EPROM Switch

[einarp]

>EX for to switch two 27128 chip you must have a 27256 and controll the pin
>for the MSB

Should be synchronized with MPU clock to avoid switching in the middle of
an instruction or data transfer. And as someone said: the program must be 
exactly the same.

>ignition timing, etc.  Also it should be done quick enough so that the micro
>does not miss an external data fetch (anyone know what the micro will do in
>this case??  will it hang or retry? ).

Quick enough _and_ outside an access cycle. If the access is to data, it would
probably result in a tiny hiccup at worst. If it happens within a code fetch,
it will with great probability set the MPU off track resulting in a hangup.
For most sensibly designed controllers that again will trigger a watchdog and
a total restart of the controller. Not the nice way to do it however.

What signal you will need to synchronize with depends on the MPU used, and the
hardware design of the system. For a Motorola 6800 family MPU the E clock 
should 
be suitable. In many designs you can use the OE or CS of the EPROM. Just a 
switch
won't do. The switch noise if nothing else will stop that from working in a 
running
system.

Other problems may occur with an adaptive system that bases it's control upon
previous experience. It may need to "go to scool" for a while after switching.
It may even counteract your changes.

Talking about hangup of controller: I had a car that often experienced a 
hiccup.
It felt like the ignition cut totally for less than a second. But it always 
happened
at the same place (within appx. 100m)! I blamed everything from a bad 
connector to my
mobile phone until I found that it was connected with this specific location.

Einar

--
einarp at ade.no  ( Maserati Biturbo Spyder )