Re: Playing samples on both new and old SIDs

From: Hársfalvi Levente (
Date: 2000-07-27 19:23:40


Carlsson, Anders wrote:

> Speaking of the VIC oscillators, they seem very unreliable, and
> the square waves seems to be produced independent on the channels;
> i.e. two notes on different channels might clear out eachother
> partly (different waveform on output) or completely (no output).
> I have noticed this very recently while writing music for an
> upcoming VIC-20 megademo. First I thought there was a way to
> time (control) this and use it to create more waveforms, but I
> dunno.

After all, I'd seriously suspect that the oscillators of the VIC and the
TED should be almost similar inside, since they seem to have very
similar features.

- Only squarewave, and a 'semi'-whitenoise waveform
- The formulas of the osc's are very similar; once I experimented with
the osc's and found 'f=Phi2/256/(128-(($900a+1)&127))',
f=Phi2/128/(128-(($900a+1)&127))' and so on for the VIC-I, and
'f=Phi/8/(1024-((frqreg+1)&1023))' for the Plus/4.
- The very same design, the same tasks

I don't think that all this could be by accident. For example, I'd do so
if I had to implement variable volume levels and didn't want to fiddle
with a hybrid design (R-ladders).

Anders: you probably use freqtab based on 127, 63, 95, 111, 119 (octave
base note frequencies - these values sure give exactly one octave
distances) - I mean the table that doesn't get out of tune with higher
octaves. If you set the lowest oscillator to 63, and the middle one to
127, you'll also get the same frq on both, am I right?

Richard, your results, as usual, lite a lamp in my head again :-). You
then probably can deduct the answer for the well known feature of the
TED - by selecting some very high, but not the same frequencies for the
TED oscillators, it is possible to get finer output voltage level steps
that can be used for playing digi sounds, achieving almost 5 bit quality
('logarithmic' scale :-), finer steps at the start). I can give you an
example code or you can also examine my SID music converter routines
(most of them make use of this feature).

It's also notable that the TED oscillators get synced together somehow
when setting very high frequency - probably, when writing the highest
frq data, the osc has only two states that it can be in. Or whatever. Or
some quirk in the TED snd source design...

> Possibly there is a difference in resolution between NTSC and PAL
> VICs (I never heard a real NTSC VIC), and I still don't know if
> there is any difference between VICs labeled 6561-101 and 6561E.
> The latter is what I got in my machine.

Also, a lamp... never heard of anyone else seen a 6561E. I have one (but
some parts should be broken in it :-( ).

I don't know if there is a difference in the resolution of the PAL and
NTSC VIC's, but one thing is for sure: the PAL VIC-I's implementation of
the PAL color encoding could be called anything but something proper.
PAL's spirit is in the phase-inversion of every odd and even scanlines
color signal, but the 6561 doesn't seem to do anything like that at all.
Once I wrote a testprogram that filled the screen with rasters, turning
on a color in every odd and black on every even lines. Switching between
'parity', I got very different colors. (I could only fill the screen
with a continuous tone if I set different color code for even and odd
scanlines - yes, it was possible anyway :-)).

Just turn on the Vic-20, and wonder why the default cyan on the borders
seems to be so 'jerky' - it's like every odd and even scanlines have
slightly different tone.

Also, no wonder why my color TV set didn't recognize a 'valid' PAL
signal to show the VIC's screen in colors... :-(

This message was sent through the cbm-hackers mailing list.
To unsubscribe: echo unsubscribe | mail

Archive generated by hypermail 2.1.1.