| ??? 04/26/07 13:51 Read: times |
#138017 - absolutely NOT Responding to: ???'s previous message |
On the Rx side this should be just fine.
absolutely NOT The 'window' is so small that the likelyhood of a small timing variation in the 'scheduling' (I know of no 'scheduler' that is precise within nanoseconds) would make you miss getting a character. The difference between a 'slightly off' 'tick' of the 'scheduler' and a 'slightly off' "other end of the serial cable" can, easily be zero or worse - negative. Trusting that something relying on a 4% difference will always work is more related to hope than to reality. Again, you can, of course, hope it will work, if you are one of those that consider 'testing' and 'hoping' a design method. It may very well work for a 'test' but the likelyhood that a 4% difference disappearing someday, somewhere (Dr. Z) is HUGE. Erik |
| Topic | Author | Date |
| Serial transmission in scheduller instead of int | 01/01/70 00:00 | |
| danger, maybe, loss of efficiency YES ABSOLUTELY!! | 01/01/70 00:00 | |
| avoiding interrupts means... | 01/01/70 00:00 | |
| Scheduller.... | 01/01/70 00:00 | |
| there is no such thing | 01/01/70 00:00 | |
| Limitation API | 01/01/70 00:00 | |
| API? | 01/01/70 00:00 | |
| Yes ASIC design fixed... | 01/01/70 00:00 | |
| debug possibilities? | 01/01/70 00:00 | |
Monitor is nice idea | 01/01/70 00:00 | |
| 1ms vs 1.04ms | 01/01/70 00:00 | |
| absolutely NOT | 01/01/70 00:00 | |
| ok but 1 byte is transmitted together | 01/01/70 00:00 | |
| if it does, then you need to KISS | 01/01/70 00:00 | |
| I took the 1ms processing tick for granted... | 01/01/70 00:00 | |
| Thanks for the inputs... | 01/01/70 00:00 |