| ??? 04/01/08 15:12 Read: times |
#152834 - adaptive delta ... ? Responding to: ???'s previous message |
You can potentially help yourself somewhat by maintaining a "moving average" of the deltas (changes) on your readings.
I don't know the nature of the signal or the environment, so I can't make any quantitative suggestions, but, if you maintain a moving average of the deltas, and add only the positive or negative moving average in any case wherein the current delta exceeds the moving average, the result just might suit you. If the delta is negative, you add the negative of the moving average, and if it is positive you add the positive moving average, but only in cases where the delta itself exceeds the moving average. You must still incorporate the entire delta into the moving average so you properly track the deltas, but average them in firmware. If the "noise" is random, then it should cancel out. Depending on the nature of your input signal, this may work out better than simple averaging. RE |
| Topic | Author | Date |
| Adding hysteresis on top of my sampling | 01/01/70 00:00 | |
| my method | 01/01/70 00:00 | |
| For this to work... | 01/01/70 00:00 | |
| adaptive delta ... ? | 01/01/70 00:00 | |
| All data is keep in eeprom | 01/01/70 00:00 | |
| What does the oscilloscope tell you? | 01/01/70 00:00 | |
| not a good ides | 01/01/70 00:00 | |
| Median filter ? | 01/01/70 00:00 | |
| are you using an RC low pass before the ADC? | 01/01/70 00:00 | |
| check here | 01/01/70 00:00 | |
Found a minor issue in hardware...not dealbreaker | 01/01/70 00:00 | |
| Low Pass Filter in Software | 01/01/70 00:00 |