How to avoid aliasing when using LFOs?

Elektron Gear Monomachine
1

Hey folks,

I just played around with my Monomachine and did some drum synthesis just for fun. Of course I know, that this is not the main purpose of the MnM … and I remember that people in this community already found out about the fact that you can not sync the start of the oscillators etc. but that is not what my question is about!!

When I was building bass drum sounds, I experimented with one-shot exponential wave LFOs to simulate an ADSR envelope for my kickdrum. When doing so with white noise or some resonant filters, it was just fine. But when modulating a deep sine wave (doesn’t matter if oscillator-generated or oscillating filter) I noticed that the LFO creates aliasing artifacts, ruining my bass drum sound. I found a thread regarding this in the old elektron forum, but with no solution. One could say that this is the digital sharp charme of the Monomachine, others might be asking if this is something avoidable?

So my question is: Are there tricks/methods to avoid this kind of aliasing sound? If not … this wouldn’t be a complete turnoff, don’t get me wrong … I am very pleased of what the MnM already gave to me … but sometimes, this really bothers me :smiley:

Thanks for your input in advance, best regards,
Boland Ross

2

Are you modulating volume or something else? I made some kicks on the Mono a long time ago but never noticed it. If volume, perhaps you are clipping the signal. The Mono is really easy to clip. Turn the DIST down to -3 or -7 and you should get a much cleaner sound.

3

Yeah, this is about the volume parameter. I suspected clipping, too and did some precautions (as you suggested), but the bell-like ringing is still there.

4

Maybe try using the “regular” envelope, limited though it is, compared to the LFO. Since the LFO will boost the signal, you will have to only use a little bit of LFO before it starts to clip. The regular envelope does not boost.

I seem to remember making kicks and only having the LFO up just a few increments before it started sounding overdriven. I want to say something like 4 or 5 was tolerable and above that it clipped.

5

I just did some further tests and I can confirm that this is not about clipping or other types of improper gain staging.

Even if DIST is set to -40, the problem occurs.

The resulting sound much more sounds like using amplitude modulation in a synth using a rectangle waveform as a modulator. It doesn’t like like saturation or digital clipping… which lets me think that the LFO is modulating the VOL parameter in a harsh way, not interpolating -> stepping the values.

After googling a bit, I found a post from 2001 by a guy who found the same thing and even contacted Elektron about it…

http://elektron-users.com/index.php?option=com_fireboard&Itemid=2&func=view&catid=9&id=54197

I guess I’ll contact Elektron Support, this makes me sad.

6

Bummer, I just ordered mine too :stuck_out_tongue:
Does this only happen with really fast LFOs? It sounds like a control-rate thing. Even Max/MSP does this with control-rate messages.

7

I hope this doesn’t ruin your new gear experience

The Monomachine is a wonderful and mighty machine, which is also very rewarding, it just has its quirks (as any complex software driven device does)
It also happens when using slower LFOs, but when increasing speed, it gets more and more audible.

I would suggest you to register your Monomachine as soon on the official Elektron website and use the support contact form to create a ticket.

Maybe if enough people are complaining, we can make a change. :slight_smile:

8

Oh, thank you, I appreciate that, and from what I’ve heard, seen, and read about the Monomachine I really doubt it will ruin my experience :slight_smile: My TX81Z does this and so did my Micron, and worse, both have terrible aliasing on the envelope attacks. I still have the TX because it sounds great but I got rid of the Micron pretty quickly and this was only one of many reasons. And of course LSDJ on the gameboy does this like you wouldn’t believe!

My Nord Micro Modular, on the other hand, is generally quite snappy and smooth. But a couple years ago I was trying to use in a pseudo-multitimbral way, so that MIDI notes from different channels would be converted to a CC controlling pitch, and a “gate” note corresponding to channel that would trigger one of several voice structures within the patch. It worked, but I discovered that even when the CC was sent first, this created a portamento effect on every note, which I found very annoying, especially after putting in the work to learn a scripting language that let me do this. So the Nord was interpolating between the CC messages to avoid this type of jumping/aliasing, and for my purposes, this interpolation made things worse!

Long story short, the solution to this problem may be even less desirable than the problem itself. But after playing with my new Monomachine for a while, if I find this unacceptable I will go ahead and create a ticket :slight_smile:

9

This sounds like a side-effect of the “control rate” of the Monomachine, which cannot be changed, and probably cannot be fixed without a whole new Monomachine with a faster processor.

Every digital synth has an audio sample rate (e.g. 44.1, 48, 96 or 192 kbps), and a separate rate for control changes, which is usually MUCH slower than 44,100 samples per second!

“Low level” software like Max/MSP, Pure Data, Csound, Kyma, ChucK and Supercollider are the only ones which allow the use to change this.

You can actually “hear” the limit of the control rate change of any digital synth (including the Monomachine), by turning up the LFO rate applied to the volume parameter of a synth. This will create Amplitude Modulation, and you will hear a note, probably in the 200hz-4,000hz range. If the LFO allows you to go faster and faster, but the note you hear caused by the AM doesn’t change (I know you can do this on the A4 because it’s LFO has the “multiplier” parameter which goes soooo fast), then if you identify the note your are hearing (which can be done easily with a spectrum analyzer), and multiply it by 2… that is, roughly, the control sample rate for that digital synth.

All of this is to say… a fast envelope will have digital artifacts, if it is going faster than the control sample rate of the synth. I never thought of this as “aliasing”, but technically it is, but it is aliasing within the control sample rate limit, as opposed to audio sample rate aliasing, which is what aliasing is usually referring to (e.g. playing a high note with too many harmonics on an old sampler/synth).

Elektron could maybe add an interpolation algorithm to solve this problem? But this would also be more expensive processor wise.

Considering that Elketron came out with the Monomachine about a decade ago, and was creating it with technology available 2-3 years before then… This just seems like a necessary compromise in the machine, due to how many features it has.

The only hope I see of this changing, would be a Monomachine 2.0, which could make use of newer, faster, cheaper digital hardware…

Personally, I think one of the most genius aspects of Elektron’s machines is how they allow you to go beyond the boundaries of “normal limits” with their limited technology. Other companies will do all they can to hide these limits, by limiting how far a parameter can go.

Like on the A4/AK/AR, the LFO multiplier makes it go WAYY beyond what the control rate can handle… but they include this so you can make crazy “control rate aliasing(?)” sounds with it! I actually think this is totally ingenious, and embracing the limits of technology as tool in itself, where EVERY OTHER Synth company in History, just about, has done the total opposite! For instance, if Roland or Korg made the Monomachine, their “solution” to this envelope problem would have been to just not let you create such a snappy envelope. Problem solved! No one would ever hear aliasing, but a few will complain that the envelopes “aren’t snappy enough”.

10

Thanks for the explanation, fisherking111!

11

np (:

12

Very interesting! I know on the Virus TI they have 4 or 5 different modes for filter sweeps that seem to be addressing the issue that fisherking111 mentioned.

The manual has a big discussion about what the pros and cons are of the various types of modes. I remember reading about this and wondering what are the Access guys smoking. But now it all makes sense.

13

This is interesting!

Am I getting this right? Let’s say …

If you have a sine oscillator playing on the A4, the oscillator is analog and gets its pitch frequency setting set by a digital controller? (for example for each note that is programmed in the sequencer)

Now … when you use an LFO to modulate the pitch of the sine oscillator, is the LFO also an analog oscillator, like on a modular synth system, or is the LFO just a mathematical function in code, which influences the analog OSC attribute?

I hope you understand what I’m trying to ask.

Best regards,
Bolandross