Re: 6702 chip

On Wed, 5 Sep 2012, Rhialto wrote:

> On Wed 05 Sep 2012 at 06:38:29 +0000, William Levak wrote:
>> INTERNAL OPERATION
>>
>> 1. When data is written to the 6702, the low bit is AND'ed with the
>>    Even/Odd register. If the result is zero, the low bit is inverted and
>>    written to the Even/Odd register, and the 6702 waits for more input. If
>>    the result is one, operation continues below.
>
> The way I have formulated this for myself is that the even/odd register
> is described as "the required low bit of the next number". The
> even/odd register is compared to the low bit, and only when it is equal,
> operation continues. Then, only when it is 1, it does steps 2...5.
> A separate step 6 would be to toggle the even/odd bit.

Logically it is the same, but the question is:  how does the bit get set. 
The simplest implementation is to use an inverting driver to copy it from 
the input register.

For step 6: toggling the bit is the same as using an inverting driver.  If 
the inverting driver already exists, then it only needs to be gated, thus 
saving on circuitry.

> Your description cleverly combines the three cases where no output is
> changed because
>
> (a) an even number is expected and given (even/odd bit = 0, low bit = 0,
>    so the AND is 0, and the even/odd bit is changed to 1); or
> (b) an even number is expected but not given (even/odd bit = 0, low bit = 1,
>    so the AND is 0, and the even/odd bit is kept at 0); or
> (c) an odd number is expected but not given (even/odd bit = 1, low bit = 0,
>    so the AND is 0, and the even/odd bit is kept at 1)
>
> but it needs a case distinction like that to realise that one
> description is equivalent to the other.
>
>> NOTE
>>
>> The original SuperPET contains sockets for four 6702 daughter boards.
>
> Perhaps that is why the output value is visible on 4 address locations.
> The addresses influenced by the circuitry around the 6702 are more than
> just those; all of "E F xxx0 xxxx" are affected, but the ones that are
> not EF00 .. EF03 all return FF (instead of the EF of the unconnected
> space). Apparently part of the address decoding is done inside the 6702.

The input and output appear at 4 loactions.  The address is incompletely 
decoded.  Address space $EFE4-EFEF is available, enough for 3 more 6702s. 
According to the schematic, address bits 0 and 1 do not even go to the 
daughter board.  Bits 2 and 3 go directly to the 6702. These could be 
programmable.

> -Olaf.
> -- 
> ___ Olaf 'Rhialto' Seibert  -- There's no point being grown-up if you
> \X/ rhialto/at/xs4all.nl    -- can't be childish sometimes. -The 4th Doctor
>
>       Message was sent through the cbm-hackers mailing list
>

wlevak@sdf.lonestar.org
SDF Public Access UNIX System - http://sdf.lonestar.org

       Message was sent through the cbm-hackers mailing list