Re: HMOS VIC sparkles

On 7/27/19 1:03 AM, Segher Boessenkool wrote:
> On Sat, Jul 27, 2019 at 12:13:14AM +0200, silverdr_at_wfmh.org.pl wrote:
>>> On 2018-03-15, at 14:24, Segher Boessenkool <segher_at_kernel.crashing.org> wrote:
>>> On later C64 machines the CS line to VIC goes active earlier, causing
>>> the sparkles.  A little cap (from CS to ground) solves this.  There is
>>> no difference between the 65xx and 85xx VIC here.
>>>
>>> (You get the sparkles because the internal data bus on the VIC is
>>> precharged; if CS is too early you get that value instead of the value
>>> on the data bus written to the colour regs, for a little while that is,
>>> half a pixel or so.  And the precharged value is 0b1111, light grey).
>>
>> .. I just had a discussion about those sparkles with a friend and couldn't authoritatively answer. Because if the above is true, then the sparkles should be visible regardless of which VIC is being used, right? You mention yourself that "there is no differnce between 65xx and 85xx here". Or is this 0b1111 only inherent to HMOS VICs? The reason I ask is because I recall "fixing" the sparkles in one machine decades ago by doing some wiring and running 6569 in the narrow board. I don't think I delayed the CS (back then I surely didn't have any idea about what was causing them to appear).
> 
> I don't think there is a difference between 65xx and 85xx VICs here.
> The main difference is these are used on different boards, with different
> PLA or equivalent circuitry, which has a much bigger timing difference
> than the different NMOS chip generations do (or actual changes on the
> chip layout, which there are a few, although not many).

We had a discussion about this years ago and I supplied a lot of scope 
plots (I still have them). Your conclusion was that the HMOS-II VIC is 
faster so you get to see the internally precharged lines.

TED has the same sparkles BTW, but they are a different color.

  Gerrit