Tapped.
It didn’t do anything. Cue sends work for preceeding and neighbor track.
v7 just uploaded. Have reverted to using -12dB stages as I think it’s clearer, and is consistent with the -4dB stages for input cue signals. Have clarified usage in the notes at bottom of diagram.
1 Like
What about preview level ?
1 Like
I’m not sure that I fully agree that the inputs are somehow attenuated by 12dB or that this is a helpful conceptual idea to perpetuate without some certainty
afaicr - there is no analog gain reduction at the inputs and anything heard via the OT is subject to A/D>D/A or D/A - the digitizing is happening at the doorstep
If the inputs are attenuated (therefore presumably digitally) then it wouldn’t be possible to record un-gained hot inputs close to FS digital
Using the device as prescribed by thinking in terms of unity positions is possibly good practice, but is it a true reflection of the stages where gain is applied and the digital representation is modified from source capture
it’s something I too would appreciate some clarification on - so I’d be careful about printed documents without caveats, it sorta perpetuates ideas that aren’t 100% proven
Yes a valid point… but I do say it’s a ‘personal interpretation’ in the title, and the notes in the latest v7 also state the machine ‘behaves as if’ there is a 12dB attenuator, so I hope these caveats are sufficient. I’m happy to accept my assumptions are wrong. My reasoning was based on measuring signals in & out, and using different routings to isolate which stage is doing what:-
- Route an input direct to the main output.
- Route an input via a thru machine.
- Route an input via a rec buff.
- Save the rec buff, copy the file to a computer and measure it.
- Assign the saved sample to a flex machine and play it.
- Import a known sample and play it.
2 Likes
I made another version of great @philtipping work, with his collaboration and very kind approval. Inspired by Elektron diagrams, not finished.
Octatrack-MKII-Gain-Staging_will5.pdf (170.4 KB)
Waiting for critics to make it better, some values may change, the idea still being unity.
For Static / Flex, I set Attributes to +12db, it could have been Vol = +63 (+12db), because I estimate that the samples are played 12 db lower.
Open to debate about this.
I think I’ll add physical inputs / ouputs. Necessary ?
11 Likes
Thanks a lot again for your effort, very helpful.
Key take-aways so far (if I understand it correctly):
- clipping at +15 dB,
- in general, unity path at 0 (-64 to 64) or 127 (0 to 127),
- exception 1: recorders/thru machines: -12dB,
- exception 2: cue from inputs: -4dB.
Edit: corrected a +/- typo.
This just makes no sense in the digital domain though …
dB is just a means to express a ratio, it doesn’t have a meaning in terms of clipping a digital signal … you can’t have +15dB relative to 0dB … what is 0dB in this scheme ? what do the LEDs really indicate ?
In the digital domain you would express clipping in dBFS and that naturally ‘maxes out’ at 0 dBFS you can’t have definitive +ve values wrt clipping
I linked to an earlier thread on aspects of the gain structure, but that related to the Mk1 which has inputs which are not as broadly accommodating as those on the Mk2 which can handle higher input voltages
I think Elektron have designed the interface to prevent clipping in most usage scenarios, so the notional 0 points on bi-polar ‘gain’ encoders are there to keep you right if you have a full set of tracks running - but are these nodes really capable of adding gain or are they attenuating and which ones if any(presumably) do add gain (besides the process in the editor and at the inputs)
To capture a near FS signal into the Mk2 may require e.g. a hot eurorack signal - it should be easy after that to monitor whether the captured audio can be clipped if these waypoints are adding gain - and a hot signal will allow it to be tested for recording/thru/pickup machines etc - I should look at this on the Mk1, quite curious - I’ve chopped and changed my mind about what’s probably happening in the OT, but I am increasingly confident it is well configured and thought through by people who know what they’re doing
The - 12db values in Machines column are not settings but theorical - 12db reduction for inputs / recordings.
As a compensation, Recordings have +12db in Attributes GAIN, and Thru needs Src / Playback page VOL set to +63 to add +12db for unity.
As @philtipping told me, it’s actually - 4db, corresponding to LEVEL 100 (from manual).
It’s a fixed value, you can’t access to it.
Thanks MK7. Agree with take-aways, except +4 should be -4.
All values are dBV so are relative to 1 volt RMS (approx 2.8v pk-pk). My bad for lazy terminology 
Feed a 15-16volt p-p signal into A, set mixer gain A/B to 0, DIR to 127, Main lev to 0, and main output shows clipping.
1 Like
The electrical i/o specifications are in the respective Elektron user manuals, though they stipulate those in dBu, likely the same for the Mk2 manual - that reference scheme would make more sense to avoid potential confusion especially if folk read across both sources of info
I corrected the +/- typo above, to avoid confusion when people just jump in and read those bullet points.
1 Like
Fair enough for the input clip level, but in the context of this diagram, I wanted to show levels as they appear on screen, so plain old dBV (or just dB as Elektron themselves use in the manual and on some controls), seemed better to me.
I did read & appreciate the numerous threads on this subject, including the Gain aGain thread, but as mentioned in the first post couldn’t find any diagrams… so as I’m a ‘pictures guy’, thought I’d have a go
If I missed it, apologies
I got the impression you don’t have a signal generator as you mentioned using a hot eurorack signal to investigate clipping. I have a generator and scope so may be able to help if you need any tests doing.
2 Likes
Very cool to see people figuring that stuff out!
Is there any chance to get the differences to OT Mk 1 described?
Thank you and cheers.
Input levels on the Mk1s were giving people issues with hotter signals, so those inputs were upgraded along with the encoders/screen etc
The Mk1 Input handles +8dBu (5.5V peak to peak)
+17dBu for Mk2 (15.5V peak to peak)
sure sounds more convenient, but does that result in folk experiencing/perceiving samples which sound more underwhelming (due to loss of level relative to imported samples)
3 Likes
Concerning settings, same for MKI, AFAIK.
In the diagram, SRC is PLAYBACK, CUE+REC1 is CUE+RECAB…
Yeah shoutout to the upgraded inputs, I was plugging modular straight in the other day, inputs red but no clipping 
1 Like
These figures don’t mean that you can feed a +17dBu signal into the MkII inputs and record/process it “as is” and expect not to clip the recordings. If you do so, you must turn the input gain down to adjust for anything that overshoots +4dBu. It means that the input stages can handle it without distorting beyond specs. I don’t see any reason for a loss of level compared to imported samples due to this hardware mod.