Micro not getting all MIDI data

integer

Hi, I have a Meeblip Micro 2.04. I have two problems.

a) It seems that the sound I get as default sound is some kind of noise. This is not so important, but I was expecting more of a "melodic bass" type of sound.

b) More annoying is that my Micro seems to miss some of the Midi data now and then. I made a loop in Ableton Live playing a 32nd note every 1/8 bar at 120 BPM. It seems that randomly, maybe every 100-200 notes something weird happens. It seems like the Micro doesn't get the Midi on or Midi off, resulting in either silence or a long note. Any ideas about the reason for this?

integer

Just to be sure, I just ran a test of the same stuff with my nowadays-not-used-for-generating-sound master keyboard. I find it also seems to be dropping some of the notes, so maybe the Midi interface is to blame...

meeblip

a) That's not right. I think I know what the problem is, and we'll have a fix in the 2.05 firmware.

b) Make sure the MIDI interface/board isn't sending out MIDI clock or MTC. I was doing some pretty extreme stuff with Live last week using a Meeblip SE and it was able to keep up. I might add a 256 byte circular buffer to the input in the near future (a precursor to adding USB) which should be able to handle some pretty extreme MIDI abuse.

integer

It seems my previous post disappeared. All is now well with b) after investing in a new MIDI interface.

As long as you're working on fixes: Sometimes when playing a note, sending a CC corresponding to a switch doesn't change the "logical" switch status within the Micro. It seems to be easiest to provoke when repeatedly playing the same note as fast as possible.

harry

Hi, I thought I'd mention that I was also having the same two exact problems. Unfortunately, I can't replace my MIDI keyboard controller, as I am building a custom case for the keyboard and synth.

A MIDI buffer/filter would be ideal, as the clock messages from the keyboard (which can't be turned off) seem to confuse the Micro, as it drops lots of notes and exhibits very strange behavior. When the MIDI is run through a filtering program on my laptop, all is well.

meeblip

@Mattias: Interesting switch problem - will see if I can duplicate it.

@Harry: Interesting. I'll give this issue some thought. My first inclination is to create a circular buffer that only captures MIDI data that the synth actually uses while dropping everything else.

integer

James: It gets weirder. This short midi clip switches noise on and off three times while playing notes. In Ableton Live, only the two first (short) noise on's have effect on my MeeBlip, but the last (longer) noise on never gets through. This is repeatable.
However, I also tried it on an old Cubase 3, and there all three of the noise on's are heard. I don't know if any of this helps. :-)
http://www.tesuji.se/tmp/meeblipCCtest.mid

meeblip

Hang on a second. :) I think I understand what's going on. The switch panel is only scanned once every 100ms, while the knobs are scanned as quickly as the analog-to-digital conversions take place.

Much as I'd like to be able to scan the switches to allow them to be played musically, there simply aren't enough clock cycles between samples to allow it (running at 16MHz, a sample has to be generated every 440 instruction cycles).

goosejackets

Cheap New-style Louboutin

Discount New-style Louboutin

Cheap New-style Christian Louboutin

Discount New-style Christian Louboutin

Cheap Louboutin Boots

.

* Via the internet Specialty Retail stores *

There were internet outlets are irresponsible, nearly always are 100 % reputable.

Especially as you visit the particular authorized webpage for Ugg boots, you can guarantee that your obtain is reliable.

Coming up from Modern australia, the shoes and boots and accessories in the main Ugg website may perhaps be your finest bet when you're looking for those rare to find styles regarding boots, athletic shoes, or products.

Many internet-based promote sites assure super lower prices for Ugg boots.However, the genuine products available for sale in many of those auctions are in reality lookalikes as opposed to the real McCoy.

As well as the footwear may are similar to authentic Uggs, the engineering and supplies are of less quality ?n comparison to the typically exceptional uggs

integer

OK, so let me first tell you about what I'm trying to achieve. I'm building a simple Meeblip parameter editor in MaxForLive, which just passes incoming midi notes through, and adds control change messages for the parameters. So, there are neither physical knobs nor switches attached to my Micro.

I was looking into the code yesterday, and although I have no experience with Atmel assembler, it seems to me that you treat incoming CC the same way, regardless of if they represent switches or knobs. The CC for switches should get written directly into the performance registers, so I'm not sure the periodic switch scanning applies to my problem. But I'm fully aware that I might have overlooked something in my browsing of the code. :-)

peter

You could also kick it with a pair of Ugg boots (sorry, very, very close to our forum migration out of this mess...)

peter

No, yes, you're correct that switches and knobs are handled both by CCs - this is for flexibility, for instance, if someone has a keyboard with a bunch of knobs and want to use those knobs to set the switch values. This shouldn't have anything to do with switch scanning issues, though. James... more thoughts?

meeblip

Yes, the MIDI value is dropped into the patch in the same manner for MIDI knobs and switches. There's also a mechanism for writing changed physical switches into the patch.

It's possible that the PATCH_SWITCH1 and PATCH_SWITCH2 values are getting overwritten by something from the physical switch scan. Will do a code walk to check - it's been a few months since I wrote that stuff.

integer

Well, the problem for me is that I would like Live to play some Midi data while I tweak the CC-knobs and CC-switches of my editor, and then I might have to press a virtual switch three or five times for it to have effect if the first time is not recognized. It's a bit hard to reconstruct the signal timing, as I would probably need a logic analyzer for this.

I have only found it to be a problem on the CC-switches, but then on the other hand, turning the CC-knobs would send a stream of CC messages which might take care of the problem automatically. Maybe I should try to provoke the problem with knobs jumping from one discrete state to another as well... Or add some kind of message queueing in my M4L patch as a workaround.

peter

Hmm... no, you shouldn't have to do that. Let me recreate this setup tomorrow and see if I can reproduce your issue.

scarkord

I've just assembled my meeblip micro and think I'm also getting the problem with the default sound as per Mattias' original post. Can't be sure, but it sounds like its defaulting to whitenoise with a highpass filter.

meeblip

Jason: I've tracked this down to an eeprom corruption issue when programming. Some Atmega32A batches seem to be more sensitive to programmer speed than others when pushing the default patch. I'll send you an updated 2.05 chip, which also allows you to overwrite the default patch.

In the meantime, you can send MIDI CC to flip the switch value.

harry

Would it be at all possible for me to get an updated chip as well? I have the same white noise issue - as a temporary solution I have just made up a few MIDI files which send the CC information to get a usable start up sound.

meeblip

Harry - Sure, send me a PM with your contact details. It's a great chance to get latest update, too - has improved MIDI handling (specifically, it now drops Active Sensing and MIDI ticks as quickly as possible).

Also, please let me know what the batch number on your Atmega32A chip is.

harry

Sorry, I forgot to mention in the PM, the batch number is v2.04

scarkord

That sound's great James! Many thanks

Howdy, Stranger!

Categories

In this Discussion