[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

The Echoplex upgrade arrives! (fwd)

With regards to the queries of Paul Dresher and others, here's a copy of 
Kim's announcement for the Echoplex upgrade, which includes a list of 
new/improved features.

---------- Forwarded message ----------
Date: Sat, 12 Jul 1997 17:32:03 -0700
From: Kim Flint <kflint@annihilist.com>
To: Loopers-Delight@annihilist.com
Subject: The Echoplex upgrade arrives!

At the NAMM show in Nashville this weekend, Oberheim is announcing a new
software version for the Echoplex Digital Pro. It's called LoopIII V5.0,
and will be shipping this month in new units and as upgrades for existing
Echoplex owners. There are many, many improvements and enhancements over
LoopIII V3.2, which has been shipping in the Echoplex for over two and a
half years now. Many of you have been waiting a long time for this; the
wait is finally over!

I'm not sure what Oberheim's upgrade policy will be for this version, you
should contact Oberheim or an Oberheim dealer for details.

LoopIII V5.0 is created and developed by Aurisis Research, a company
quietly formed six months ago by Matthias Grob, Eric Obermuhlner, and
myself to continue developing the Loop(R) technology invented by Matthias
years ago. This is our first big release as a company, and we are quite
proud of it. LoopIII V5.0 represents a huge amount of effort in technical
development, business negotiations, and evangelizing the possibilities to
Gibson. It was a daunting task to say the least, but we now have a high
quality software product in the Echoplex and a firm commitment to the
product from Oberheim and Gibson. We would like to thank all the users
whose enthusiasm for the Echoplex motivated and inspired us. We never would
have made it without you!


If you are interested, here is our release documentation, for LoopIII V5.0:

LoopIII  version 5.0
>From Aurisis Research

 Who is Aurisis Research?

Aurisis Research was founded by three music industry veterans to develop
exciting new musical instrument technologies. We license these technologies
to major manufacturers in the industry. Our primary products are real-time
looping instruments.

 What is Loop(R) ?

Loop, our centerpiece technology, is a real-time sampling and looping
musical instrument. It encompasses an efficient and intuitive user
interface allowing musicians to create, perform, and manipulate live loops
in ways that have never been possible before.  The revolutionary sampling
functionality of Loop has redefined the art and technique of looping. Loop
has been in development for over seven years, and can be found in the
Oberheim Echoplex Digital Pro and the Paradis LoopDelay. LoopIII is the
third generation of the technology.  Version 5.0 is the latest release.

 What is LoopIII version 5.0?

Our goals for version 5.0 centered on playability. A good musical
instrument is defined by the subtleties and nuance of playing it, and with
this release we sought playing perfection for LoopIII. We made many
enhancements and improvements to the existing functionality, to improve the
playing experience with the Loop.  All features that had been intended for
the LoopIII function set have been implemented. In addition, we fixed every
bug that has ever been reported to us, as well as many that no one ever
found. Stability has been improved everywhere to insure that no musician is
ever let down by the Loop. We did everything we could to make sure that
Loop continues to be the preeminent looping tool and an excellent musical
instrument in it's own right. We hope we've succeeded!


 *Loop Sound Copy*

 The long absent loop copy function arrives!  This function allows the user
to easily and intuitively copy his current loop into the next loop, during
performance. When NextLoop is pressed to jump to a reset loop, the copy
process begins. While copying, the loop continues to play, so the
performance is seamless. Even better, during the copy, overdubs can be made
on the new version! And just like the Multiply function, the new loop can
be made to have as many multiples of the original as will fit in memory.
With two button presses, you can take a one bar rhythm in loop 1 and turn
it into a sixteen bar vamp in loop 2 with a melody recorded over the top!

 *Loop Time Copy*

 This function is very similar to the sound copy, but just copies the
length. And, in the same fashion, you can overdub during the copy and
create multiples of the original length. Keep your loops in a tight groove!


 *Improved Use of Memory*

 A fundamental change was made in the software architecture to fix the
problems with Retriggering in cases where memory is low. Now we can
retrigger loops that take the whole memory at any time. Before, when using
retriggering functions like Mute-Insert, Mute-Undo, Mute-Multiply-Beat,
Sample triggering with SamplerStyle = One and even syncing to external
clocks, the retrigger would generally not work when the loop was longer
than half the total memory available.  This improvement is possible because
we now do real-time "defragmentation" of the loops. We unify the fragmented

 *Retriggering Improved*

 The Retriggering function, typically executed with Mute-Insert, has been
rebuilt.  It is now protected from interference from Overdub and Feedback,
and is generally far more reliable when used along with Reverse and Undo.
Once the loop has been triggered, repeated pressing of Insert retriggers
the sample for stuttering effects, rather than going into Reverse or 

 *MIDI Sample Triggering Improved*

 The entire function for MIDI sample triggering had to be rebuilt. It is
now considerably more consistent and usable. Changes are:

- The volume is now corrected when the switching is executed.  This fixed
problems where the volume setting for a loop would be retained from
previous switching, especially when the SamplerStyle or Velocity parameters
had been changed.

- The velocity of a midi trigger now sets the volume relative to the volume
set by midi continuous control. So the Continuous Controller set by
VolumeCont now serves as a master volume.

- The switching is fully dependent on SwitchQuant. So, for example, if
SwitchQuant is set to Cycle and a midi trigger is received, the new loop
will not start until the current one reaches its end. Setting SwitchQuant
to Off makes it behave most like a Sampler.

- If Velocity=off, and SamplerStyle=Att, the key switches between full
volume and silence.  This fixed problems where loops would play with the
volume completely off or at a setting from a previous switch's velocity
value. Now if Velocity is turned off, the loops always play at a consistent

- If Velocity=on, the velocity is used to set volume independently of the
setting of SamplerStyle.

- NextLoop now always plays the loops at full volume. Before it would play
them at volumes set previously by midi loop triggers, which was confusing.
It is now independent of the settings and usage of SamplerStyle and

 *Exiting from Quantize*

 Its easy now to get out of the quantizing (ooo) state. Just press the same
key again and the function happens immediately. This might be useful if you
normally Quantize but as an exception do not want to wait for the end of
the current cycle, or if an expected sync signal does arrive. Additionally,
Undo will cancel the impending function and the quantized wait state,
returning you to where you were before.

 *Improved MIDI Clock Sync*

 The Echoplex's ability to synchronize itself to midi clocks has been
substantially improved. This was achieved through a combination of bug
fixes, internal parameter adjustments, and complete rewrites of some
software functions. The Echoplex stays locked to the midi clock source far
more reliably than it did before.  It used to have trouble when the clock
drifted faster and the sync signals arrived early, especially when the loop
was longer than half the memory size. That is no longer a problem. The
Echoplex is also more tolerant of wider deviations in the midi clock. So if
the clock source is drifting from its original rate, the Echoplex will
continue to sync to it over a wider range. The amount of drift we allow in
the clock before giving up is determined by an internal parameter, which we
essentially doubled. If the clock has drifted beyond this range we assume
it is intentional and the Echoplex stops trying to sync to it. If the clock
returns to a reasonable rate, Loop will sync up to it again.  Also, Loop
now recognizes if MIDI clock was present and then stopped,  which prevents
waiting endlessly for a clock that never comes.

 *Recording of Many Loops in Sync*

 If AutoRecord is on and we are synced to an external source, AutoRecord
turns into the Next-Insert, or Time Copy function.  This allows the
Echoplex to record multiple loops, all synced to the external clock. To
avoid confusion in performance, we also allow the Record button to
terminate this function and round off the recording to the next cycle
length. (instead of terminating immediately and defining a new cycle
length, as InsertRecord usually does).

 *StartPoint Location with Insert and Multiply*

 Before, if several inserts and multiplies were done, the StartPoint would
end up in strange locations. Now it is much more intuitive, and maintains
its location at the beginning of the first audio sample. The exception to
this is when Insert is done right at the very beginning of the loop, in
which case we assume that is now the StartPoint.

 *Display for Insert*

 The display was not very intuitive during inserts. Now, when inserting in
the middle of a multiplied loop, the total number of cycles is shown
counting up. When Insert is ended the cycle count reorients itself to the
actual location.

 *StartPoints with Insert*

 Loop is now much better about keeping track of cycle number locations when
insert is done several times on a loop. Before, the beginning of the count
could end up almost anywhere after several Inserts were done. The cycle
count now stays with the audio in a much more obvious way.  Functions that
retrigger the loop work in a much more obvious way, since the StartPoint of
the loop stays in an obvious place.

 *Audio Recalibration / Improved Undo*

 Loop is now able to recalibrate the audio hardware to account for audio
degradation associated with thermal variations in the analog to digital
conversion parts. This probably improves audio quality in many subtle ways,
but most importantly is the improvement to the Undo function. There has
always been a problem where the audio hardware sometimes developed DC
offsets over time. Loop misinterpreted this as an audio input, which meant
that new memory would be used continuously and the Undo function would not
work well. The thermal recal totally fixes this, making Undo much more
reliable! Audio recalibration is done very quickly, but it does require
that audio be off very briefly. To avoid any pops, the recal is done with
each reset, and repeated periodically if Loop is left in reset. This means
that a user still has the rare potential to encounter the Undo problem if
they power up, begin a loop immediately, and never reset. This is actually
fairly rare, and we have even improved this situation by making the startup
sequence longer, allowing the convertors to get closer to their operating
temperature before the initial calibration. And in the extremely rare case
where a unit does develop the offset during use, as soon as the user does
Reset, the problem is fixed. Before it could only be fixed with a power
cycle. So even this worst case is much better than the offset situation in
the previous version!

 *Noise Reduction*

 We employed some software tricks commonly used in communications systems
to reduce system noise. This will likely reduce noise that could sometimes
be heard in the audio.

 *Stereo Operation*

 Operating two Loop based devices together for stereo is now far more
reliable and consistent.
- The synchronization functions are now much smarter and all the problems
with one unit getting confused by sync signals from the other have been

- Synchronizing stereo units with external devices, either as a clock
receiver or transmitter, is now far more reliable as well.  Before this
could only be done by removing the BrotherSync cable or using particular
midi cable configurations.

- Many bugs have been fixed where midi messages sent to the master from an
external source were not correctly passed on to the slave.

- Several bugs that caused functions on the slave to execute differently
than on the master have also been fixed.

- The problem where feedback on the slave would be set slightly low,
resulting in long term loop degradation, has been fixed.

- The parameter defaults are now set for stereo. So a new user will not
have to do anything for stereo to work.

- Velocity messages for LoopTriggering now transfer to the stereo slave

- Volume continuous controller now passes to the stereo slave correctly.

- The stereo units now operate together very nicely!


- When a multiplied loop is in Reverse, the cycle count now counts 

- Accuracy of Reverse is improved, mainly in case of retrigger functions.

- Multiply-Reverse is now possible. Due to processing limits cycle counts
cannot be maintained so all multiples become one long cycle.

- No more pops!

 *Delay Mode*

 In DelayMode, the output is now off while recording. This is not
delay-like, but better. Before there would be odd situations where the
previous delayed audio would appear in odd places as a new delay time was
being set. Also, during Mute, the feedback and input sound continues
updating the unheard delay sound. This way you can record audio into the
delay while the output is muted, and then turn the output back at a later
point to hear what is in the delay.

 *Sample Dump*

- various bugs fixed, plus automatic workarounds for some bugs found in
other samplers and software.

- Handshaking, or "closed loop" now works properly and really does go
faster than the standard open loop dump.

 *Manufactureability Improvements*

 Power supply variations on some units could cause slightly low readings
for Feedback, causing problems where the loop would decay even with the
know all the way up. Units had to be tested for this problem in production
and fixed before they could ship, which was inefficient. Some units
undoubtedly shipped with such a problem, and there is the possibility of
developing it as the hardware ages. Changes have been made in software to
tolerate worst case voltage variations across all the relevant circuits.
This corrects the problem, with the tradeoff being a slight reduction in
feedback resolution at the top end of the scale. It doesn't appear to be
noticeable, and is definitely worth it for the improvement in reliability.

*** BUG FIXES ***

- A lot of pops and clicks have been removed.  (including the ones
occasionally found in Reverse)

- NoteOFF is always treated as if it had velocity 0. Thus, the NoteOFF
command of modern keyboards with dynamic key release are understood to
release the keys of the ECHOPLEX.

- There was an attack noise at end of PlaySample. (the Mute-Insert
function)  This has been fixed.

- MuteMode was not saved properly in the EEPROM.  After power cycling the
display would be correct but it would always be in the "Continuous" mode.

- MIDIFeedBack send did not work well. It did not even reach the max and
min value. Now it is sent every 30ms.

- Some display errors, mainly in connection with Next or Quantized,  have
been removed.

- In some special situations of NextLoop, with Quantize on or empty loops,
we got stuck or jumped immediately. These cases are fixed.

- The sound sometimes faded after a lot of repetitions. A bug with feedback
was fixed that should cure this. A similar bug with stereo operation was
fixed, so slave units won't have mysterious loop fading problems.

- In confirm, NextUndo did not influence the running state, which was not
useful because you want to arrive in the new loop Playing. So NextUndo now
always does StopMute and StopOverdub.

- Quantize/sync and threshold are now "anded."  If both are on, Record is
executed only at next beat after the signal appears.

- Record-Undo is improved. (Undoing an accidental press of Record)  It is
maintaining the rhythm now and also uses the new defragmentation procedure,
so it is possible with a loop as long as the entire memory minus the
accidentally made loop. Unfortunately, the Multiply structure is lost so
the loop becomes one longer cycle.

- Overdub can be operated in parallel with other keys much more reliably.

- After an even number of Multiplies, AutoUndo did not coincide with the
start of the loop any more.

- When certain functions are called now, the old data on the Quantize stack
is erased. That fixes a Long Multiply problem and prevents the situation
where we come back from the parameters and are stuck waiting for something.

- Sync was redone. If it arrives early, we call Retrigger. Before it did
not work when more than half the memory was occupied.

- Insert-Multiply was wrong. It should now work as described in the manual.

- The Syncs do not interfere with the long-presses of real buttons any 

- Undo LED now turns red when the key is pressed.

- The amount of variation allowed for syncing to a changing clock speed is
not a technical problem, but a constant we set. This limit is imposed since
the speed of the music in the loop does not change, rather the loop end is
chopped off or the beginning repeated. Too much variation in the external
clock tends to cause a rhythmic mess, so we assume the variations are
intentional and stop trying to sync. Nevertheless, we have now doubled this
constant. This tends to improve the reliability of the sync without causing
too many problems, and seems to be a better value.

- Source# only goes to 114 to leave space for all midi functions.

- Clock out for 8ths/cycle = 1 did not work at all, this is fixed. The
whole algorithm for counting and generating midi clocks changed and all
measures are more accurate now.

- The whole Memory overflow calculation is new. (did not work the first
time in a new loop)

- The Multiple display now counts down while Reverse is on. But while
multiplying in reverse it is still counting up (how would I know where to
start counting?).

- In case of quantizing to external Sync, the second press now executes the
function but does not erase the Quantize flag. A reset then erases it.

- NextMultiply sometimes fell directly into Mute, now its fixed.

- Multiplication happens independently from loopstart and consequently the
counting is not synchronized with the loopstart.  In many situations, the
difference between the loopstart and the counting is only due to limited
operating accuracy. Therefore, RETRIGGER now eliminates this difference if
it is smaller than a constant BeatCountRoundTime (actually 100 blocks =
70ms). This avoids the short appearance of the last number when we trigger
the loop and strange flickers when syncing. These still appear if the
multiplication was far off the loopstart and thus presumably intentional.

- NextMult cannot work forward if we are coming from a reversed loop. So
now, it is taking over the direction.

- If the MoreLoops parameter is changed, and then immediately set back to
the current value, the loops are no longer reset.

- Midi LoopTrigger commands now pass to the slave correctly, including case
where AutoRecord is on.

- LongNext now brings you to loop 1 when the current loop is reset.

- Velocity messages didn't get transferred to the stereo slave, now fixed.

- Volume continuous controller not sent to slave. Now fixed.

- Quantized multiply ended at the very end of memory left it waiting
forever. It now ends the multiply correctly.

- When syncing to midi clock and using Quantize, functions sometimes
executed a cycle too late. Now they always execute at the end of the cycle,
like they should.

- Insert could pass the end of memory in some cases. Fixed to recognize the
end correctly and kill the Insert automatically.

- Multiply of max-length loops tried to work anyway.

- Some functions did not execute right if Overdub was held down. That's

- NextLoop sent over midi sometimes did two NextLoops when switching to a
reset loop. We recognize the note offs better in this case.

- Master setting slave feedback to 125 instead of 127, causing loop fades.

- Inserts done in the middle or beginning of a loop could cause it to go
over the maximum memory. This condition is now recognized.

- Front switches now work when a pedal in the overdub jack is held down.

- Multiply-Undo messed up Undo so that the loop could be undone back before
it's original point in memory, bringing back old loops. Now it stops in the
right place.

- When the stereo master had feedback turned all the way down, it did not
set feedback correctly on the slave after a Record. Now it does.

- Tempo of midi clock out was lost during quantized reverse. Now it is kept

- LED's are set correctly during quantized Loop switching. Some available
functions were not correctly indicated.

- Plus many other very minor ones....

Kim Flint                   | Looper's Delight
kflint@annihilist.com       | http://www.annihilist.com/loop/loop.html
http://www.annihilist.com/  | Loopers-Delight-request@annihilist.com