A few questions prior to purchase

[DrJustice]

I'm on the brink of ordering an Origin. However, researching the instrument uncovers some ambiguities and brings up a few questions that are important to me:

* About Programs and Multis:

In the manual, section 8.2.2.2 ("Multi Preset") says: "A [MULTI] consists of a selection of up to four Program presets along with various settings for use in both live performance and host sequencing/recording situations. In a Multi, the FX as well as step
Sequencer, arpeggiator, and 2D Envelope and Galaxy modulation settings are shared among all Programs."

In chapter 6.2 ("The program preset") it is mentioned that the FX, arpeggiator, sequencer, 2D envelope, Galaxy modulator are parts of a program.

This information conflicts with, or at least creates confusion in relation to chapter 6.1 ("Overview of the global structure of Origin") and 6.3 ("The Multi preset") which show a multi as having 4 programs each with a sequencer and 3 FX.

If the multi mode indeed shares the three FX, one arpeggiator, one sequencer, one 2D envelope and one Galaxy modulator among all four programs, that raises some questions of importance to those wanting to use the Origin as a multitimbral sound module:

 1 - will not the programs loaded into a multi sound dramatically different as they lose their individual FX, arp, seq, 2D env and Galaxy mod?
 2 - as you load programs, which programs FX, arp, seq, 2D env and Galaxy mod takes effect?
 3 - if saving a program that has been used in a multi, which FX, arp, seq, 2D env and Galaxy mod settings gets saved with the program; the current sounding ones or the originals of that program?

* About the sequencer user pattern data:

Chapter 7.10 ("Using the step sequencer") says: "Each sequence used in a program consists of three sub-sequences which can be used to modulate different parameters. This way, you can easily set up complex sequenced modulations, all controlled hands-on using Origin’s [Sequencer] section. You can also find up to 200 sequencer presets that you can assign to any of the existing programs.".

This paragraph, as as well as the ambiguity of the sequencer sharing between programs and multis mentioned above brings up another question :

 4 - are the user sequencer patterns stored with each individual program (or multi, depending on the above issue) or are they stored in a single global bank?

* Some additional questions :

 5 - is the Origin per sample or block iterated?
 6 - how does it handle feedback connections between modules (relates to 6)?
 
 DJ
 --

[Sean W]

DJ,
the current manual was written before Origin V1.0 was released and contains some inacurate information. The new manual is being prepared for the upcoming V1.2 release and will be corrected and updated will al the new features introduced since V1.0.

Each Program has its own effects (3 per program, serial or parallel configuration), arpeggiator, step sequencer (3 sub-sequences per pattern), 2D envelope and Galaxy module.
http://www.arturia.com/evolution/en/products/origin/details/guided-tour.html?docId=826

In a Multi (up to 4 Programs), each Program (routed to its own stereo output) retains all its modules/parameters.
Therefore, a Multi can use up to 12 effects (3 per program), 12 sub-sequences (1 pattern per program), etc.
http://www.arturia.com/evolution/en/products/origin/details/guided-tour.html?docId=848

Regarding the sequencer, there is a "pool" of 256 patterns (128 preset patterns, 128 user patterns) stored in 16 banks of 16 patterns.
Each pattern contains 3 sub-sequences, each with assignable trigs, accents, slides.
This pool is shared by all the programs, but each program remembers which pattern is used by the program (bank and pattern), and the routing of each sub-sequence.
The pattern for the "active" program can be switched on the fly.
http://www.arturia.com/evolution/en/products/origin/details/guided-tour.html?docId=828

Not sure what you are asking in your additional questions.

[DrJustice]

Thank you for the prompt reply Sean!

I'm glad to hear that the multi mode does not compromise any functionality. I now see the guided tour is actually a better reference than the manual in some ways. I would suggest that a more complete block diagram is added in the next revision of the manual.

Regarding my last two questions, I wonder if the whole system is iterated once for each sample period or if it is block iterated, meaning that each module is iterated several times before moving on to the next module. This relates to real time performance, as in latency in the signal chain, and thus also the behaviour when using feedback paths.

Edit to add: is an overview of the CPU load for the various modules available?

DJ
--

[DrJustice]

So I've ordered mine. In part due to seeing that the support and development continues.

I guess I'll find the answers to those last questions when it arrives....... 

DJ
--

[Cord]

Congratulations! You won't regret it. This is a deep machine and if you ask me, I find the updates coming slow but therefore the OS is very stable. I mean, very stable. I saw different and I rather wait a couple of weeks for quality work.

[Superwaldi]

Congratulations, Dr. Justice.

I guess, you couldn't put your hands on Origin before the purchase. I'm curious about your experience.

[synthguy99]

I ordered mine just based on experience with the Mg Modular V.  I loved it, but hated the latency which I couldn't get rid of on my PC.  I could never get the Origin out of my mind, but cringed at a retail price of $3200 US.  When I saw it for $2000 here in the States for a limited time, I decided to take the plunge, and have been amazed at its power and flexibility.  Even with a gob of synths on hand, it's what I go to first when I fire up my arsenal.

[DrJustice]

Thanks guys! I reckon it's going to be big fun.

@ Superwaldi : that's right, there's nowhere around here to get a hands-on experience. Thus I try to research it properly, not least by reading the manual. It all looks good to me after Sean clearing up a couple of points

DJ
--

[DrJustice]

OK - I've had my Origin for a day now, and can honestly say that I'm impressed. Running firmware 1.2.1 without any problems.

Now to answer my own questions about block or per sample iteration: I have tested this and it definitely seems like the Origin is a per sample iterated system - Philippe Wicker, do you want to confirm this, please?

DJ
--

[Philippe]

I have tested this and it definitely seems like the Origin is a per sample iterated system - Philippe Wicker, do you want to confirm this, please?

It's block based. The block size is currently 32 samples (.725 ms at 44.1 Khz sample rate). The order of module computation is determined in such a way that most of the time the latency due to the processing of the 32 samples time slice is only 1 buffer (so .725 ms at 44.1 Khz).

[DrJustice]

OK. However, I made the following patch (there is also a KBF going to the OSC and filter and an ENV going to the two OUT modules just to get the patch going steady by triggering with a 1 step sequence):

Code: [Select]


    OSC-----+----FILTER----OUT1(pan left)
            |
            +--------------OUT2(pan right)


When I record the stereo signal, there is not a phase difference of 32 samples between the left and right channels. I've tried with the filter at 10kHz and a slight bit of resonance just to be sure that the left channel is filtered and the right channel is not. The recording shows a phase difference of 1 sample period. How can this be?

DJ
--

[Philippe]

Imagine the following patch:

OSC1 -> FILTER1 -> VCA1
OSC2 -----------------> VCA2

When the voice is triggered (eg because of a NOTE ON) the OSC1 is first required to generate 32 samples, then these 32 samples are passed to the FILTER1 which generates 32 filtered samples. These 32 filtered samples are then passed to the VCA1 which may apply an envelope if any.
During the same time frame, OSC2 is requested to generate 32 samples which are then passed to VCA2 which may apply an envelope if any.

The key point is that all these computation steps occur within the *same* time frame. A time frame here is the interval of time corresponding to 32 sampling period, that is 32 / 44.1 = .725 ms, 44.1 KHz being the sample rate.

[DrJustice]

Aha, so another way to say that is that it's iterated starting from the 'back end', not from the output, right?

So if one makes a feedback connection around a single module (e.g., OSC out -> OSC in), am I right in thinking that path has a delay of a single sample period? And if making a feedback connection around several modules, the delay will be 32*n_modules sample periods?

Please bear with these questions, I'm very interested in finding out the behaviour when using feedback.

DJ
--

[Philippe]

32 samples in both cases

[DrJustice]

I now understand the iteration and buffering scheme used in the network. Thank you very much for explaining that !

DJ
--