LFO rates [& How random is random?]

Elektron Gear Analog Keys/Four
#1

Two things

Firstly, if you want a regular LFO shape to cycle every bar, you set the factor of speed and Multiplier to 128 (or -128)
so e.g. Spd 1 Mult 128, Spd 4 Mul 32, Spd -2 Mul 64
this is in free mode btw !

The random LFO is 16 times faster, so at the same rate as above, the random value will change every trig step, well kinda, certainly at that rate !

If you want to have a random modulation exactly every step, then you set the factor to 128 - however, the phase is not on the trig, it seems to be out of phase with the trig pulse, easiest to see at a slow tempo, if you slow the rate down to per beat (ie factor is 32, ie Spd2*Mul16) the phase issue becomes worse often, the lfo is changing on the 3 not the one ! This is after the sequencer is reset to zero, so the LFO is not starting at one …
So point one is about phase !

Point two - is about random, if you then switch the mode to Trg, ie restart the lfo on a trig then you’d hope it would work, it does with other shapes, but with random it starts with zero every time, so it’s not random, at least that’s how I see it

I think it is reasonable to expect to be able to sync with trigs a random LFO that starts at the onset of the Trig

What am i missing, or is this a feature oversight - what’s a workaround if any ?

Enlighten me

EDIT:some issues fixed in subsequent os updates (i.e. non random random) !!!

7 Likes
LFO speed and multiplier explanation
LFO Speed/Sync
A few insights on Digitakt LFO
I thought I understood the LFO's
Tempo Delay
LFO Sync
LFO Speed/Sync
Noise generator on one track gets “reset” by TRG on another?
Analog Keys Overbridge lfo speed stepped note sync
LFO types : How to hold?
Lfo
LFO panning
I finally solved and understand the LFO's
#2

To set a RND value every trig, try using the Hold setting on the random lfo. That should set a new value every new note in the sequence.

The mod parameter HLD-

• HLD makes LFO1 run free in the background, but when a note is played the output LFO1 level is latched and held still until the next note is played.

1 Like
#3

hey - thanks for the feedback - that’s a good workaround for a small subset of my intended uses, that indeed ticks the box of what to do if you want the classic sample and hold thing - but, but, it requires trigs, full trigs, so if you want to do something with the filter on a droning single trig (even a trig muted one) you get no changes at the lfo freq intervals - it’s helpful to keep that suggestion in mind, thanks, but it doesn’t provide me the full solution i’m after

i’m still receptive to ideas and indeed discussion about the phase issue and the ‘random’ issue

#4

as an addendum, it is possible to knock the LFO back in phase by using trig mode, then do a trig mute and put the lfo back in free - it’s just that any time the values are changed the phase is lost and it is not zero if you start from the beginning of the pattern - free is very free, seems to ignore start phase

the point about the random LFO starting with a zero value still stands, you can hear it if the lfo frequency is is set to twice the trig rate (so 256 in this case) - every note starts of with the root note, the second half of the trig is random ! not random

my interest is not entirely academic

1 Like
#5

very interesting topic!
I was trying to figure out how the LFO speeds really work but it put my head in.

Nice job figuring out the 128 factor, I couldn’t even come up with this…

but yeah some properties of the LFO seem to be:

  • it is always locked to sequencer tempo, it is never really free.
  • it doesn’t stop if you stop the sequencer though
  • if you change speed while it’s running, any pattern-aligned phase will get messed up, so it needs to be retrigged if you want to match the phase to bars or steps…(the same goes for the arp…)

what I find weird is the behaviour you get if you put it to HLD mode with its speed turned up… (imo try it with any of the sweeping waveforms like ramp or triangle, and routed to OSC pitch with a depth of 12… just fill up a pattern with 16th steps to generate oldschool S&H melodies with a range of two octaves) well… experiment with the values…

i expected this to be very “shifty” when speed is set to some odd value, .e.g. taking a good number of cycles before the LFO phase matches with pattern steps, but this is not the case! or I can’t really pull it off… the LFO will repeat its little melody pretty quickly…
BUT every now and then there will be a single odd note which falls out of the pattern, adding variation to this LFO melody… some kind of moiré-pattern effect… anyway it’s beyond me lol.

1 Like
#6

Need to have a further look at that, explorations pretty much turned out as expected, no surprises unfortunately, all a bit digital, not very analog- your suggestion raises one of my issues, it’s actually tricky to dial up a musical interval accurately e.g the octave (sure you can use a square lfo and tune the depth by ear, but there’s no way to know from the knob because the high resolution encoder provides low resolution feedback. You can monitor the nrpn output, but a means to know when you are on the MSB values could be handy

I suppose it makes sense that the lfo doesn’t restart at the reset of the sequencer, so my gripe about phase is perhaps just something to accept for what it is - a bit of detail about the pseudo random shape would be useful, I presume it’s more than a constantly repeating graph of stepped noise

I might have said this before but I wish there was clarity on the correlation between the start phase parameter val and the val in degrees, so you could guarantee a tri lfo started at max value -beyond auditioning it

1 Like
#7

Nice work with this. Just wasn’t getting it quite the way I liked before. Just always a bit off. Had the Speed at higher values mostly. All clear now.

#8

Thanks for the ‘128 = 1 bar’ tip. Just wondering, is there any reason why an LFO speed of 128 gives a 1 bar LFO cycle? Found the LFO speed a little confusing, so trying to understand the logic behind it.

#9

Due to the nature of it being an 7bit integer (128 values) it is ever so easy to divide (and multiply) by two, i.e. 7bit to 6bit to 5bit etc … therefore it’s just super easy to get musical periods from a repeating LFO built on those foundations, it’s part of the musical paradigm to maybe work with 1 bar or 2 bars or 4 bars or 8 bars, so a binary related number just fits with the grid … it’s very logical, but there’s no tie in with the midi standard, just a coincidence :3lektron: :thup:

1 Like
#10

What I once found confusing, now makes complete sense…ha! Going to make my LFO tweaking much more enjoyable. Thanks a lot the explanation avantronica :smile: :thup::thup:

1 Like
#12

Im trying to do some fake sidechaining with an LFO but the phase is incorrect and doesnt start with the sequencer??? is this correct?

#13

Are you in Free mode ? Try Trig mode.

4 Likes
#14

ill try thanks

#15

Any updates on this?

Did you manage to get the random to get a value synced to every step?

How to fake sidechain (4 on 4) with the LFO?

Edit: I followed the above steps and everything went fine. 128 is the number!! (I was going mad trying to find it by myself). Thanks!!