Attn: FM and Chip synth addicts

[Kayster82]

Been using these for a couple of weeks...These are awesome!! I will write a review, these VSTs needs more exposure!

[Echoes in the Attic]

Ahem. This thread should read "Attn: FM and Chip synth addicts who use windows".

Would love to see a mac version!

[Sendy]

Well you guys get Sinevibes plugins, so I figure we're about even

Is it me or is there something unique about the Sega Master System squarewave tone? It doesn't really sound like the others. It has a sort of cute but almost "whiney" character which SPSG really captures well. Anyone know what I'm talking about?

[ENV1]

Is it me or is there something unique about the Sega Master System squarewave tone? It doesn't really sound like the others. It has a sort of cute but almost "whiney" character which SPSG really captures well. Anyone know what I'm talking about?

Maybe it is a 'derived' waveform, you know, like with the TB-303 which 'derives' its square from the saw?

Just a guess, but this could easily explain such an 'off-standard' character...

[Aly James]

Nope, there is no vodoo on the generated square wave, which is a 50/50 duty cycle pure square.. BUT,
The pitch resolution is really non standard, as it is derived from a clock based counter there is only 1024 possible frequencies (the range differs according to the master clock speed the chip is clocked with) that means that it will be slightly detuned compared to other elements in a mix

You can also notice (if you take the direct output without filtering) that the chip generates a positive voltage only for the high portion of the square, which result in a DC offset @ 0Db... where standard waveforms usually goes from +1 to -1 the output here is +1 to 0.
you can see how this behavior change the position of the waveform when it is filtered, instead of having a straight line.



On a side note, the Timer modulation trick tthat produce a "SID lead" sound type uses an emulated Z80 Interrupt routine that write to the volume register at audio rate, when you write to the volume level max/min in a loop at nearly the same frequency than the actual playing tone it is like producing another square wave on top of the real one..
The result is that it creates a slow phase shift producing a real variation of the resulting square duty cycle
something that the chip itself cannot do without this trick.

All of the above make it pretty unique indeed.

[Sendy]

Nope, there is no vodoo on the generated square wave, which is a 50/50 duty cycle pure square.. BUT,
The pitch resolution is really non standard, as it is derived from a clock based counter there is only 1024 possible frequencies (the range differs according to the master clock speed the chip is clocked with) that means that it will be slightly detuned compared to other elements in a mix

You can also notice (if you take the direct output without filtering) that the chip generates a positive voltage only for the high portion of the square, which result in a DC offset @ 0Db... where standard waveforms usually goes from +1 to -1 the output here is +1 to 0.
you can see how this behavior change the position of the waveform when it is filtered, instead of having a straight line.



On a side note, the Timer modulation trick tthat produce a "SID lead" sound type uses an emulated Z80 Interrupt routine that write to the volume register at audio rate, when you write to the volume level max/min in a loop at nearly the same frequency than the actual playing tone it is like producing another square wave on top of the real one..
The result is that it creates a slow phase shift producing a real variation of the resulting square duty cycle
something that the chip itself cannot do without this trick.

All of the above make it pretty unique indeed.

Thanks for the explanation! I'm going to go with the character coming from the pitch resolution, because SMS chiptunes have such a unique character to them, you've only gotta picture the music to Alex Kidd (which has been seared into my brain) to hear that kind of slightly off sing-song quality it has.

Would I be right in assuming that if I'm using the fake-PWM and fake Sync effects, it would be wise to oversample the synth?

[Aly James]

Would I be right in assuming that if I'm using the fake-PWM and fake Sync effects, it would be wise to oversample the synth?

For SPSG VST, The Tone Generator is band-limited so as the AY emulated Hard Buzzers.
On the other hand the Timer interrupt routine that is written on the volume register for the "SID Voice" effect isn't, so you might benefit from a higher sample rate in this particular case.

For a Higher FM quality in FMDrive you will benefit from working or switching your project to 96Khz before exporting to 48 or 44.1 because the FM process itself will have a broader range, especially for all the generated FM harmonics. The YM2612 chip on the Megadrive was outputting at ~53KHz.

[Kayster82]

Talking about tone sound... isn't the FMDRIVE pure sine wave unique? I mean it has some harmonics in it even in clean mode (HD) no modulation, at least in my 1.24 I can hear some high harmonics on lower midi notes.
The Megadrive 1 Mode (YM2612 Ladder) is impressive, if you set a long release (I have double check that in Audacity) the sine wave is slowly turning into a square wave as the sound dies.

[Aly James]

Talking about tone sound... isn't the FMDRIVE pure sine wave unique? I mean it has some harmonics in it even in clean mode (HD) no modulation, at least in my 1.24 I can hear some high harmonics on lower midi notes.
The Megadrive 1 Mode (YM2612 Ladder) is impressive, if you set a long release (I have double check that in Audacity) the sine wave is slowly turning into a square wave as the sound dies.

Indeed, the Sine wave is calculated from the first (rising) quarter...
Explanation:
A ROM inside the YM2612 contain a log-sin waveform table, containing one quarter of a sine wave, 256 samples long, the resulting waveform is then 1024 sample long.
This as been set to default since 1.24 this is why you can hear harmonics even in High Quality Mode.

The Ladder effect which happens on the MD1 Mode is an emulated bug on the discrete YM2612 built in DAC, which had been removed by Yamaha on later MD model 2 which feature an ASIC (Application-Specific Integrated Circuit , basically a chip with many function inside) with the FM chip built in (based on an YM3438).

This bug, which make the sound so particular, happens because of an analog problem handling the sign bit (Analog value matrix of the DAC is 8x32, 8bit (256 levels), 9th bit is a sign...)
When it crosses zero to go negative, there is some voltage finding its way.. it happens even if a channel is muted on the real hardware.
This analog DC voltage happening only on the negative part, slightly offset, produce a square wave at low volume that is mixed with the actual output.
As it is mixed with the output, logically, it becomes more prominent as the sound approach max attenuation. You also hear a change in tone at the very end because the pulse width of the square wave is reducing...
Look at this graphic, better than words



And this is what you can observe on an operator with very quick decay and no modulation,
so you are almost right,
it ends up being almost a square wave



Hope that helps!

[Kayster82]

Some really good YM2612 informations here.
Thanks a lot for your great work, it really is much appreciated!

What you call the "ladder effect" is clearly present in one of your FMDRIVE demo, the Pandora Box track at soundcloud, it sounds wicked :
https://soundcloud.com/alyjameslab/pand ... mdrive-vst