Jump to content

TK.

Administrators
  • Posts

    15,198
  • Joined

Posts posted by TK.

  1. You are using a STM32F4! Of course, this changes the situation.

     

    The idle value is ok, same at my side with this core.

    Query is working at my side.

     

    Maybe the only difference: I'm using the latest bootloader

    Could you please update to the latest version (V1.018) as well?

     

    Since the query function doesn't work during runtime, you've to press&hold the blue button on the discovery board.

    Re-connect the USB cable

    Keep the blue button pressed!

    WinXP should install the USB MIDI driver for the bootloader (if this hasn't been already done)

    Restart MIOS Studio, select the right USB MIDI interface and upload the bootloader update application.

    All this by keeping the blue button pressed (to prevent that the MBSEQ app will be started)

     

    Release the blue button during the hex upload - the bootloader app will be started thereafter.

     

    Once the bootloader has been updated, upload the MBSEQ app

     

    Best Regards, Thorsten.

  2. It also crashes at my side.

    A quick analysis of the pattern data shows me that it's corrupted - I can't say why.

    Do you remember the moment when you stored the A7 pattern? Was there something special? (e.g. power interruption or similar?)

    I don't see an easy way to restore the pattern.

     

    In order to avoid the hang, you've to overwrite 2:A7

    E.g. after power-on select group 2, press MENU+SAVE, select 2:A7, press the SAVE button and confirm the name.

     

    Best Regards, Thorsten.

  3. Actually a great idea, so that I couldn't resist to hack this into the firmware:

    http://www.ucapps.de/mios32/midibox_seq_v4_088_pre9.zip

     

    (Copy/Paste are available at GP15/GP16, and can also be executed with the dedicated buttons)

     

    But: accent won't be copied as the user might expect it.

    It will work if the track has enabled the accent trigger layer, if not the accent flags will get lost.

     

    The euclid generator doesn't get use of the accent flags, but directly sets the note velocity values instead to simplify the editing of a generated pattern.

    The "Live Pattern" copy/paste function would need a certain intelligence to determine this special situation, but it's too error prone...

     

    So: the copy&paste approach will only work reliable with gates.

     

    Best Regards, Thorsten.

  4. Here the changes: http://svnmios.midibox.org/comp.php?repname=svn.mios&compare%5B%5D=%2Ftrunk%2Fapps%2Fprocessing%2Fmidibox_cv%2F%401033&compare%5B%5D=%2Ftrunk%2Fapps%2Fprocessing%2Fmidibox_cv%2F%40830

     

    A V1.2c doesn't exist, and the changes between V1.2b to V1.3 don't explain the issue that you noticed.

     

    Could it be that you used a customized version, and therefore also changed the version number?

    To which pins are CLK and Start/Stop connected?

    According to this file: http://svnmios.midibox.org/filedetails.php?repname=svn.mios&path=%2Ftrunk%2Fapps%2Fprocessing%2Fmidibox_cv%2Fsetup_j5_enabled.asm

    it's RC0 and RC1

     

    Best Regards, Thorsten.

  5. This one should work: http://www.reichelt.de/AK-450/3/index.html?&ACTION=3&LA=446&ARTICLE=4012&artnr=AK+450&SEARCH=AK+450

     

    Btw.: is it only me, or do you guys also have the impression that prices increased in the last weeks?

    E.g. my favorite turkish fast food restaurant recently increased the price for a Doner Kebab from 3.50 EUR to 4.50 EUR

    And now I notice that Reichelt increased the price for this cable from 3.50 EUR to 4.50 EUR as well! ;-)

     

    Best Regards, Thorsten.

  6. Here another mockup:

    http://www.ucapps.de/mios32/midibox_seq_v4_088_pre8.zip

     

    Instead of the repeat raster, a pattern can be selected.

    In the same page, the pattern can be edited by pressing (& holding) the SELECT button + the GP buttons (or encoders)

     

    I got rid of the idea to provide a "delay" trigger, because it conflicts with the already available groove function.

    So: it's better to use the groove function independent from the pattern selection.

     

    Instead, only gate and accent are editable.

     

    The "copy track to pattern" function isn't available yet, on the other hand it could complicate the handling too much.

    Of course, it would be nice to use it for taking over existing patterns, or transfer generated euclid patterns...

    On the other hand: where to place such a copy function? And is it really worth?

    I mean: a 16 gate/accent pattern is easy to remember, just enter it again with the pattern editor - something what don't has to be done so often, because I guess that once somebody found well working patterns, he won't edit them anymore.

     

    Best Regards, Thorsten.

  7. With the pattern based approach a step offset wouldn't be required anymore (one reason why I find it useful :)

    The Euclidean generator could be used to generate a pattern.

     

    Usage could be the following way: create a pattern in EDIT page as usual (this allows you to use the euclid generator as well)

    Once you are happy with the pattern, transfer it into an ARP pattern slot and play it with the keyboard.

     

    Best Regards, Thorsten.

  8. As mentioned by flyweight303, the feature would be more powerful if it would work like an arpeggiator.

    What I've in mind are 16 or 32 soft-configurable patterns with:

    - gates

    - accent

    - delay (for swing)

     

    how many steps would be required for such patterns? 8 or 16 or even 32?

    Memory consumption would be 3*(steps/8) bytes per pattern - no problem for STM32F4, could work with STM32F1 as well, but problematic for LPC17 (since RAM is almost fully allocated)

     

    But: for LPC17 I could at least provide pre-configured patterns (stored in flash)

     

    Best Regards, Thorsten.

  9. The MIDI message shouldn't be sent immediately (it could be an option in future, but I added added some code which prevents this).

     

    Check your MIDI router and bus settings - only a single router node & MIDI bus should forward MIDI events.

     

    Best Regards, Thorsten.

  10. Something overloads the CPU, this is indicated by the "system" command which shows 216% in your first snapshot.

    Actually it should be around 50% on a STM32F1 core (when no sequence is running) and around 60%..80% when the sequencer is running.

     

    I need more input: you've to find out, which configuration causes the overload - I can't help you to find this remotely...

    I know that there are many many many mode & parameters which could influence this (users requested so many features ;-)) - but at least I can say that you are using some kind of configuration which I haven't selected by myself in the past (I'm using a STM32F1 as well)

     

    It could make sense to start with a new session.

    Just enter "new <name-with-8-characters"> to create a new one

    If this already helps, then please attach your previous session directory to this posting.

     

    Best Regards, Thorsten.

  11. Ok, here a quick & dirty implementation as a starting point, and to identify potential issues:

    http://www.ucapps.de/mios32/midibox_seq_v4_088_pre7.zip

     

    Usage: enter the Utility->Live page, then press GP8 to enter the (hidden) "Repeat Config" page.

    On a common track you will see a "Repeat" on/off switch and the repeated note length at the left side, and 8 repeat configurations at the right side.

    On a drum track you will be able to select the drum instrument in addition at the left side, means: each drum instrument can get it's own configuration.

     

    Repeat configurations: 1th, 2th, 4th, 6th, 8th, 16th, 24th, 32th note

     

    What is already working:

    - once enabled, notes won't be played immediately, but will be repeated according to the selected repeat raster

    - in drum mode, an individual raster can be specified for each instrument. Notes which are not defined as an instrument in the drum track will be passed immediately and won't be repeated

    - channel aftertouch controls the velocity of all repeated notes

    - polyphonic aftertouch allows to control the velocity of the corresponding note

     

    What is not working & known limitations:

    - no support for recording yet

    - no overdubbing takes place yet. Which means: if the sequencer step plays the same note, it will be sent twice

    - probably won't work properly if sequencer plays glided or sustained notes

    - triplets and dotted notes can't be specified directly, because the repeated note has to match with the track raster. Means: if the track is configured for triplet modes, you will get triplet notes for all notes played by the track. There is no solution to overcome this limitation

    - due to the same reason, 32th notes will only be played if the clock divider is configured for 32th notes in the "track divider" config page

    - due to the same reason, 6th and 24th notes result into funny patterns - I left it as it is, because it results into an euclidean-like effect (try it ;-)

    - configuration won't be stored on SD Card (because it's a moving target...)

    - the same configuration is used for all tracks

     

     

    There is no way to overcome the "track raster" limitation due to conceptional reasons.

    However, we can also see this as an advantage: euclidean-like patterns can be created this way ;-)

    A possible extension would be to add a note offset, e.g. to shift certain instruments into the off-beat

     

    Best Regards, Thorsten.

  12. The SID chip is clocked at 1 MHz, register write accesses have to be synchronized to this clock to prevent data loss, so that typically 2..3 clock cycles will be consumed to ensure that the write operation took place.

     

    During this time a PIC can only execute 30 instructions.

    A LPC1769 would execute more than 300 instructions during this time, and it shouldn't do anything else (or it has to service the write operations with a high-prio timer, which is another story).

     

    This results into a significant performance loss, and therefore the direct access doesn't make much sense.

     

    Best Regards, Thorsten.

×
×
  • Create New...