blue-2:
	BUG FIXES:
	* Finally fix the bug that the G8 LEDs freeze whenever the LED gauge
	  of the [pot] circuit is visible (thanks to the @klngvrhltnss's
	  reluctance to give up on this)
	* [euklid]: Fix shifted offset after reset (found by @kikegas)

magenta-17:
	CHANGES:
	* [motorfader], [faderbank]: New input "resetvalue". "startvalue" still
	  exists. Splitting up this allows user interfaces with some kind of
	  reset or set button without loosing the storage to the SD card.
	* [pot]: resetvalue correctly honors discrete now.

magenta-16:
	CHANGES:
	* [pot]: Replace input "startvalue" with "resetvalue".
	  It does basically the same: set the pot's value after a reset. But
	  now it does not affect the storage of the pot's last value on
	  SD card. That also fixes a bug where the pot's state was not
	  correctly saved.

	BUG FIXES:
	* [pot]: Fix bug where the virtual value of the pot -
	  if selected at boot time - slightly moves after each patch reload
	  (via button or MIDI). Found by @klngvrhltnss.
	* [button]: Possible fix weird behaviour for a multi state
	  button if "states" is not explicitely set.
	* [clocktool]: Fix hickups for negative delays. -0.1
	  is now correctly something different as 0.9. Also now correctly
	  allow delays > 1 and < -1.

magenta-15:
	BUG FIXES:
	* [button] (and probably others): State was not restored
	  in situations where preset was *not* patched.
	* [button]: State of presets was not restored.
	* [motorfader]: Fix restoring of state from SD card.
	* [pot]: Improve delay for selecting the first or
	  last entry when using descrete (found by @jpizzo)
	* [pot]: Fixed presets that now also work without select
	  (found by @jpizzo, @Mario Paler and @oddharmonics)
	* [pot]: Fixed virtual pot logic when just startvalue
	  is used but neither preset nor select
	* [pot]: Slope now works again

	NEW FEATURES:
	* [potbank]: Removed potbank again, for the while. Didn't
	  seem to be useful while wasting flash space, my time
	  and still turned out to have some bugs.

magenta-14:
	NEW FEATURES:
	* All circuits need less memory now. Precisely spoken, each
	  existing input jack (even if not patched) now needs
	  four bytes less.
	* The double press on the master's button now creates
	  multiple state files, like STATES1.TXT, STATES2.TXT
	  and so on.
	* [potbank]: New circuit for managing eight pots in once,
	  saving lots of memory if you have many similar pots.
	  (as a gift for @jpizzo in order to make his awesome
	  Supergrids patch get even more awesome)
	* [pot]: New discrete mode for selecting an integer
	  value, e.g. for preset numbers, sequence lengths
	  etc. Including special visual LED gauge mode for that.
	  Simply set discrete = 8 for example (based on an idea
	  by @alienlebarge)

	BUG FIXES:
	* [midifileplayer]: makes now sure that there is at least a tiny
	  fraction of time between a note off and the next note on,
	  even if they happen at the exact time in the MIDI file.
	  That way a subsequent [midiout] circuit will now see the
	  new gate impulse, which was formerly lost. Note: this still
	  might not be enough to trigger an external analog envelope, but
	  that's than at least not my problem ;-)
	* [contour]: Disallow negative velocity with unpredictable behaviour.
	  The velocity is now limited to the range 0.0 ... 1.0
	* [pot], [algoquencer], [button], [buttongroup]: current state
	  was not correctly saved when the jack "loadpreset" was used.

magenta-13:
	NEW FEATURES:
	* [droid]:
	     - The [droid] circuit now has a new parameter called
		   m4faderspeed, which sets the power / speed of the motor
		   faders of your M4s.
	     - The [droid] circuit now has a new parameter called
		   m4notchpower. That allows you to tune the force feedback
		   of the virtual notches in the M4s.
	* [motoquencer]:
	    - New color and blink codes for gates that are randomized
		  (see manual)
	    - New input "defaultgate", which sets the
	      default gate statet when you clear the sequence
	    - New inputs for live recording into the sequencer with a
		  keyboard (cv/gate, but can be routed from midiin, of course)
	    - new inputs copy, copymode and paste, pastefaders,
	      pastegates for copying and pasting steps or step parameters between
	      pages or presets.
	    - New trigger luckygatepattern. And that is now
	      correctly used for randomizing when luckyfaders or luckybuttons
	      is triggered in the according fadermode or buttonmode (it was
	      luckyties)
	    - New parameter "cvnotches" for using the step CVs for setting
		  integer values (e.g. for switching or selecting something).
	* [motorfader]: New input "reset"
	* [pot]: Now supports 16 presets
	* [button]: Now supports 16 presets
	* [buttongroup]: Now supports 16 presets
	* [matrixmixer]: Now supports 16 presets
	* [faderbank]: New inputs "startvalue" and "reset"
	* [droid]: New trigger input "calibrate" (mainly used in the DMMDM
	  labs for initial calibration during the test procedure)
	* [polytool]: New circuit for changing the polyphony of voices.
	  You can either map one single CV/gate input to multiple
	  CV/gate outputs or vice versa You can even map three inputs
	  to two outputs if you like.

	BUG FIXES:
	* GENERAL: LED registers R17 ... R32 did not work while an
	  LED gauge was visible (from pot circuit). This has been fixed.
	* [motoquencer]: Fix stuck end step when you dynamically reduce
	  numsteps. Also disable unused faders (those for steps that
	  go beyong numsteps). Found by @ImaginedWorld.
	* [algoquencer]: Fix gate length of rolls. Previously "gatelength"
	  was interpreted as seconds, but just for the rolls. Correct is
	  that it is a fraction of the clock cycle. For rolls it's now
	  a fraction of the clock cycle divided by the number of rolls. That
	  way the gate is always audible even if the gatelength is close to
	  1.0.
	* [arpeggio]: Fix startnote: the note numbers where garbled. Now
	  2 really plays the 2nd/9th (was the 3rd). 3 plays the 3rd (was
	  the 5th) and so on.
	* [arpeggio]: Fix startnote(2): The start note now exactly replaces
	  the first note that would be been played, not the first two
	  or similar.
	* [chords]: Make selectfill1 ... selectfill5 work. They were simply
	  ignore, which is a shame, since out-of-scale chords can sound
	  very interesting (found by @effiksmusk)!
	* [chords]: Fix wrong note output when just one output is used
	  (e.g. C# instead of D, found by @Klngvrhltnss)
	* [midifileplayer]: Works again! It was totally broken due to
	  the introduction of the miniin-circuit in blue-1. Sorry for the
	  mess.
	* [midiout]:
	  - Fixed pitch tracking (avoid bogus additional note on
		and note off event, which lead to stucking notes on Logic and
		maybe other DAWs).
	  - Fixed missing initial update of CC number 0
	  - Delay initial CC updates by one second (configurable). That
	    avoids missing the initial CC states after a Droid cold start
		(found by @tmp).

	M4 Firmware orange-3:
	* Fix: Increase sensitivity of touch buttons.
	* Fix: Add histeresis in fader notch force feedback (reduces jitter)
	* Fix: Fix bug in fader stuck detection, which led to too high
	  fader speeds and overshooting (and probably jitter as well)
	* Make fader speed configurable (see droid circuit)
	* Make notch power configurable (see droid circuit)
	* LEDs below the faders light purple whenever the motor is
	  operating. That helps to detect "stuck" faders, where the
	  motor tries to pull but does not succeed.

	NOTE: You need to update *both* the master and the M4s. Otherwise
	they won't be able to communicate!

magenta-12:
	NEW FEATURES:
	* GENERAL: allow to load empty patches (might seem useless, but removes
	  some edge cases and allows you to "clear" your master)

	BUG FIXES:
	* Fix: Fix bogus error message about free RAM when you have too many
	  circuits. The numbers in the error message are now precise.
	* Fix: Fix stickyness of ledvalue and ledcolor in motorfader and faderbank.
	  The LED colors would stick even if the circuit is not selected anymore.

	M4 Firmware orange-2:
	* Fix: Fix saving of internal counters

magenta-9:
    MANUAL:
	* More information about the M4
	* Added chapter about the fadermatrix

	BUG FIXES:
	* GENERAL: avoid certain problems at start time no controllers haven't
	  delivered values yet
	* [motoquencer]: transpose works now (was simply ignored)
	* [motoquencer]: cvbase and cvrange now work as expected
	* [motoquencer]: Limit start and end step to the number of steps
	* [motoquencer]: Gate probability setting 5 (odd turns) would
	  never play. works as expected now.
	* [motoquencer]: Reset previous possible start/end change if
	  buttonmode (now) is not even patched
	* [motoquencer]: Do not apply random CV in composition mode
	* [motoquencer]: trigger to clear does not clear the other presets anymore
	* [motoquencer]: avoid overwriting preset 0 in certain conditions
	* [motoquencer]: the end step could be garbled during power cycles. That led
	  to incomplete play of the sequence.
	* [motoquencer]: last edit was not saved when working with instant presets
	* [motoquencer]: changes made by "I Feel Lucky" are now correctly saved
	* [motoquencer]: use true random number for randomization, so you get
	  real new random decision every time
	* [fadermatrix]: renamed input "fader" to "firstfader" in order to consistent
	  with faderbank and motoquencer

	NEW FEATURES:
	* [motoquencer]: New outputs currentpage and currentstep

	M4 Firmware orange-1:
	* Fix: Avoid data loss when there are two notches (like fadermode = 6
	  or 7). Changes of the faders would not be detected always, so
	  data was lost or carried over to other pages. Should be fixed now.
	* Fix: Avoid data injection from one situation into another when
	  using notches = 1
	* Fix: Get rid of minimal twitch when switching faders and the
	  new situation has the same value as the old one.
	* Fix: Avoid potential jitter of fader

magenta-8:
	BUG FIXES:
	* GENERAL: Fix factory reset: it wasn't deleting the settings
	  of the individual circuits.

	* [motoquencer]: Fix: Presets work now
	* [motoquencer] + [algoquencer]: Fix loading preset 0 at the
	  beginning if preset is used without triggers. Also fix
	  loosing changes in current working set if presets have
	  not been switched.
	* [motoquencer]: Some new musical features of the ``gate propability''.
	  please refer to the manual!

	M4 Firmware cyan-4:
	* Fixed non-notched fader mode (was completely broken, sorry!)


magenta-7:
	NEW FEATURES:
	* [buttongroup]: New outputs outputs1 ... output32. They output 1
	  (or valueX) when the button is active and 0 otherwise. This also
	  works when the group is deselected (the led outputs don't work
	  then).
	* [motoquencer]: If buttonmode is invalid, make LEDs black
	  (just the current step is white as always)
	* [motoquencer]: slight color adaption for tied notes (violet)

    BUG FIXES
	* [motoquencer]: reset was inhibited by The Metric Saver™. That
	  has been fixed.

	M4 Firmware cyan-3:
	* Touch buttons keep working in "low power" mode.
	* More force feedback in notched mode, notches feel better
	* More force feedback in bend wheel mode (notches = 1)
	* Fix rare fader vibrations in notched mode (hopefully)
	* Fix spontanous modification for virtual fader values during
	  power shortage.
	* Show LED animation with red-yellow-green-blue when link
	  to master is not working or M4 is disconnected from master.
	* Improve LED PWM (smoother color grading)
	* Fix behaviour when number of notches changes
	* Prevent bogus ghost touch presses right after module starts.



magenta-5:
	NEW FEATURES:

	* GENERAL: The maximum number of controllers has been raised from 16
      to 32. You can attach 32 controllers as long as you don't exceed
	  the maximum total allowed power. We will give exact numbers of
	  the power demands of each controller later.
	* [togglebutton]: Remove "switch" input, remove reset2..4 and
	  output2..4. Please use the new [button] circuit. That supports
	  overlaying (even better, because it is not limited to 4 overlays)
	* [matrixmixer]: New inputs "reset" and "startvalue". reset brings
	  the matrix into its initial state. startvalue defines what that
	  state is: 0 = all buttons off, 1 = buttons on the diagonal on,
	  2 = all buttons on. Using startvalue disables persistence on the
	  SD card.
	* [algoquencer]: New trigger input "clear", which unsets all buttons
	  from the current page (steps, alternate, accents).
	* [buttongroup]: Supports 32 buttons now (was 16)
	* [effiquencer]: Reactivated Superbooth prototype of Motorsequencer
	  under this new name. This circuit will be removed soom, nevertheless.
	  It will be replaced by the new circuit "motoquencer".
	* [motoquencer]: New take on motor fader sequencer.


	BUG FIXES:
	* [buttongroup]: Fixed saving of state (that bug also could
	  have garbled the state of other circuits, as well)
	* Fix nasty bug that if the controllers attached to your master
	  didn't reflect the active patch, it would try e.g. to
	  load droid28.ini. This has been fixed. Now if "droid28.ini"
	  or the like does not exist, it falls back to droid.ini


blue-1:
    NOTE: blue-1 needs a bootloader update before you can install
	      it. This is performed with green-8. Read the manual for
		  details. If you come from green < 8, please first update
		  your master to green-8 as an intermediate step.

	NEW CIRCUITS:

	* [pot]: New helper circuit for pots. It combines the functions of both
	  switchedpot and notchedpot and is consistent with the new
	  select/selectat method for overlaying things. That way it also
	  isn't anymore more restricted to 8 virtual pots. Also the pot
	  has a nice LED gauge for showing it's virtual value and some
	  other new cool features.
	* [button]: Same here: New helper circuit for buttons. It combines
	  the functions of both togglebutton and fourstatebutton and also
	  implements the new select/selectat method. Also it has a detection
	  of long button presses.
	* [timing]: New circuit for flexible beat timings. Shuffle/swing is
	  the easiest task. Move up to eight successive beats slightly back
	  and forth in time to create interesting groove patterns.
	* [midiin]: Receive MIDI streams via the new X7 expander via USB
	  or TRS/DIN. With all the jacks of midifileplayer and some more.
	* [midiout]: Send MIDI streams via the new X7 expander via USB
	  or TRS/DIN.
	* [midithrough]: Route MIDI data through the X7, while splicing
	  in our out your own MIDI events

	NEW FEATURES:

	* GENERAL: Support for the new X7 expander. The X7 gives you
	   - MIDI in *and* out via DIN/TRS (3.5mm jack) and USB-C
	   - Access to your SD card via USB (X7 behaves like an USB stick)
	     without any additional software
	   - Alternatively super fast patch upload via MIDI Sysex
	   - Four gate outputs (G9, G10, G11 and G12)
	   - Most complete MIDI implementation found in Euroland, both
	     for MIDI in and out.
	  The name "X7" comes from: 1 USB jack, 2 MIDI jacks, 4 Gate jacks
	  makes 7 jacks. And U1M2G4 was simply to long and clumpsy. Hence
	  X7 ;-)
	* GENERAL: Support for the new B32 controller. The B32 is an array
	  of 4 by 8 buttons, just like those on the P2B8.
	* GENERAL: Support for the new P4B2 controller. The P4B2 is similar
	  to the P2B8, but has 4 large pots and 2 buttons.
	* GENERAL: Support for the new S10 controller. The S10 has two
	  rotary switches with 8 position each and 8 toggle switches with
	  two positions each.
	* GENERAL: Remove the upper limit on the number of constants. You can
	  now have as many as you like. Constants are fixed numbers in your
	  patch like 1.0 or 17.8.
	* GENERAL: A double click (fast double press) on the master's button
	  now creates a file names "STATES.TXT" on your SD card. There your
	  find the current values of every input, output, pot and other
	  interesting stuff.
	* [midifileplayer]:
	   In general the midifileplayer has greatly benefited from the
	   new midiin circuit, as both share the same MIDI implementation.
	   That means lots of new jacks and access to MIDI features that
	   hadn't been accessible, some of which are:
	   - Can now output up to four arbitrary additional
	     MIDI CCs (in addition to modwheel (CC#1) and volume (CC#7)).
	   - Can now handle MIDI program and bank changes.
	   - Add jack "tuningpitch" for setting the tuning pitch.
	     This is now consistent with "minifonion" and other circuits.
	   - Support for channel pressure and polyphonic key
	     pressure (both aka as "after-touch"). See the new "pressure"
	     outputs.
	   - New inputs "lowestnote" and "highestnote" for defining a
	     range out of which no notes are being played ("keyboard split")
	   - New inputs and outputs for getting the individual gates of
	     up to sixteen specified fixed notes, called "notegates"
		 (e.g. for drum sequencing)
	   - NOTE: "dropnotes" has been renamed to "voiceallocation".
	     Please check your patches and update them if you have used
		 that parameter.
	* [buttongroup]:
	   - Increase the number of buttons from 8 to 16
	   - The inputs value1 through value16 have now the useful default
	     values 0, 1, ... and 15. So in many cases there is no need
		 anymore to use them.
	   - Allow maxactive = 0. This allows to create an empty "ghost
	     button group", that - in combination with select - can be
		 used to make sure all LEDs are off when you work with nested
		 selectable buttongroups and similar contraptions.
	* [nudge]: support select and selectat
	* [calibrator]: support select and selectat
	* [notebuttons]: support select and selectat
	* [matrixmixer]: support select and selectat
	* [euklid]: New input "outputsignal" that makes euklid
	  useful as a stepped modulation CV. Brings in kind of switch
	  capabilities.
	* [fold]: Remove default bounds of 0 and 0.1. Now leaving
	  minimum or maximum unpatched leaves that boundary
	  "inactive". That's more helpful and more intuitive.
	* [compare]: New input "else". Makes the whole thing more
	  intuitive. Also the asymatric behaviour of less has been
	  removed.

	BUG FIXES:

	* GENERAL: voltages near zero V (< 0.5mV) will not light the input
	  LEDs anymore. Such super small voltages are not uncommon and are
	  usually practically zero, so the LEDs of input1 ... 8 are silent
	  now.
	* GENERAL: remove circuit order dependency when using L registers
	  as an input. They were all 0 at the beginning of the patch and
	  not at the value that had been set last. Thanks to @Joscha for
	  finding this bug!
	* GENERAL: Fix wrong line number indication for unused patch
	  cables if the name of one cable is the prefix of the
	  name of another one.
	* [algoquencer]:
	    - If you use both lengthbutton and length, length
	      should set the default length when the circuit is used for the
	      first time. This didn't work and has been fixed now.
	    - Now applies "repeats" correctly to the
	      "barleds" and to the alternative pattern. Note: if you
	      want the old behaviour for the alternative pattern back
	      (which can make sense), set the new parameter
	      "alternaterepeats = 1".
	* [slew]: Fix freeze that @rss encountered.
	* [crossfader]: This mixing distribution between the two
	  neibouring outputs was reversed. While you could compensate
	  that in a setup with just two inputs by inverting the
	  fade input, with three or more input signals there
	  where hard cuts. This has been fixed (found by @klngvrhltnss).
	* [chord]: Make two and three voiced chords work as expected
	* [clocktool]: At the very start of the DROID and divide = 2,
	  let the first clock through, not the second. Also when
	  just using trigger and gatelength, the very first clock
	  output was missing after DROID start. This has been fixed.


green-8:
	New features:
	* GENERAL: The maximum number of constants has been raised from
	  32 to 40.
	* GENERAL: increased maximum droid.ini size from 20000 to 64000 bytes
	  (spaces and comments are not counted!)
	* GENERAL: Added bootloader update function in maintainance menu
	  (this will be needed for updateing to the upcoming blue firmwares)
	* GENERAL: Internal restructuration of the flash memory in order
	  to make more space for future firmwares
	* GENERAL: Experimental support for the new M4 motor fader controller
	* GENERAL: Experimental support for the new X7 interface/gate module
	* GENERAL: Fix computation of taptempo inputs (affects all circuits
	  with a taptempo input). This didn't work as expected. Now you can
	  either use a steady input clock as a reference or a manual tap.
	* GENERAL: Input LEDs: display values closer than 1 mV to zero as
	  black. This avoids low blue flickering if you patch a level of
	  practically zero volt to and input.
	* GENERAL: Fix spurios "all white blinking" error and freeze afterwards
	  when reloading the DROID patch.
	* GENERAL: Fix "all LEDs blink yellow" error when trying to load
	  a patch from the SD card sometimes. This happened when you removed
	  the card afterwards and reinserted it, but no circuit state had
	  to be saved meanwhile. The card not initialized anymore. Switching
	  the rack off and on again helped. This has now been fixed.
	* GENERAL: Support for the new B32 and P4B2 controllers
	* GENERAL: Show error message when trying to use B, P or I registers
	  as outputs, because that's very likely not what you intended anyway.

	* sequencer: Allow longer sequencers by chaining multiple sequencer
	  circuits (super easy!)
	* buttongroup: new select input for easily overlaying buttons with multiple
	  functions
	* buttongroup: New outputs buttonpress and longpress
	* algoquencer: same here: new select input for sharing buttons between
	  multiple instances (i.e. tracks) of alogquencer.
	* algoquencer: supports now 16 presets and two different load/save
	  methods. See updated manual for details.
	* cvlooper: Automatically adapt internal tape speed when external clock
	  changes. That way you can slow down or speed up the loop's content
	  by later changing the clock speed.
	* switchedpot: new convenciance input "bipolar". If set to 1 changes the
	  output range from 0...1 to -1 ... 1 (by the internal operation *2+1)

	Bugfixes:
	* sequencer: Fix the meaning of the "steps" input. Now if this jack is
	  unused or zero, it will never reset the sequence. This makes live
	  easier when using repeats.
	* clocktool: Fix missing first or second gate just after starting your
	  DROID. The reason was a too long gate length that two consequitive
	  gates merged into one. The gate length now defaults to 10 ms, which
	  fixes the problem (thanks to @klngvrhltnss)
	* clocktool: Fix reset handling: If the reset was not synchronous with
	  a clock signal, one clock would be missing, so for example if you
	  had "divide = 2" the clocktool would start with a missing beat (offbeat)
	  after a reset. This has been fixed (thanks to @klngvrhltnss)
	* algoquencer: Fix crash after a couple of seconds if length is greater
	  then 16. This bug has been introduced with the new pitch1..pitch16
	  jack (thanks to @antti_tiihonen)

green-7:
	New features:
	* GENERAL: enable fast 4 bit mode for SD card (hardly noticable,
	  but hey, these 3 additional wires would be wasted otherwise)
	* GENERAL: If your DROID patch has an error (blinking LEDs), the
	  reason for that error will be written into a file in the SD
	  card with the name DROIDERR.TXT. So if you are courious why
	  your patch won't load: simply look into that file!
	* GENERAL: Improve ADC sampling so that the input latency is better.
	* GENERAL: Support for new P10 controller!
	* sample: new input "gate" for continous sampling.
	* minifonion: new inputs noteshift and selectnoteshift

	Bugfixes:
	* GENERAL: Fix rare hickup where all input LEDs go white and
	  the inputs do not read correct values anymore. That was introduced
	  in green-6.
	* copy: The inverted signal of 0.0 was -1.0. This has been
	   fixed. It is now 1.0.
	* lfo: Avoid total DROID freeze in case of high negative
	   rate value, which would create a frequency of practically
	   zero.
	* switchedpot: Virtual pot values changed after DROID reboot,
	   sometimes the value even doubled. This has been fixed.
	* clocktool: Fix missing the very first clock after reboot
	  when it comes to clock division. It started the count one
	  too late. Thanks to @klngvrhltnss.

green-6:
	New circuits:
	* chord: Four voice chord generator (similator to that of the
	    Sinfonion)
	* arpeggio: Creates melodies by rules and pattern (similar but
	  much more flexible than that of the Sinfonion)
	* adc: Converts an (analog) input value into a binary representation
	    of up to 12 bits.
	* dac: Convert a binary representation of up to 12 bits into an
	   (analog) value
	* fold: Keep CV within defined boundaries by folding it by fixed
	    amounts (e.g. octaves)

	New features:
	* GENERAL: Significantly cut down the RAM usage of the circuits,
	   so you make even bigger patches. For example one togglebutton
	   now needs just 224 instead of 544 bytes.
	* GENERAL: Now a backup file of the DAC calibration is saved
	   to the SD card if one is present.
	* GENERAL: Drop the maximum number of internal patch cables.
	* GENERAL: The state of the circuits is no longer saved in the
	   flash but in a file in the SD card named "DROIDCAL.BIN".
	   This frees 128 KB of flash for more circuits. It also allows
	   copying the settings from one DROID to another. The disadvantage
	   is, of course, that you now need the SD card in the slot all
	   the time if you want your state to be persisted during two
	   power cycles. The DROID patch is still written to the flash
	   and does not need the card for being loaded.
	   NOTE: After the update you will loose your current settings *once*.
	* sample: New input timewindow: introduced short bypass phase after
	  a trigger. Tackels problems with many digital sequencers.
	* compare: More intelligent default value for ifequal.
	* algoquencer: Create a 32 steps or longer pattern with the new
	  parameters "alternatebutton" and "alternatebars".

	Bugfixes:
	* sample: The trigger input didn't work. It did essentially nothing.
	  Therefore the whole circuit was useless. This is now fixed. Thanks
	  to @Des_Rois for finding this!
	* algoquencer: If you set the length shorter than the number of
	  buttons, you would get "ghost copies" of the current-step
	  indication light. This has been fixed.
	* notebuttons: They didn't save their state. But do now.
	* cvlooper: The gate input did only work when a low gate
	  was *precisely* 0. So using I1...I8 directly as a gate
	  input didn't work. G1...G8 worked. Also internal reliable
	  values like directly mapping B1.1 to the gate input. This
	  has been fixed. Thanks to @Fyde for finding and posting
	  this bug!
	* contour: Fix contour stucking at sustain level if just
	  trigger, attack and release are used.

green-5:
	New circuits:
	* transient: create (possibly very long) transients.
	* midifileplayer: Play MIDI files from the SD card.
	* superjust: Creating a perfect pure intonation of up to 8 voices

	New features:
	* algoquencer: Make the "morphs" parameter's range more useful. It's maximum
	  at 1 is now exactly one morph per step. It has a higher resolution with
	  less morphs at the low end of the range.
	* droid: new input "ledbrightness" for setting dimming now all the
	  LEDs from the master and G8.
	* contour: new parameter "startfromzero". When set to 1, a trigger
	  or gate completely resets the current level of the envelope.
	* calibrator: new parameters "ledup" and "leddown". When used the
	  LEDs in the nudge-button indicate the current amount of
	  adjustment.
	* GENERAL: The maximum number of circuits is no longer limited to 128.
	  Now, if the actual memory consumption of the circuits in your patch
	  allows it, you can have as many circuits as you like. The actual limit
	  when using just very simple circuits is at about 500 now.

	Bugfixes:
	* GENERAL: color for missing SD card is now white as the manual
	  states (was cyan).
	* GENERAL: Fix permanent freeze (solved by factory reset) in case of
	  too many used interncal patch cables. Thanks to mvdirty.
	* GENERAL: Fix P2B8 pots not reaching exactly 1.0 in some cases (just
	  0.99975)
	* math: Fixed quotient, sine and cosine.
	* algoquencer: Fixed activity parameter when no buttons are used. It would
	  not be effective and always 100% of the beats would play a gate.
	* algoquencer: Use better random number generator, which produces more
	  variations in the patterns.
	* algoquencer: Fix a LED bug which green-4 has introduced. If the manually
	  set pattern length was less number of buttons, several LEDs would
	  flash at the same time, instead of just the one of the current step.
	  Also fix a bug that the last LED would never be lit.
	* algoquencer: Morphs would never go back totally to zero once they were
	  active. This has been fixed.
	* algoquencer: Fix influence of "variation" on pitch. It was inverted!
	  Also it didn't honor the dejavu setting and was always random.
	* algoquencer: Fix length of pattern if you change your configuration
	  to use fewer buttons than previously and also use the lengthbutton.
	* algoquencer: Use same relative amount of accents as configured with
	  the button, when doing variations. Previously in variations the accent
	  propability for generated variaton notes was simply 50%.
	* switch: Fix reset input. When it got a trigger the complete DROID
	  rebootet. This has been fixed now.
	* quanizer: Fix steps input. It would divide 10 octaves, not one.
	  Now steps = 12 really gives 12 steps per volt. With the bug it
	  was 12 steps per 10 volts.
	* slew: new input "gate" for enabling/disabling slew limiting.
	* contour: Fixed bug where contour was hanging at 10V. Thanks to
	  mvdirty for finding and nailing down this bug in a small patch!
	* burst: This circuit was actually missing, not detected, could
	  not be used. Now it's ready to go! Thanks to mvdirty.


green-4:
	Bugfixes:
	* GENERAL: O1 ... O8 outputted for a short time after reboot a value
	  of -10V. This has been fixed. Now the outputs start at 0V when powering
	  the DROID up. Once the circuits run (a few ms later) correct values
	  are output. When you reboot the DROID (e.g. with a button press) the
	  outputs now stick to the last output value - without going to any
	  other value inbetween for a short time.
	* GENERAL: light LEDs even for very small non-zero values (was black below
	  a certain level)
	* contour: Fix hanging sustain phase, when you use the gate input but
	  send an ultra short gate (below 1 ms).
	* math: difference output did never work. Was stuck to 0V. Instead the sum
	  output was showing the differnce. But now works!
	* logic: negated is now correctly responding the threshold.
	* algoquencer: LED indicating the current step was always one step ahead. This
	  has been fixed and the LED now correctly shows the current step.
	* algoquencer: Fix clock/reset timing detection. Now if reset and clock
	  happen within a window of 5 ms, it does not matter in which order they
	  happen. In such a situation always the first step of the pattern will
	  be played. This is a great help for dealing with reset/clock via external
	  sequencers, where the reset signal is shortly delayed by the internal
	  processing of that sequencer.
	* sequencer: same here: clock and reset within 5 ms are considered to have
	  happened exactly at the same time.
	* switch: same here: reset and forward/backward within 5ms are considere
	  to be simultaneously and the switch is being reset to 0.

	New features:
	* GENERAL: During the (very short) start phase the LEDs will be lit
	  one after another in cyan now. In the (naturally almost unimaginable) case
	  that the DROID cannot complete its startup procedure the last lit LED
	  will indicate the position where it's hanging.
	* burst: new circuit for sending a series of triggers

green-3:
	Bugfixes:
	* contour: Fix mode with trigger: now correctly skip decay
	   phase and directly go into release.
	* contour: Fixed shape input. It had no effect, but now has.
	* switch: Fix case with one input and multiple outputs. Also remove the
	  splitting of input and output switching. It was nonsense. Now there is
	  just forward/backward/reset for triggerd switching and offset for addressable
	  switching. Both can be combined, of course.

	New features:
	* contour: decay, release and swell times are now 20 times longer.

green-2:
	* GENERAL: Fix freeze in case of certain P2B8 errors
	* random: min/max now are named minimum/maximum (for consistency)
	* copy: min/max now are named minimum/maximum (for consistency)
	* algoquencer: start with all enabled, accents on downbeats


green-1:
	Bugfixes:
	* GENERAL: parse droid.ini with Mac-style CR line ends correctly
	* GENERAL: Make taptempo detection more sane when tap cable is unplugged
	* lfo: fix taptempo input in LFO (didn't work at all)
	* lfo: phase input of LFO now works correctly
	* contour: better balance of timing of log / lin / exp shapes:
	  The perceived "duration" of the decay / release phase does
	  not depend on the shape so much anymore.
	* algoquencer: removed output "startofpattern". Didn't work anyway.
	  Please use "startofbar" instead.
	* algoquencer: fixed "rollcount": is now the number of *additional* beats
	  as the manual states.
	* algoquencer: fix missing pitch changes when rolls are active.
	* math: fix 'minimum'. Was doing the same as maximum.

	New features:
	* GENERAL: Allow "input = A - B", by automatically converting this
	* GENERAL: Allow division by constant numbers, e.g. "input = P1.2 / 12"
	  into "input = -1.0 * B + A"
	* GENERAL: You can now switch between several DROID patches by simply
	* MANUAL: New symbol 1.2.3 for jacks that read or output integer numbers
	  holding buttons on your P2B8s while pressing the load button.
	  Please consult the user manual in section 4.12 for details.
	* togglebutton: new output "negated"
	* slew: new inputs slewup and slewdown
	* queue: new circuit: clocked CV shift register.
	* spring: new circuit: physical spring simulation.
	* nudge: new circuit for modifying a value in steps using two buttons
	* calibrator: new circuit for fixing badly tracking VCOs
	* lfo: new output "output" and input "waveform".
	* algoquencer: gatelength is now relative to the input clock
