SID Emulation

[LunaRecording]

I absolutely love CCS64 (I've used them all, nothing works nearly as well). I have been a C-64 user since I was 4 years old, and grew up admiring many of the amazing sounds that were squeezed out of that SID. It inspires me to this day, and I use CCS64 for my own music (I know, nothing is as good as a real SID, but it's simply too noisy for my taste - even a HardSID picks up a lot of RFI/EMI which is rather audible in any recording, even with a noise gate).

I'll get back to my real question - or, rather, suggestion.

I've noticed that the emulation sounds very good for both 6581 and 8580 modes. However, both modes tend to have a lot of DC bias within the emulator. This results in a lack of headroom (worse s/n ratio), as well as problems with annoying "pops" during signal processing. Of course any wave editor can remove the DC bias to some degree, but it only centers the wave; even when it is centered, any quiet portions still have a DC bias.

I was wondering if this was actually typical of the 6581 and 8580 chips, meaning that the DC bias is present in both. If so, I was wondering if fixing the DC bias in the emulator would result in a loss of audio quality during emulation.

Thank you again Mr. Sundell for creating such an outstanding emulator. Keep up the good work.

[LocalH]

I am under the impression that the DC bias is integral to the methods used to play 4-bit samples through the volume register. That's the whole reason why 6581's are better than 8580's for 4-bit digis without some special tricks - in the 8580, CBM lessened the DC offset. In one of the later issues of C=Hacking netmag, there was a *very* extensive discussion of why digi's in general work, and different methods of generating them.

[LunaRecording]

I understand that there is a certain amount of DC bias associated with the SID (especially the 658x versions), and that this was used to play 4-bit samples through the volume register. However, it appears that the volume register isn't the only thing that can change the amount of DC bias. For instance, if I use a sweeping low-pass filter with CCS64 the DC bias changes when changing the cutoff frequency and resonance. I was mainly wondering if this was authentic to the SID's behavior.

[zaphod]

VICE (with ReSid engine on) has much more authenntic SID emulation. If it does it in VICE, it' realy how the sid would do it.

That said, every real SID is different, especially in the filter.