Re: Blurry picture

From: Segher Boessenkool <segher_at_kernel.crashing.org>
Date: Tue, 15 Nov 2011 03:47:28 +0100
Message-Id: <159B140F-9D9D-41C2-BAD9-EA12E73B2B5E@kernel.crashing.org>
>> Sorry for the mistake. I think that may have been an earlier  
>> design I was remembering from Al Charpentier's notes about having  
>> to add a PLL to keep them in sync.
>
> That was the VIC-II and all its derivates. In the end the decision  
> to supply dot clock and color clock to the VIC-II independantly  
> resulted in the need for the PLL and in the end gave us the little  
> MOS 8701. Even though MOS reworked the VIC for the C128, the 8701  
> is still present there.

The MAX machine used dual crystal oscillators (in discrete parts).

The C64 divided the colour clock (~18 resp. ~14 MHz for PAL/NTSC) by
9 resp. 7, then by 2, and then fed it to an MC4044 PLL taking its
feedback from Phi0 (so that the PLL multiplies frequency by 8).

The C64C uses the 8701 instead.  It doubles the frequency (by doing
a XOR of the clock with itself delayed a bit), and then divides by
4.5 resp. 3.5 (the duty cycle isn't 50/50).  There also is quite a
bit of unused circuitry on the chip, bonded out to undocumented pins;
perhaps there is (an attempt at) a PLL in there?  The "pass it through
some slowish inverters" fixed delay thing is ugly (but it works ;-) )

I don't think it would have been possible to make an 8701 in the
older 65xx NMOS process, 36MHz is too much.

The VIC (6560/6561) uses circuitry quite similar to the 8701 (but
much simpler).  It leaves out the doubling step, the dot clock is
only about 4MHz.  PAL/NTSC isn't switched by a pin, but by changing
the metal layer (break/make some traces), this simplifies things
further.

Are there die photos of the VIC-IIe somewhere?  It would be interesting
to see what changed on there (what exactly changed functionally,  
anyway?)


Segher


       Message was sent through the cbm-hackers mailing list
Received on 2011-11-15 03:00:05

Archive generated by hypermail 2.2.0.