Gain aGain

This is a follow up post which perhaps questions a few of the assumptions made in a thread from a few years back … some of the statements have been taken to be a statement of fact (not least because of what is stated in the manual and in terms of how evaluations were thought through originally)

I’ve revisited this afresh in a new thread in light of the general frustration with managing levels throughout the OT, particularly for capturing Live (and especially now we know the inputs are easily pushed by comparison to the MKii) … but it also relates to a passing comment i picked up from someone at :3lektron: that some the truths that permeate on Elektronauts about the OT are not quite what they pretend to be.

So, one of the recent comments I also read related to the absolute need to avoid going into the red and i had a long held question in my mind about whether the ‘Gain’ on the Mixer page was indeed a gain or an attenuator with on offset to encourage better level preparation for stacked tracks

There is one truth about the OT inputs, the do pass every signal through an AD converter, so even if you pull up the Dir, it is now Digital and coming out again via a DAC, but the legend tells that ‘every input is subject to a -12dB gain reduction’ and that is optionally re-applied via the Gain Parameter on the Sample Attributes page in the Editor (this defaults to +12dB) … in the manual it is documented as follows

So i’ve got used to these two statements (often red really does equate to awful clipping btw), but tonight I tried to see how hot I could push the Inputs and opted for a loud unfiltered Doepfer DE MK1 Osc … manually dialling the VCA level up (to 4 fwiw) until the waveform was nearly clipping the screen (wait what ! not merely taking up a quarter of the screen with -12dB hit !) as I had already set the Gain to +63, the Meter was way way into solid red (same as track button) - I also set the same test with the Mixer Gain at 0 and it does indeed look like a -12dB drop (i could have regained to check)

So i’m wondering, as i have previously wondered, is the OT offering real Gain or is it all managed attenuation … have these recordings gone through a gain reduction (inputs -12dB ?) and then boost (+63) or has the +63 nulled the ‘gain reduction’ (i.e. what is 0 really) (note the inputs were way way into the red, but there’s no clipping) - Note the sample attribute Gain is applied only to playback (it’s not applied to the sample)

I’m wondering if we can find out what’s really what by doing a bit of thinking/research (or asking the question direct) it may all stack up as originally proposed … is there any further comment from MKii users (besides smug ones :wink: ) ? At the end of the day, this is not an academic exercise, i was a little disappointed about how quietly i had the Rytm set to feed the OT without it seemingly overloading, just keen to bottom out what’s happening inside if possible

Here’s two images of the direct recording (T1 + CD) then (CD + Edit) {no vertical zooming by Enc D btw}

Depending on whether we uncover some new truths this may yet get merged back into an older thread

I’m wondering if @PeterHanes and @sezare56 or @Open_Mike have any MKi opinions or if @AdamJay @LyingDalai or @William_WiLD have any MKii thoughts to throw into the pot, or indeed anyone else ?

I’m hoping to build up a visual picture of all the gain/attenuation stages for all signal paths through the OT, especially for the inputs and Thru machines


edit : update below, there is a significant clarification

23 Likes

Really interesting but unfortunately I can’t make decent tests without my mixer/soundcard/DAW plugged. I need to finish the room for that, and it’s a total mess.

I’d need to calibrate inputs too. I made a test with a passive guitar directly plugged in input A. It seems to add real gain, as I allways thought, but now I’m not 100% sure.

2 Likes

I’ve been preaching about pushing hotter signals into the OT for awhile now as I feel I’m in a minority about claiming to get a good mix, my meters may flash red here and there and I never hear clipping, but I tend to keep it just below that…

I never actually use the gain parameter in the mixer. I always have the Rytm piped into AB and run it direct. The Rytms output is hot enough that I’ve never messed with OT’s gain, I just use the Rytms volume knob, which I set and forget…
Mackie vlz4 with channel faders and mains at unity goes into CD and I control the connected source levels from their own volumes. Nothing I’ve connected has had to be boosted, usually the volumes are around 3/4 and below max, so I haven’t messed with gain there either.

I monitor sometimes from my mixer and sometimes through the OT and switch mid performance, so I need the OT to output the same level signal as what’s coming in, and also not be affected by which method of monitoring I’m using on the OT…

The following are my “OT Unity Gain” settings:

-Master volume: 19
-If using a master track: track level 127

-Mixer direct monitoring: dir 127
-Flex playback of buffer: track level 127
-PU: track level 127, monitoring level on rec pg2: 127
-Thru: track level 127, amp vol +49
-Any values not listed: default

I always run 24bit for both flex and recorders, I don’t know if that makes a difference in all of this vs 16bit…
For the recordings I do always leave the +12db as is…

I’ll also mention again the noise gate, it gives you a db level of where it cuts off. Setting it around -14db takes a pretty healthy signal to get through. I’m curious how this measurement plays into the whole equation as well.

That’s what background info I have for now, I too would like to know what exactly is going on with the attenuation/boost scenario, just where does clipping actually occur, and what the differences are between compensating here or compensating there…
I do think I get a decent mix though by pushing hot signals, so really even though I’m curious about the details, I haven’t had to think about this for awhile because it’s been working nicely and sounding good…
I can’t really test anything for a few weeks as I’m moving, but I’ll stay tuned…

9 Likes

i remember back in the days i was confronted with a muddy sound out of my OT, although my samples have been top-notch

a member told me to lower the track-level below 100 … so i usually use 96 and it works

in sum i found the leveling-process quite complicated, a bottom-up process

sample-level
track-level
main-level

and if resampling for transition is involved it takes another leveling-step

so if :3lektron: is ever going to built a new Octatrack DPS-2(!) an automated leveling-process would be nice

2 Likes

did you never felt, that any signal pushed into the red sounds ‘hot’

Samples from DAW, normalized to 0db ?

No,but yes, not selfmade but bought from Loopmasters

I just made a test with a 0 db Sine wave on 8 tracks. I had to set tracks Level to 90 to avoid clipping.

1 Like

I’d imagined that it may have been set up to have about 4 hot tracks tip it over, on the basis that there’s no likelihood you’d have 8 hot signals

In the test above, everything is pre-track of course, I’m just looking at getting good signals in (normally I’m averse to adding ‘gain’ at inputs to focus more on getting a hotter source sorted)

it may well do, I thought so too, but I’m just a little bit curious if ‘Gain & 0’ in the Mixer is just a way to manage clipping/levels - I guess the electrical specification of the inputs needs to compared to the source which needs to be quantified (i do have an old uni scope I picked up somewhere but I’ve no idea if it’s working/calibrated) … it may be possible to use the A4 CV side to offer up ‘known’ signal levels (i.e. Value Linear with sin LFO at audio rates)

tbh, I normally do too, I’m surprised I didn’t set this up, I’m sure iirc I just ran a new vanilla project with default settings (16b)

ah, may be good to have this in the kit bag to make sense of some of this

Not in this particular illustration, for sure I’ve noticed issues when it wasn’t far into the red, so maybe in this case I just had a lower signal at 0 state than normal (and to be clear, I was setting up VCA levels with max Gain) so it may only challenge the notion that you shouldn’t touch the red (but it depends on the Mixer gain)

I do not have made precise analytics on Gain staging with a focus on Inputs Management signal because I never SERIOUSLY Deal with sample/record material on OT inputs and never really have to deal with a mixing hub scenario like a live instrument on inputs to thru my setup.

At the beginning I had 3 Elektron Gears where OT part the hub to get signal for 2 devices and OT making his own sampler thing in between. And I was unhappy with the results… I remember the theory regarding Midi notes level as 127 is max (possibly peaking) and I started to put everything between 90-100 and it was indeed way better [ at the END ]

But still I was preoccupied by my general main outputs for my Live and I Fastly invest in a mixer to deal with audio signals with all tools I know from the DAW on a computer. I deal my Main Outputs like this all my Gears separated (as I would mixing a track in a DAW but with a difference I do not go with Headroom as far as with a full internal mixing where I was at -16dB/-12dB but more like -8dB/-6dB so including the stack loudness I get something clean and deal later with compensation if I need it on the mixer and later) as a Bus > through Analog Heat for Shaping my Master and give it more punch : generally there we check with Audio Engineer to make it Good into the PA (and he reserve him a little range here to deal if I go crazy but I pay attention to that by starting a bit lower myself and up the level progressively to my Max zone.

So as you can realize I do not have to worry with Inputs in my scenario because i globally use pre-recorded material and I deal with that at the computer level (because i come back to hardware after years ITB and now the beginning of my young years on Gears for me start to be too far away for remember really)

I have Latin Percussions, Rode Mic but I record it with my RME directly to the computer…
For now I doing all myself and no collaboration for now LIVE with Musicians

but it’s indeed very important…
And as I will introduce more creative ideas making on the road with OT MK2 and Percussions / iPAD i probably check things and MAY have thoughts to throw into the pot…

i bookmarked this thread and the old one and coming back with more analytics on OT gain staging because I do feel it’s important for me as well to always get the most clean and usable recording signal.

1 Like

that’s where I’m coming from, if there’s a way to get a better snapshot of your live inputs I may be more inclined to just sample into the OT to capture short ideas rather than hooking up other recording options

1 Like

yup same for me with ideas I have in mind recently I will have to deal with that too, and I think OT still be best candidate to start directly to shape something from recorded material…

in this analytics way of thinking I would give here my thoughts based on OT MK2 of course…
Practically sure to have some things to write next week. Sticked note on my fridge to come back on that thread :wink:

1 Like

Yeah off course it’s a particular test with exactly the same signal at 0db.

1 Like

I’ve always found to get the sampled level back I can always get there within the +63 via amp page, I think you might be right about it being attenuation rather than gain there. I just record healthy signals then mix by cutting rather than boosting, Then I can always bring the overall level up a decent amount at the end without clipping. Failing all else if I can’t get decent levels from something (which is rare) I stick a pre on it

1 Like

Indeed, the OT clearly works well on the whole for what it’s intended to do, but I’m on page 1 wrt capturing the sample at its best, useful info like this is for page 2 and so on :wink: … the gain attribute on the sample settings page defaults to +12 for captures, but that’s for playback … if you want to take a nice captured sample out via USB disk mode I think i’d rather do fine level tinkering in other software, but getting the best signal possible will be a bonus, thus the focus on the Mixer Gain and input attenuation at the moment, first things first, especially wrt those pesky LEDs

1 Like

Ah, i think i have discovered that there’s been a bit of a prior misunderstanding, possibly on my part and it’s been perpetuated …

I tried Setting the Gain to 0 on the mixer to see how far i could push the DE, it could go way beyond the 1/4 scale captures you’d see if the inputs were categorically being lowered by -12dB … but not far enough, the VCA seemed to level off after about 7>8 and it was maybe 2/3rds or more full-scale on the OT and the shape was good/clean (red inputs)

So i stacked up some 0dB sine samples on three tracks on the AR and that produces a punchy sine level to push into the OT at Gain level 0 (i was able to get closer to full scale and could eclipse it by adding more stacked SINs introducing distortions to the waveform)

This is the result … it’s a single harmonic, no clipping and it’s clearly not attenuated by -12dB (i.e. the sample)

Applying Gain will introduce a shape change as the Sin is clipped at the peaks and troughs, so it’s absolutely contradicting what i’ve dropped into discussions from time to time based on the assumptions put out in the older thread experiments

That’s not to say i have everything understood, it’s just clear that the previous assertions are wrong - ps - fwiw, this signal was also way way into the red, but it’s not distorted (or Gained)

so i need to rethink input metering and using the OT as a capture device, it’s better placed than i assumed before for some of my usage cases, that’s not to say the way jot’s set up by default isn’t best for typical usage scenarios, but it ought to be possible to capture decent samples with a bit of prior configuration especially at source

I also think that Gain in the Mixer IS gain !

It’d be easier to test all this out with proper signal generating equipment, but i’m satisfied a few mistruths have been addressed, albeit posed originally as questions (and fuelled by the comments in the manual about applying automatic +!2dB gain on captured material)

I’m still not convinced there isn’t more info to uncover wrt all of this, but i thought this was significant in terms of my understanding of what was occurring

4 Likes

The manual’s wording is pretty unambiguous but I wonder if what’s actually going on is that the LED metering is set up to give about 12db headroom, and rather than being padded -12db like it describes the reality is that the LEDs go red at -12dbfs s if you follow their directions for setting input levels and get it so it’ peaking just below red, you’ll end up with an audio file that has 12db of headroom.

It wouldn’t make any sense for the manual to be written the way it is in that case, but it WOULD explain a lot of the unexpected gainstaging behavior.

EDIT: after work this afternoon I’ll do some experiments focusing on thru machines.

3 Likes

Has elektron ever stated why there is no digital FS peak metering for the inputs?

I am having trouble grasping the implications of htese findings… These apply to OT mk I? So on the mk I, redlining the input meters is the way to go?

fwiw I think that the whole -12dB ”pad” was a design mistake and we’d all be better off if such a safeguard were never implemented in the first place… Digital audio fundamentals are not rocket science, if some dweebo clips the mixbus by overly hot signals its not fair the rest of us need to suffer from it too… Even MPCs clip the internal mixbus if you don’t mind your track levels, and most users learn to get on with it jus fine.

Does anyone happen to know the internal bitdepth of the OT mixbus? 32bit float? 24bit fixed-point? other?

2 Likes

I only have a MKi. It’s more nuanced than some of us maybe thought before, but it’s clear you can in some circumstances (without gain or attenuation) get good recordings when it’s solid red (or even if you need to use gain obviously) and there’s not necessarily massive level reduction at the point of capture either, so you’re good to go for quick and dirty sampling with decent levels (without resorting to edit boosts etc) - makes it a bit more usable, but clearly there’s conflicting forum chat about the significance of hitting red at all - because i’d always thought the gain reduction was a given, I never sought to maximise the digital levels the way you might in an audio editor for best signal to noise (it’s always easier to attenuate than to correct for low gain captures)

5 Likes

In general for setting levels I’ve used the noise gate and set it around -14db and adjusted source until it’s peaks pop through, and I’ve looked at the waveform to in the editor to analyze its height(amplitude). I’m always a bit concerned with how far I push it (red meters) to reach what seems still an OK level so even though it sounds good, I drop it back a bit…

A general concern has been about not being able to record hot signals into the OT, I’m always sampling gear with their volume knobs most of the way up without clipping…

My settings with high track levels may work out for me because most all my samples are captured by the OT, I don’t know if I used pre recorded samples especially normalized if it will still sound OK. I also don’t use volume boosting fx…

I believe track level default is 108, so it seems that should be fine?
Lately I’ve been reconfiguring for track levels around 100, because I often directly cue the inputs which then output to cues at a level equivalent to 100 on mixer direct…

I’ll occasionally load a regular sample and I’ve noticed on one going through a few fx that affected volume, I had to drop the track level to around 50 to match my live sources…

This may be why I feel I get better sampling mix results, because everything I’m mixing is audio passed through the OT or sampled by the OT, I’m not using pre recorded loud samples to have to mix around… And again this may be why I can get away with high levels everywhere…
@avantronica, I know your main focus is the input levels, but I have to bring outputs back into this in case it’s possible to make too hot of recordings that cause us to have to drop track levels…

If normalized samples clip the OT when using high track levels, and noisebuddy and others have experienced muddy sound, can we drop the master instead of the track levels?
I wonder if you adjust track level, master track level, master level, sample gain, amp vol, if they each have a separate discrete effect on headroom? Is it the sum total of all of them that will end up clipping? Might amp vol and sample gain have a discrete affect but levels be additive?

Be interesting to find out how these all affect each other, and to bring back to topic what would be a desirable recording level to not have to decrease anything on the output stage.
In other words, does a recording start getting too hot for overall summing before clipping occurs, causing us to drop output stage parameters when we could have captured a lower, more desirable strength recording that mixes well with default output settings?

3 Likes