Re: Blurry picture

>>> Another way I have been thinking about is to rig up a +12V regulator
>>> and use a 6569R5 in a board made for the 8565R2 and supply pin 13  
>>> with
>>> +12V. Shouldn't be too hard, the voltage doubler circuit is still
>>> there, just needs a 7812 plus some capacitors. If we still don't see
>>> grey dots then, this suggests it's something with the 856x-VIC.
>>
>> Before doing such major surgery,
>
> That's not really major surgery, it would be fully reversible and  
> mean only one bent pin (13) on the 6569R5 and some open wiring. I  
> have a demo that will produce the grey dots reliably.

It's major surgery in the sense that it changes lots of things, since
the VIC generates most of the relevant signals.  It will not prove the
problem is caused by the 85xx VIC; it can merely prove that the problem
does not manifest itself with the 65xx VIC.

And it won't tell us much as to _why_.  Just graphing the bus signals
will tell us much more: it will show where the important differences
are, so we can focus on that :-)

>> maybe we should get some more data first:
>> get scope traces of all the relevant signals (clock, VIC #CS, R/ 
>> #W, #AEC,
>> #RAS, #CAS, some A and D pins) on various different machines. I  
>> couldn't
>> find a single bus trace online.
>
> I can see if I can get around to do some of those.

Excellent, thanks!

> Which signal in relation to what interests you most?

How many channels do you have?  If only two, let's start with
Phi0 and VIC #CS?  If you have more, throw any of dot clock,
#AEC, CPU R/#W, #RAS, #CAS, or maybe CPU data and address into
the mix.  I'd like to get a full picture of the timings on the
bus.  It's hard to say which signals are more important without
seeing them first, but roughly that order I'd say.

Can you test both a "dots" machine and a "no dots" machine?

>>> I could, but the C64C I have access to at the moment (the others are
>>> stored away) doesn't show the vertical bars. Most C64C only show  
>>> them
>>> somewhat, they are really visible on a C128 though.
>>
>> None of my C64C do (and no dots either!)
>
> There is a problem with using the term C64C. I own one that says  
> that on the bottom, but still contains a 250466 board, that means  
> 65xx-parts and 82S100 PLA. It's better to use the board part  
> numbers. The board with the large PLA is the 250469.

Yeah, like I'll remember those numbers :-P

Sure you're right that it's inexact; just assume when we say "C64C" we
mean "85xx machine".  The 64-pin chip is not a PLA FWIW, if we're in
pedantic mode anyway ;-)


Segher


       Message was sent through the cbm-hackers mailing list