Re: CIA old/new?

On 1/5/21 6:54 PM, smf wrote:
> On 05/01/2021 12:22, Gerrit Heitsch wrote:
> 
>> Unless you use the color clock or a multiple of it as dot clock, you
>> will have dot crawl. And if it's not a whole number multiple, you need
>> a PLL or another circuit that can accomplish the same.
>>
> Right, they changed the dot clock late and then added the PLL.
> Eventually the PLL got removed. It's kludge city.

Well, yes, but the 8701 does the same thing, creating dot clock an color 
clock from a single crystal. And the only reason, as far as I know, they 
went with the 8701 was the 74LS629, that chip could only be gotten from 
TI and Commodore had a bit of a history with TI from the calculator days.


>> Manufacturing issues aside, the chip designs were OK for back then.
>> And the fact and a lot of them still work more than 30 years later
>> does show their manufacturing quality wasn't too bad once they got the
>> process figured out.
> 
> The industry is completely different now, they didn't have the tools
> back then etc etc.

Especially since MOS didn't have the latest machines even back then.


> Stringing capacitors onto lines that shouldn't have them etc is not a
> good sign :D

That's the way it was done back then when you needed to adjust a signal 
timing. Take a look at how the ZX Spectrum generates /RAS and /CAS for 
32 KB DRAM that is not controlled by the ULA.



>> TED and 7501/8501 had problems if made in 1984. Versions made later
>> didn't die as easily.
>>
> Quite what they were doing making  TED in 1987 is anyones guess...

The latest datecodes I have seen on TED are from 1986, but there was a 
8360R3 it seems.


>> Compared to what? Which other soundchip with the capabilities of the
>> 6581 existed back then?
>>
> Why does that matter? The first design isn't automatically good.

Yes, but even that first design was better than anything else availbable 
back then.

  Gerrit