here we go:

without jumper i have always connection from O4 to I1 (i checked them all)

with jumper there's following connections:

O5-I5

O4-I1

with cable these connections (no jumper):

O1-I2, I3

O2-I2, I3

O3-I2, I3

O4-I1, I2, I3

O5-I2, I3

(i checked all outs with every in)

all midi loops were sending and receiving the long strings correctly.

so i hooked the MBSidV2 to the O3 and I3 and managed to upload the previously working firmware (with the previously working LCD) and it was also detected by MIOS Studio.

thanks a lot so far, but seems like there's quite a mess on my board

one thing i maybe should mention is the upload didn't start automatically after powering the sid. instead the application booted normally, then i started the transfer and that worked.

Edited May 5, 2011 by rosch

[rosch]

could be i was wrong with successfully uploading the sid firmware, it might also have been an assumption, after all (as the previous application still existed)

today i tried to upload midibox 64 and midio 128 firmwares to 2 pic18F452 i just burned the bootloader on.

the connection is there, the pic 18F is recognized and when i upload the firmware it seems to work.

after it's finished mios2 says "waiting for reboot", but the reboot never happens.

then it reports "bootloader up & running" again

so the gm5x5x5 seems to work, and not to work, at the same time...

i'll hook that sucker to a scope and see if i can find anything.

Edited May 9, 2011 by rosch

[TK.]

confused... let me meditate about this issue ;)

Best Regards, Thorsten.

[rosch]

i must say i'm not at all sure about my soldering. there could have been a problem from the very beginning which i just haven't noticed.

anyway, no hurry here!

thanks!

[lucem]

confused... let me meditate about this issue ;)

Best Regards, Thorsten.

Oh, the good old days of guru meditation ;)

[rosch]

hm i've built another one, red pcb (i kept it all stock this time, with leds)

1st time i plugged it in the main led lit up (although not too bright) and windows informed me that a "USB Verbundgerät" had been installed and it would be better to reboot the pc, which i did.

after that there was no main led on anymore. i tried to send some midi notes to a synth but failed.

strangely, when putting the usb plug in, the following led sequence happens (video)

actually, the smt soldering job seemed to have worked well & easily. but i might have messed sth up...

could it be there's something wrong with the optos? (as only midi ports 1, 4 and 5 seem to light up on start (should they do that at all?) (or maybe somth wrong with some leds, duh?)

so the main led goes off after half a second, but the power out pins show 4,98V still.

wrecked?

am i the midibox dummy??

edit: i forgot to mention, i used the same 6N138's and hexinverters from the old one, as i don't have any others here right now (order them today)

http://www.youtube.com/watch?v=lNXZIR5u1PM&feature=youtu.be

Edited May 30, 2011 by rosch

[TK.]

It's strange that Windows asks you to reboot the PC, this never happened at my side with USB MIDI devices (regardless if GM5 or my own USB projects are used).

But I must also say, that I typically test the devices under WinXP and Win7 only, no Vista (are you using Vista?)

It's normal that the GM5 status LEDs only is lit as long as the USB MIDI driver hasn't connected to the core - thereafter the LED turns off.

It seems that your Windows connects very fast - thats ok!

The MIDI IN/OUT LEDs are flickering randomly at power-on, this also seems to be ok...

In order to ensure that the problem isn't related to your synth: could you please try a loopback test?

(e.g. with MIOS Studio)

Try it also with the GM5 windows driver, just to exclude that any other driver on your PC modified the original USB MIDI driver Microsoft

(e.g. some people noticed this with a Logitech Webcam driver)

Best Regards, Thorsten.

[rosch]

thanks, Thorsten, i got it working now!

the soldering looked really clean, but i thought maybe there's a bad contact somewhere.

so i went over the whole chip again, with flux and solder wick, after that it worked!

i made the feedback tests, and all ports work fine in mios studio (haven't tried the strings yet, though)

now it seems to me that the behaviour of the leds was actually normal.

i've just tried normal midi use but i'll try to upload some firmwares tonight.

really glad it's working, so far.

thanks for your help!

[TK.]

Hooray! :thumbsup:

Best Regards, Thorsten.