| ??? 08/27/08 11:59 Read: times Msg Score: +1 +1 Informative |
#157792 - Sample a longer run at high frequency and analyse Responding to: ???'s previous message |
I wouldn't call that oversampling. If you took one sample every 0.4ms then you really did sample at 2.5kHz.
In your case, you sample at a rate of 20Hz, but you try to reduce noise by averaging a quick burst of 125 reads. Are your samples so noisy that you need 125 readings to produce a single output sample? Your bandwidth is limited by your 20Hz sampling rate, which makes you unable to measure any frequencies above 10Hz. Do you have the bandwidth to make a longer log at 2500Hz sampling rate? If so, I would recommend that you do that, and then analyze that curve and look at the frequencies and base any filtering on the results of that analysis. |
| Topic | Author | Date |
| Integrate decimate... | 01/01/70 00:00 | |
| Strength reduction | 01/01/70 00:00 | |
| >> = Divide and << = Multiply | 01/01/70 00:00 | |
| Decimating | 01/01/70 00:00 | |
| I have also used this approach | 01/01/70 00:00 | |
| Thought process in frequency domain.. | 01/01/70 00:00 | |
| Low-pass filter | 01/01/70 00:00 | |
| Low pass filter selection... | 01/01/70 00:00 | |
| Sample a longer run at high frequency and analyse | 01/01/70 00:00 | |
| Good suggestion | 01/01/70 00:00 | |
| For your application... | 01/01/70 00:00 | |
| Rooling average concept... | 01/01/70 00:00 | |
| Moving Average Filter = Rolling Average Filter | 01/01/70 00:00 | |
| General concept vs. specific algorithm. | 01/01/70 00:00 | |
| Correlated and uncorrelated noise | 01/01/70 00:00 | |
| Your language ... | 01/01/70 00:00 | |
| 'Synchronous'? | 01/01/70 00:00 | |
| Similar but not identical meaning | 01/01/70 00:00 | |
Yes, sounds waayyy better... | 01/01/70 00:00 |