Re: When to use a 74LS245 to buffer the data bus?

[Rubywand <rubywand@swbell.net>]

Glenn Jones writes ...
> 
 ....
> 
> Yves, I am interfacing an ethernet module to the Apple2 bus. The module has
> two chips on it...one is a custom chip W3100A and the other is a RTL8201L
> PHYceiver chip. I am not sure how many TTL inputs this would be equivilent
> too.
> 
> The concept is the same as the TFE ethernet board that Adam Dunkels created
> except that the Wiznet chip is amost 1/4 the price and has a TCP/IP stack
> built into the 3100A.
> 
> I have a prototype wired up without the 74LS245 but IIgs hardware reference
> pg 180 talks about loading and driving rules and this has got me thinking
> htat maybe I should be using a 74LS245 for the data bus.
> 
> If you are interested, you can find out more information about my project at
> http://members.rogers.com/rg.jones/AppleII_ethernet%20_project.htm
> 
 ....

     Usually, you would not use a bus transceiver unless there were some reason
to isolate the peripheral bus-- like the peripheral (local) data bus is used by
a controller on the board to talk with other chips on the board.

     If you have peripheral board outputs which are seperate from inputs,
having an 'LS244 octal tri-state buffer (or a couple of LS365's or LS125's) on
the peripheral data outputs is pretty standard practice. It guarantees the
peripheral can drive the system bus and can improve peripheral response (when
peripheral chip outputs can be tied to 'enabled').

     If your peripheral IC's handle data only through a bidirectional bus, you
can check IC documentation to see if adding a transceiver or pairs of buffers
is a typical option supported by the chip(s) and when adding buffers is
recommended.

     Otherwise, it may be best to leave space for a few extra chips on the
prototype and try out the board with no special buffering. 



Rubywand