Finally Got Around To

[toneburst]

..building my PreenFM2 last week. Very easy build, everything worked first time, and I love the metal case! Very pleased with it.



I've mainly been flipping through presets and DX-7 patches so far, but hopefully I'll get a chance to build some patches from scratch soon.
Hoping that randomiser feature comes along soon though, just to give me some inspirational starting-points.

Keep up the great work, Xaxier!

a|x

[toneburst]

One thing I have noticed, sound-wise is that a lot of patches only sound good within a certain range. A lot of sounds turn into wobbly mush at lower octaves, I guess because the lower-pitches operators are transposed into the LFO range. That's a problem with FM synthesis generally, of course. The other thing is the lack of operator feedback means that some of the DX-7 patches don't sound as they should, I suspect. This is especially the case with some unpitched percussion parches, I suspect (though I don't have an actual DX-7 to test this hypothesis).

a|x

[Mmarsh]

There is a difference in the sound engines of the Yamaha synths and the Preen.  Yamaha synths use Phase Modulation while Preen uses Frequency Modulation.  MY understanding is that feedback is easy with Phase Mod but not so much with Freq Mod...

[toneburst]

Yep, I knew that. Wonder if the DX-7 translation code could be tweaked to simulate operator feedback more exactly. Maybe some kind of extra oscillator waveform that mixed noise and sine wave in variable proportion... Just thinking out loud really.

I understand that exact simulation of existing (and now vintage) hardware isn't a priority, however. It's good that the Preen is able to import DX patches at all.

a|x

[Xavier]

Thanks for the feedback... 

Feedback is not easy to add and in adition is less usefull when you have the choice between waveform for each operator... which was not the case on the DX7.
Also at the maximum polyphony CPU usage is not far from 100%, so it's hard to add features at the oscillator level.

The sound difference is more due to the internal engine. The DX7 is very hard to emulate closely. Take a look at hexter code to see what i mean.
Also phase / frequency difference make the spectrum totally different as soon as you have 3 cascading operator or more.
So some DX7 sound sounds good others not.

Yeah i know, i must find time to add a randomizer 

[widdly]

I think that part of the difference  in sound across the keyboard is the lack of level scaling in the PreenFM.

DX7 has level scaling settings that change  the volume of the oscillators depending on which key is playing.  It is very useful in FM patches to get good modulation in bass notes without the shrill mess in the high notes.  At extreme depths you can use level scaling to split the keyboard so you have say a bass sound below C4 and a piano above.

You can partially simulate key scaling on the PreenFM2 by routing key number to the IM and operator volumes in the modulation matrix.

Since the scaling only occurs at note-on and is just a scalar of the oscillator volume, I imagine the PreenFM2 cpu could handle DX7 level scaling easily.  I suppose the difficulty would be squeezing the parameters into the menus.  The level scaling on the Dx7 has two curve settings (+lin,+exp,-lin,-exp), a breakpoint and two depth settings for each oscillator. 

One way of adding this feature would be to implement a few level scaling functions as modulation sources in the modulations page.  Then you could use these to modulate the volume, fm index or any other destination.

[toneburst]

@Xaxier
Interesting, thanks for the info. I will treat the imported DX patches as starting points. Many of them sound good in their own right (though there's a limit to the number of minor variations on the piano/clavinet/harpsichord I can listen to ).

@widdly Interesting... that explains all the DX-7 presets with 'split' in the name...
+1 for a scaling functions in the mod matrix.