Ghost frequency in FM oscillator

[Costis]

Hi All,

I'm pretty new to Dune 2 and loving it already.

While playing around with FM osc and watching results in a spectroscope I noticed that if both amounts for operators A and B are set to zero there is still an inaudible frequency (around -57 db) which is double the main one. This works with both algorithms. I would expect a pure sine there with no additional frequencies - is this intentional?

I can reproduce this with Melda's MAnalyzer and Robin Schmidt's Signal Analyzer.

[Mark]

There are residual signals coming through from operators A and B. You can check this by entering negative values in the mod matrix for FM Amt A/B with Const as source. I imagine there's a reason for it, suppress denormals/NaNs maybe? Anyway, as you say, inaudible, so, not critical.

Welcome to the forum btw

[Costis]

Thanks Mark!

Yeah funny you can indeed route consts at -100 to both amounts and its gone - so zero is not zero... I was guessing the reason is avoiding division by zero too.

While we are at it, how can one replicate the FM osc using sines and the matrix? When sending osc1 output to osc2 semi or fine in audio rate I am getting low frequencies below the central one which does not happen with the inbuilt FM oscillator. I used a high pass filter to get rid of them as a workaround. Basically I am trying to prove that Dune 2 can actually do 7 operator FM (contrary to some reviews that claim it has a basic FM implementation).

[Mark]

I believe Richard is able to remove the lower frequencies that one normally gets with FM. When you do it outside of that then you will get those lower frequencies. If your using integer ratios then there shouldn't be any need to high pass except for effect.

BTW, I'm guessing you're doing Osc3 -> Osc2 FM Amt A >> Osc2 out -> Osc1 FM Amt A >> Osc1 out? I've been doing that to get more complex timbres, but I think the modulation of FM Amt is technically RM, but close enough to have fun

[Costis]

I was more like thinking Osc3->Osc2 Semi then Osc2 -> Osc1 Semi then Osc1 out. Wouldn't that make it pure FM?
That's how I get those left sidebands though.

Dexed (a free DX 7 emulator) pretty much has the same behavior as the inbuilt Dune 2 FM osc so I must be doing something wrong with the routing.

[Mark]

That would be real FM yes, but how do you do 7op with that config? D2 like Dexed are both Yamaha FM which is really PM.

[Costis]

hmmm you got a point there. Now that I think about it If Osc1-2 were in FM mode, with this setup I would still be doing a weird FM/PM mix since I would be modulating the frequency of the whole signal of each osc instead of the first modulator. I would probably have to modulate the frequency(ratio or fine) of the first modulator in each OSC to do this properly but these are not available destinations in the matrix I believe.

Maybe Richard can implement this in the future so that no one can imply that D2 is not a full fledged FM synth as well.

[Mark]

The configure I suggested is good enough if you think of RM as a milder sounding PM, plus you can do feedback RM to add a bit more welly.