Re: TI-H: Radio Link


[Prev][Next][Index][Thread]

Re: TI-H: Radio Link




I know it is bryan...  :)  The lower end microcontrollers have a smaller
address space (or something like that) which means a faster speed...
Sometimes when you are checking or setting bits, you need a nop after each
sbi/cbi/sbic/sbis and things like that...  If you don't, sometimes you get
a false reading becuase its going tooo fast...

Someone told me this on the AVR technical mailing list...  His name I
forgot but he made these sticks taht look like 30pin SIMMS and program
chips...  Something like the SlimStick.  Also, David Vanhorn told me about
this problem too...

>Actually, the TI protocol is asynchronous... in theory. However
>since all programs are fundamentally functions over time (I.E.
>one thing per cycle, repeating as time progresses) nothing
>is truly asynchronous, including the TI protocol. The AVR
>was so fast that it was waiting for a response from the calculator
>before the calculator had time to recognize the AVR's response,
>and thus was bypassing critical loops in the TI byte send/receive
>routines. Even at 1 MHz I've had to add several nops
>just to give the calculator time to change its I/O state.

Microchip didn't lie.  That number came from Atmel and my Ti92...
1/10000000 is 200^10*-9 or 200nsec...

>I'm betting Microchip lied a bit about those 200ns numbers,
>as the AVR seems to be at least a dozen times faster than the
>PIC. I suppose it depends on a lot of things though.
>
>Bryan
>
>Grant Stockly wrote:
>>
>> >You reversed the per-cycle times for the AVR and PIC, by the way.
>> >The PIC is about 200ns, the AVR is around 62.5. Time = 1 / Frequency.
>>
>> I know...  at 10MHz you get 200ns...  at 16MHz you get 62.5...  Your doc
>> said you were using a 10MHz crystal and so your routines were
>> misconfigured...