| ??? 02/25/02 20:50 Read: times |
#20139 - RE: FIR Filter |
Move CPU intensive math to assembly and optimize it by careful manipulation with registers. Use as much as possible the XCH instruction since it can replace 2 MOVs, but the variable location is then dynamic so a good 8051 simulator will be of great help for the debugging. Do not use floating point but fixed point instead. Put as much as possible to the coefficient data table and use it in your math equations.
I have implemented IIF fileter of 2nd order for realtime single frequency detection. Detection time is 60ms. Sampling frequency is 5200Hz/8bit running 16MHz. Special 24bit x 16bit fixed point multiplication has been developed, while constant part of equation has been evaluated and moved to the table. So simple index of the 16bit coefficient data table defines a detection frequency and energy treshold level. Franc |
| Topic | Author | Date |
| FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
| RE: FIR Filter | 01/01/70 00:00 | |
RE: FIR Filter | 01/01/70 00:00 |



