@FelipeTellez - I did :) A bit of Python code to open the WAV file and capture the frequency peaks, and then a bit of javascript to generate the animations that I then record. I think it turned out pretty ok :) Glad you're enjoying them!
Nice, scientific summary! So the result is "happily", oscillators are behaving mostly very different at least at fm-ing. The tune-drift of some of the OSCs are just an offset I estimate and you can adjust the tuning to be in tune with other OSCs again? To avoid this completely, you could also use SYNC, which results in other/more dynamic "material". What I found quite interesting overall is the fact, that while adjusting the modulation depth, some of the sidebands "wandering" downwards the frequency and some upwards. Somehow a bit like a carousel. Should use the spectrum analyzer more often...
@kitkatandy909 - The downward wandering took me a while to figure out... it sure looked weird to see some going down while others were going up, but this is result of that carrier drift. Let's say you've got a carrier at 440 and a modulator at 330. The upper side bands would be 770, 1100, 1430, 1760... The lower sidebands start at 110 and then, when the math says it would go to 110-330 = negative 220, it reflects back around to 220. Then -550 goes to 550, and -880 goes to 880. (This is all going to be my main video). Now imagine that the carrier drifts upward from 440 to 490. Those reflected sidebands would start by appearing to go down -- the sideband at -110 would drift upward to -60. But because it's reflected around zero, it actually appears to drift down from 110 to 60. So some of the sidebands are going up, but any of the reflected sidebands will appear to drift down -- and then back up again when it hits zero. Wacky. :)
Hi, I make some quick FM calculations and your sidebands spectrum shown on at time 1:15 - 1:25 is wrong, because there can not be frequencies 330 and 660 Hz.
You're right, the image I was showing at the time was just a "stock image" that I had ready, not something from the specific "carrier at 440 & modulator at 330" tests I would run. That's a good catch.
What made you choose 440hz and 330hz modulator? I'm not really a big fan of FM mods because they can easily get crazy and unmusical, but surely you could have used two frequencies which sounded better together? Very interesting experiment and info nevertheless, but as I observed, I wish more modular synth demos sounded like music, and less like an annoying science experiment.
@pixelwash9707 - To some extent, this was an experiment. The best sounding frequency choices are when the modulator is an even divisor of the carrier -- like 440 & 220. However, in those cases, you can't see what happens when the lower sidebands drop below zero hz -- they just wrap around to lie on top of the positive sidebands. In the 220 hz case, you'd get lower sidebands at 220 hz, 0 hz, -220 hz (which is just 220 hz) and then -440 hz (which is just 440 hz), and so on. But in the 330hz case, you get lower sidebands at 110 hz, then -220 hz (which is 220 hz), which adds a new sideband in where there wasn't one before. It was specifically that "what happens when you go past zero and start to add in new sidebands" behavior that I wanted to compare.
Are all of these examples linear FM? Are they AC coupled? I have some Animodule VCOs that I really like, but the triangle wave is slightly offset so I need to use a high pass (I actually have a purpose-built passive filter in order to AC couple) if I'm doing linear FM with them. I wouldn't expect the humble audio sine to be offset, but it might be.
@andrewduncan529 - Yes, these were all linear FM. And I *think* they were all AC coupled. it's quite common for the FM input to go through a small value capacitor to filter out the any DC offset. I know there are a couple that let you switch between them, but none of these do.
@Fluxwithit - It's actually something I built. I capture the audio, run it through a program that does the spectrum analysis/FFT and outputs data that is used to generate the visuals. It's a bit more work, but this way I can make it look/act exactly as I want.
You "WROTE" your freq analysis tool!? If you did, that is mega cool. Loving your explanations😊
@FelipeTellez - I did :) A bit of Python code to open the WAV file and capture the frequency peaks, and then a bit of javascript to generate the animations that I then record. I think it turned out pretty ok :)
Glad you're enjoying them!
@@SoundsofVoltage It looks very profesh! kudos!
the spectrum is nice. but i'd really have liked to also see the waveform at the same time. anyway, thanks for this nice comparison!
Great video.
Thanks!
Nice, scientific summary! So the result is "happily", oscillators are behaving mostly very different at least at fm-ing.
The tune-drift of some of the OSCs are just an offset I estimate and you can adjust the tuning to be in tune with other OSCs again? To avoid this completely, you could also use SYNC, which results in other/more dynamic "material".
What I found quite interesting overall is the fact, that while adjusting the modulation depth, some of the sidebands "wandering" downwards the frequency and some upwards. Somehow a bit like a carousel. Should use the spectrum analyzer more often...
@kitkatandy909 - The downward wandering took me a while to figure out... it sure looked weird to see some going down while others were going up, but this is result of that carrier drift.
Let's say you've got a carrier at 440 and a modulator at 330. The upper side bands would be 770, 1100, 1430, 1760... The lower sidebands start at 110 and then, when the math says it would go to 110-330 = negative 220, it reflects back around to 220. Then -550 goes to 550, and -880 goes to 880. (This is all going to be my main video). Now imagine that the carrier drifts upward from 440 to 490. Those reflected sidebands would start by appearing to go down -- the sideband at -110 would drift upward to -60. But because it's reflected around zero, it actually appears to drift down from 110 to 60. So some of the sidebands are going up, but any of the reflected sidebands will appear to drift down -- and then back up again when it hits zero.
Wacky. :)
@@SoundsofVoltage Quite interesting at first sight! Thanks about clarification of the "maths" behind this. Thats why I love your videos so much 👍
Thank you
Hi, I make some quick FM calculations and your sidebands spectrum shown on at time 1:15 - 1:25 is wrong, because there can not be frequencies 330 and 660 Hz.
You're right, the image I was showing at the time was just a "stock image" that I had ready, not something from the specific "carrier at 440 & modulator at 330" tests I would run. That's a good catch.
What made you choose 440hz and 330hz modulator? I'm not really a big fan of FM mods because they can easily get crazy and unmusical, but surely you could have used two frequencies which sounded better together? Very interesting experiment and info nevertheless, but as I observed, I wish more modular synth demos sounded like music, and less like an annoying science experiment.
@pixelwash9707 - To some extent, this was an experiment. The best sounding frequency choices are when the modulator is an even divisor of the carrier -- like 440 & 220. However, in those cases, you can't see what happens when the lower sidebands drop below zero hz -- they just wrap around to lie on top of the positive sidebands. In the 220 hz case, you'd get lower sidebands at 220 hz, 0 hz, -220 hz (which is just 220 hz) and then -440 hz (which is just 440 hz), and so on.
But in the 330hz case, you get lower sidebands at 110 hz, then -220 hz (which is 220 hz), which adds a new sideband in where there wasn't one before.
It was specifically that "what happens when you go past zero and start to add in new sidebands" behavior that I wanted to compare.
its spinning
Are all of these examples linear FM? Are they AC coupled? I have some Animodule VCOs that I really like, but the triangle wave is slightly offset so I need to use a high pass (I actually have a purpose-built passive filter in order to AC couple) if I'm doing linear FM with them. I wouldn't expect the humble audio sine to be offset, but it might be.
@andrewduncan529 - Yes, these were all linear FM. And I *think* they were all AC coupled. it's quite common for the FM input to go through a small value capacitor to filter out the any DC offset. I know there are a couple that let you switch between them, but none of these do.
nice video. what scope are you using , very clean layout.
@Fluxwithit - It's actually something I built. I capture the audio, run it through a program that does the spectrum analysis/FFT and outputs data that is used to generate the visuals. It's a bit more work, but this way I can make it look/act exactly as I want.
@@SoundsofVoltage brilliant