LED messages leaking to other LED

a place to share your patches and get help with the Max/MSP Dev. Kit.
TomSwirly
Posts: 79
Joined: Sun Aug 07, 2011 9:04 am

Re: LED messages leaking to other LED

Postby TomSwirly » Wed Feb 29, 2012 10:30 am

Oh, very interesting!

I'm a little confused, though - let me restate what you're saying to make sure it's right.

1. There's no overall speed limit on MIDI messages to the softstep.
2. However, for LED messages only, if you send them too fast the machine gets confused and "leaks" messages.
3. Sending three extra, empty CC messages after LED messages only fixes this issue.

I wonder if simply waiting a little time after sending LED messages *only* would fix the issue...? Assuming that Max is using MIDI running status, 3 CC messages consume 6 bytes, at about 3K bytes/second that's 2ms.

I'm a little nervous sending out CC 0, because that's "bank select" - if this managed to "get out" and get sent to a hardware synth, it would likely change what sound it was making.


I confess to not having seen this issue in testing - but I've been more focused on the text display.

I'll have some chance to work on this today in Javascript and will report back.
TomSwirly
Posts: 79
Joined: Sun Aug 07, 2011 9:04 am

Re: LED messages leaking to other LED

Postby TomSwirly » Wed Feb 29, 2012 10:32 am

> It sucks that it involves a workaround to get it working consistently, so hopefully they can fix it in firmware (have they been good at addressing issues with firmware? I've only recently got one)).

We did get at least one firmware update.

I'm of the opinion that we on this board can do a better job of writing an interface box to the SoftStep than KMM anyway :-D so as long as we come out with something that handles this, it's more or less invisible to anyone who's using our stuff.
User avatar
Rodrigo
Posts: 25
Joined: Wed Feb 22, 2012 11:49 am
Location: Manchester, England
Contact:

Re: LED messages leaking to other LED

Postby Rodrigo » Wed Feb 29, 2012 11:03 am

The leaking is pretty bad as far as I've encountered. I agree that it would be nice if it just had a mini pause rather than sending a long dummy message, but it could be that that force the firmware to process what it has dealt with whereas sending it 3messages, pausing 2ms, then sending it 3 more might not do it.
TomSwirly
Posts: 79
Joined: Sun Aug 07, 2011 9:04 am

Re: LED messages leaking to other LED

Postby TomSwirly » Wed Feb 29, 2012 12:56 pm

Only one way to find out! :-D

I do agree (as I said in PM) that simply throwing a few extra messages in is a lot simpler in pure Max. But overall I think that a Javascript solution will be the best strategy - partly because it lets us do more complex things like "scrolling messages" that might be quite hard in pure Max, and partly because, if it's mostly or entirely table-based, it should be quite easy to adapt it to run on the QuNeo by simply changing the data.

But "the proof of the pudding is in the eating" - i.e., let's see what I come up with in version 2. I have some time to work on this this week.
Peter Ostry
Posts: 23
Joined: Fri Jul 22, 2011 1:56 pm

Re: LED messages leaking to other LED

Postby Peter Ostry » Wed Feb 29, 2012 6:03 pm

C'mon guys, beeing scared about "long" messages ...
You send a lot more than 12 CCs all the time :-)

Rodrigo wrote:I'm a little nervous sending out CC 0, because that's "bank select" - if this managed to "get out" and get sent to a hardware synth, it would likely change what sound it was making.

First, on most synths CC 0 0 alone does nothing. But anyway, you write programs for MIDI controllers. You are perfectly able to control where your messages go to. Otherwise all the gear around you would constantly go crazy.

However, you are not bound to CC 0. You can send anything that is meaningless for the SoftStep. Send CC 127 if you want. I find it just better to have a simple, easy to identify message like the zeros and – important – to check with KMI if such a message can do any harm.

I got already a response from KMI and they say that the message seems to be ok. But they will ask the firmware developer also. I will post it here after I got the final answer. And then, hopefully, they will never trigger a function with CC 0 0 (if they do, we will append another message).

It would be nice if the LED leaking could be fixed in the firmware but I wouldn't expect this very soon if ever (and if they fix it, we just take our [append] out of the patch).
TomSwirly
Posts: 79
Joined: Sun Aug 07, 2011 9:04 am

Re: LED messages leaking to other LED

Postby TomSwirly » Wed Feb 29, 2012 7:28 pm

> First, on most synths CC 0 0 alone does nothing.

I actually know without looking what my primary synth does with this input - it goes to Trumpet.

> You are perfectly able to control where your messages go to.

Unfortunately, this just isn't the case with Max For Live due to its well-known deficiencies in MIDI. All your MIDI inputs get mixed into channel 1 and all your MIDI outputs get assigned to channel 1, no matter what you set them in your Max program.

Moreover, again due to the numerous deficiencies of Ableton, I often have to drop messages out the IAC buss "at the bottom" and then read them "back in the top" to trigger automation - and this has meant in the past that I had to do clever things to work around messages for different machines coming on the same channel.

It is EXTREMELY unfortunate that Ableton chooses to devote no resources to fixing these major issues, as well as other MIDI issues I've systematically documented, but for the moment I'm stuck with it.

Now, in this patch I'm hoping to entirely bypass the problem by directly going to MIDI using lh_midiout - but you could call me "gunshy" on this matter...

Anyway, this is just chat. The 0 0 0 0 0 0 solution is the best of a bad job. Ideally, KMM would fix the software not to do that, but this works perfectly OK... it just makes me nervous for the reasons I've detailed above.
Peter Ostry
Posts: 23
Joined: Fri Jul 22, 2011 1:56 pm

Re: LED messages leaking to other LED

Postby Peter Ostry » Thu Mar 01, 2012 3:05 am

Peter Ostry wrote:> First, on most synths CC 0 0 alone does nothing.
TomSwirly wrote:I actually know without looking what my primary synth does with this input - it goes to Trumpet.

Interesting, have never had such a device. All I know (not many ...) need CC 0 (xx) and then CC 32 xx to set the program.

I am afraid there is no other number to suggest. I assumed CC 0 would be pretty safe and mentioned CC 127 as a possible replacement but this wouldn't help, it can switch the polyphony. CC 0 127 probably? Or is this a common bank number? Maybe you want to try to send some other message type which is not interpreted by your gear. I believe the Softstep wouldn't care which message he cannot interpret. It's just a mock attack to gain time.


Peter Ostry wrote:> You are perfectly able to control where your messages go to.
TomSwirly wrote:Unfortunately, this just isn't the case with Max For Live due to its well-known deficiencies in MIDI.

I see. So I spoke too fast and apologize for this.


TomSwirly wrote:The 0 0 0 0 0 0 solution is the best of a bad job.

Correct. And the better best of the bad job is the "double dummy" 0 0 0 0 0 0 0 0 0 0 0 0 because it does never allow LED leaking.

----------

Is there a MIDI documentation for the Softstep anywhere?
Or did everybody monitor the data stream to the device?
TomSwirly
Posts: 79
Joined: Sun Aug 07, 2011 9:04 am

Re: LED messages leaking to other LED

Postby TomSwirly » Thu Mar 01, 2012 8:45 am

Oh, pshaw, no need to apologize, we need to be kept honest.

And the case could easily made that I'm a bit stupid for continuing to use Max For Live. :-D

(And come to think of it, I don't remember whether my synth really does banks on 0 or 32. However, I remember for a fact that I only need one CC for it, not two.)

To the best of my knowledge, there's no data sheet for the unit. The "official" mechanism of communicating with it is either through one of the official editors, or through the ss Max patch.

I made extremely good progress last night with the Javascript program, however, and will almost certainly have something to release today.
Peter Ostry
Posts: 23
Joined: Fri Jul 22, 2011 1:56 pm

Re: LED messages leaking to other LED

Postby Peter Ostry » Tue Mar 06, 2012 3:27 am

Finally we got a "go" from KMI support and the firmware developer:

    "The dummy messages should be safe to use. You're not going to do
    anything bad by sending those 0's. They're not doing anything special."

Therefore my suggested solution to avoid leaking of a LED message to the next LED while sending CCs is to insert a "double dummy message" after each 3-element LED message:

    [append 0 0 0 0 0 0 0 0 0 0 0 0]
TomSwirly
Posts: 79
Joined: Sun Aug 07, 2011 9:04 am

Re: LED messages leaking to other LED

Postby TomSwirly » Tue Mar 06, 2012 12:20 pm

I am also using this strategy in the JS driver (http://ax.to/softstep).

I'm still not 100% happy with this as a resolution - it begs the question of whether, for example, you can get other confusions by sending other messages too fast. I'd really like to know exactly what one can and cannot do with the unit - there has always been the implied possibility held out by KMM that you can render the pedal useless by, for example, sending it too many messages too fast (though as far as I know no one has ever done that, and I did some pretty serious testing early on before we got that warning).

But it works, and it's easy enough.

Return to “SoftStep Max/MSP Dev Kit forum”

Who is online

Users browsing this forum: No registered users and 1 guest

cron