SYSEX out bug

matti · 10 Sept 2009

AngularMomentum wrote: I have weird behaviour of the sysex out component.

When I connect a string element just before the SYSEX component, it will spit out the wrong DATA...after fiddeling around a few times with disconnecting the string element and reconnecting it sends out the right data...

I am not sure, but I do suspect the MIDI out port being responsible...when disabling the port when the data is correct, and enabling it again gives me again the wrong data. It is not consistently occuring, so it is hard to tell where this is coming from exactly...I can get this result in several ways.

Darn. That sounds bad... I have a couple of sysex-tools that work just like that(haven't done any diagnostic on the output yet tho). I better remove the string primitives just to be sure.

vst-control.de · 2 Jan 2010

The same here! This string goes in the string to sysex module: F04136002320011D00F7 and this string comes out: F01F6F0F3F0F1F5F0FF7 Sometimes it works like it should, sometimes not! Please bugix!

vst-control.de · 6 Jan 2010

I think when an output of a String to Sysex module is not connected or goes nowhere this error occurs (more often). I have made VST-Controllers for the vintage Synths Roland JX-8p, MKS-70, MKS-50 and Alpha Juno. My next Project will be a Controller-Plugin for the Roland Juno 106. This Bug is really annoying because i can work around it a bit but i still don't know why and when it happens and what to do to completely avoid it. I can monitor the String that goes in, but out comes something competely different, like I posted before: in: F04136002320011D00F7 out: F01F6F0F3F0F1F5F0FF7. In the Project for the MKS-50 i have over 50 String to Sysex modules. When the error comes up, sometimes all Modules don't work right, sometimes only one. I get the same behaviour in compiled VST-Plugins. Please help!

rl · 7 Jan 2010

I've build a Sys-ex loader for my K5000 Synth, where everything seems to work well. Maybe you could post a part of your schematic to analyze your problem.

vst-control.de · 7 Jan 2010

I just figured out that the VST-plugins work as it should when you avoid the following:

1. An output of a "String to Sysex" module is not connected or goes nowhere

2. more then one Synthmaker VST-Plugin with "midi out select" component is loaded at the same time. In Cubase 5, a single Plugin works perfect, but when i use two: i get wrong sysex data, Note hangs and crashes. So i can't control my JX-8p and Alpha Juno at the same time 🙁

Would be great if that will be fixed someday

malc · 7 Jan 2010

Thanks for narrowing this down so well. That gives us a good idea of where to look.

sudio · 8 Feb 2010

I have similar problems controling an access virus B via sysex.

sysex in works fine, also with very long messages (e.g. complete controler dump)

string to sysex midi out works *sometimes* i.e the external sysnth recieves a wrong or incomplete message. Unfortunately I cant confirm the points of vst-control.de, seems nearly random when its workig. Seems also not be related to the length of the sysex message, so I think that it is not a speed issue of the midi connection.

Can i do anything to help to find the bug... ?

Iman · 21 Feb 2010

Same Here, also I cannot specify an exact situation that causes "StrSysex" to malfunction it is Totally random. may related to the string input or trigger signal but It has not been clear to me yet.

I send "F0 00 00 0E ... " and I receive "F0 00 00 0E ... " (OK) after some times I receive "F0 01 01 61 61 11 01 01 11 F7" !!! again I try and I receive "F0 00 00 0E ... " but shortly after "F0 01 01 61 61 11 01 01 11 F7" again !

It is really strange because I used it for 1-2 days without any error but today It made me mad.

This bug completely makes "StrSysex" useless please Fix this.

Iman · 26 Feb 2010

Finally I managed to create an OSM that shows the bug. I think the problem is about trigger input of StrSysex. when you connect more than one link to "doit", specially when the second link comes from another module StrSysex will generate wrong data.

sunsynth · 26 Feb 2010

Cannot reproduce that 'bug' here... tried it many times...

MichaelBenjamin · 26 Feb 2010

me neither, but i can imagine that depending on how you connect the triggers that it could mess up the data when stressed in a random way (like the green system randomly ticks). make sure the trigger flow is as uncluttered and definite as possible. but it could also finally be some obscure bug with the primitive.

Iman · 27 Feb 2010

sunsynth wrote: Cannot reproduce that 'bug' here... tried it many times...

Here in my PC with Win7 the bug exists, I recreated the schematic on my laptop with Win XP and I could not reproduce the bug there! so it may related to system resources. In my laptop the bug shows itself in another circumstances, totally random.

Did you Open my OSM or build it yourself? it is important how and when you connect the second trigger input.

sunsynth · 27 Feb 2010

2nd try on Win7:

i can confirm the bug now 2: even if i only click on the trigger button the output value randomly changes from F00C0C0C0C0C0CF7 to F0001102036669F7 🙁

My 1st test was on XP SP2...

MichaelBenjamin · 27 Feb 2010

i also tested on xp.

DigitalWhiteByte · 27 Feb 2010

On my machine, I could not reproduce this problem. 🙁 They always displays the correct values.

Win 7 x64

sunsynth · 27 Feb 2010

Hmm, that's strange or better good 4U 😉

btw: I use Win7 32bit

malc · 26 Mar 2010

I've managed to track this down and a fix will be in 1.1.7 beta8. If you wouldn't mind testing it when it gets released (very soon) that would be great.

The issue turned out to be memory related. Not altogether surprising given the random nature of this. Basically there was a 2 character string buffer being used for each hex element and this needed a null terminator. Depending on the state of the memory around it you could get a zero sometimes, other times you might not be so lucky, hence the randomness.

Thanks for pinning this down to a reproducible schematic. Even though it had me scratching my head without it I'd have been in no man's land.

Iman · 28 Mar 2010

Thx malc

It seems that 1.1.7 Beta8 has sorted the problem in my system (win7 32bit). but it will be great if others confirm this too.

sunsynth · 29 Mar 2010

Cannot reproduce it here either 🙂

So confirmed - seems 2 be working (Win7 x86)