[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: Flight Computers

On Wed, 01 May 2002 21:44:55  
 Phil Karn wrote:
>>The SHARC had so much horsepower that our analysis of the processor
>>bandwidth loading is that with all of the above running we used about
>>15% of the processor bandwidth.  Most of the time was spent waiting
>>for the next task switch.
>Most interesting. What kind of response deadlines did you have, and
>did you ever have any problems with excessive interrupt latencies?
>As far as I can tell, the distinguishing feature of a "real-time" OS
>is that it bounds the interrupt latencies associated with protecting
>critical sections of code in the kernel. But as you say, just using a
>faster CPU with a "conventional" OS may work just as well.

The only interrupts that we used were associated with the 16550 UART.   We used the recieve and transmit buffers to increase the time between interrupts.  Also, a good Interrupt Service Routine is very small, no more than a few instructions.   If I remember right what we did is move the data into or out of the buffer (Only takes a microsecond or two to do) and set a flag for the task that handled the data.

Now the SHARC is not that fast of a processor in terms of MHz.  Only runs at 40MHz but it is really fast in terms of cycle efficiency.  ALL instructions run in a single clock cycle and this makes life really nice and pretty deterministic as well.

For a spacecraft there is not much that goes on that needs to happen in timespans of less than 50 milliseconds.  Specialized communications applications at the data rates that amateur satellites use is still not that demanding unless you want to do software radio and that takes a DSP anyway.


>>This is a reason that I am a very strong advocate of doing whatever it
>>takes in radiation shielding to protect the processors.  This brings
>>so much greater reliability to the system than trying to use the last
>>ounce of power from an ancient processor.
>Agreed. Problem is, the energy distribution of charged particles in
>the Van Allen belts has a very long tail. It's easy to stop nearly all
>of the particles with relatively little shielding, but those really
>high energy particles are almost impossible to stop. So you need
>rad-hard parts and error-correcting memory. Or you can stay in LEO,
>where the radiation environment is a lot more benign.

Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org