| ??? 02/25/09 00:32 Modified: 02/25/09 00:58 Read: times |
#162788 - The PLL approach isn't that hopeless... Responding to: ???'s previous message |
Per said:
A PLL is very good when you have slowly varying frequencies and need to get higher update frequencies than you would get by waiting a full period. As an example, a PLL is idealy suited to measure the mains frequency at a resolution of 0.01Hz, without needing to wait 100 seconds. Per said:
But I bet you had quite stable signals so your PLL had the time to follow the drift? That's not that hopeless as it might look at the first sight. If you display the speed at a resolution of 1MPH, then allowable tolerance at a speed of 1MPH is +-50%. Or would you display the vehicle's speed more precise than 1MPH? Per said:
In this case, It sounds more like the speedometer for a car, where you may go from zero to 50% of the speedometer range in less than 10 seconds. When the speed is rapidely changing, there's no need for an accurate measurement either, especially if the display reading is updated only twice per second. Think only, how much the speed can change in 0.5 seconds. Also, much of the error would average out due to the higher PLL frequency. And if you finally decide to define a minimum speed of 10MPH instead of 1MPH, which would be much more realistic, then the PLL option begins to look interesting. Carefully setting the capture range, realizing two time constants for rough and fine tuning and other design details should do the trick. I'm intelligent enough to realize, though, that a modern microcontroller can offer way more sophisticated solutions. Kai |
