Sammich SID troubles - midi in - help!

[recompas]

I've already looked through the other posts on the forum, but no luck. I finished constructing my sammichSID last night and was trying to get the SID program uploaded. I followed all instructions, and upon trying to test the display, I had no response. I tried replacing the optocoupler twice, and no luck. Midi Out is working, and MIOS boots up. The message I get in MIOSStudio is:

00000533676095 ms | Continued SysEx: F7 03 00 00 00 00 00

I am running a Macbook Pro

2.33ghz core 2 duo

w

os x 10.6.2

and a m-audio usb uno as my midi interface.

The interface appears to work, as I can use the keyboard in MIOSStudio to play one of my synths via the uno.

Is there a good way to narrow down where the problem may be? I was going to try a loop of the midi in/out on the sammichSID board by moving the midi jumpers and just connecting the in/out (tx/rx) lines to see if it echos back on MIOSStudio, or would this blow someting up?

Any help very appreciated. Is there a good way to test the optocoupler?

Thanks,

Travis

[nILS]

For a loopback test try looping the midi device only to see if it correctly sends out sysex. If that works you can try what you described (it won't blow up anything). If that doesn't work, check the parts around the MIDI in (resistors, optocoupler mounted the right way around)...

[philetaylor]

What happens when you try to upload? What do the MIDI Monitor screens show?

There are problems with MIOS Studio  and the (to quote Wilba) "crappy Mac Java MIDI library". This could be what you are seeing, there has been mention on the forum of installing the latest version of Mandolane http://www.mandolane.co.uk which is much better than the standard Java MIDI library

If this doesn't work, TK is currently writing a version of MIOS Studio written in C++, an early alpha release is available for developers which TK may let you try....

Uploading an app to the Core, requires the MIDI subsytem being able to reliably transfer large sysex streams. Many implementations (both operating system and manufacturers drivers) seem to be quite poor at this which I find amazing!

Cheers

Phil

p.s. As nILS has suggested, you should also double check all connections/components round the MIDI ports :)

[lylehaze]

Uploading an app to the Core, requires the MIDI subsytem being able to reliably transfer large sysex streams. Many implementations (both operating system and manufacturers drivers) seem to be quite poor at this which I find amazing!

It's not so surprising really..

EXCEPT for SysEx, MIDI is a medium ruled by messages of 3 bytes or less.

So once you get that working, ALMOST everything works! Then all you have to do is manage these SysEx messages that could be ANY LENGTH LONG.. so you have to decide how big of a buffer to provide, and how to handle the possibility of messages bigger than that buffer, and all of a sudden your nice, simple little program is a collection of exceptions and workarounds.

And how do you test SysEx message passing? I would start by creating tests that use the smallest and largest possible SysEx blocks.. But, there is NO "largest".. So many programmers will make some compromises.. based on whatever is in front of them right now.

If your system was built around receiving the message before forwarding (or handling) it, then you have to impose SOME size limit on SysEx, and at that moment you are no longer universally compatible.

To Thorstens credit, all SysEx messages are broken into packets no larger than a specific size, so we have a reasonable limit on managing this problem within our software..

As long as those "regular" 3 byte messages get around OK, it works.. And SysEx becomes the less reliable part of MIDI.

I guess my driver writing experience is showing. :)

LyleHaze

Edited February 10, 2010 by lylehaze

[recompas]

sure enough, it was the driver. I tried the mandolane driver and it worked a charm. It was really weird, the sysex was getting received on the computer end - it just couldn't send any out (though midi msgs for notes etc worked fine). Anyway, I'm quite happy it wasn't hardware related as I had triple checked everything. This is probably one of the most well documented and put together kits I have encountered, thanks Wilba!