Re: General 6809 card for 6502 machines

yes effectively, the delay is not related to the clock but the chip 
itself...


Le 19/10/2017 à 22:54, Mia Magnusson a écrit :
> Den Thu, 19 Oct 2017 22:28:10 +0200 skrev didier derny
> <didier@aida.org>:
>> the liber809 is built specially for the atari 800xl  (and some other
>> similar)
>>
>> the 6502 used on atari is not the standard one   3 pins are differents
>> RW is not on pin 34 but on pin  36  (on sally  34 is NC)
>> there is a HALT signal on pin 35  (nc on a 6502)
>> the clock used on the atari is 1.75Mhz   (not 1mhz)
>>
>> to produce Q  the liber809 is using a digital delay line  calculated
>> for a 1.75Mhz
>> so if your clock is 1Mhz Q arrives too late
> Woudn't it rather arrive early?
>
> I.e. to delay 1,75MHz one quarter cycle you would have a delay of about
> 143 ns but for 1MHz it would be 250 ns. With 143ns at 1MHz the four
> phases of the 6809 clock would not be equally long, but each would be
> within spec for the timing requirements for the 6809.
>   
>> in the past I had a project to put a 6809 on a 6502 socket
>> or a 6512 on a 6809 socket :)
>> but never had time
>>
>> I'm trying to check on how I could make that type of adaptors but
>> compatible with several clocks
> A PLL would probably be the only way if you want to be able to use a
> 1MHz 6809 in a 1MHz 6502 machine and with the same hardware use a
> faster 6809 in a equally faster 6502 machine. But it takes some time
> for a PLL to synchronise to it's clock so it might not work in a
> application which switches speed like for example the C128.
>
> But how many different 6502 machines could this be relevant for? As
> there is already the liber809 for Atari we could perhaps ignore Atari.
> That probably leaves us with many machines running at about 1MHz, some
> running at 2MHz (C128 if we emulate $0/$1) and also the BBC computer. I
> don't know of any computers with 6502 that runs faster than 2MHz, but I
> haven't looked for them.
>
> Another way to see it is which speeds 6809's are available at, and just
> make the delay fit each possible CPU speed. By looking at Wikipedia the
> Motorola 6809 were made for 1 and 2 MHz while Hitachi 6309 were made in
> another version running at 3,58MHz (and according to Wikipedia usually
> possible to overclock to 5MHz).
>
> Someone needs to read the data sheet to see if I'm making a correct
> guess that the phases don't have to be equally long. If the as I
> suspect don't, then the hardware could just be set for the kind of CPU
> that's inserted, using a jumper, and then working fine at any clock
> frequency that the specific CPU can handle.
>
> It seems like standard 6502's were made for 1, 2 and 3MHz while the
> 65C02 can run up to 14 (!) MHz. So it's a question of which speeds to
> support.
>


       Message was sent through the cbm-hackers mailing list