| ??? 05/04/06 03:05 Read: times |
#115516 - Cures... Responding to: ???'s previous message |
Russell said:
If you've got low level signals, either amplify at the sending end or use differential signalling - this can mean just using a differential input using a ground wire and signal wire from the source unit. Ground noise should be cancelled out. Yes, "should". But unfortunately, due to differencies in the source impedances of signal wire line and ground wire line common mode rejection by this methode is rather poor in many cases. Also, often this ground wire isn't referenced to the true ground from where the signal wire leaves, but to some different ground, often contaminated by different ground noise. Only a true symmetrical signal routing will do the trick! Russell said:
Some signals run in shielded cables, the shield is only connected at the ECU end. Which makes the shielding inneffective at high frequency interference... Why not learning from the PROFIBUS? Shield is connected at both ends to chassis ground, but signal wires are optoisolated, which gives you both: Perfect shielding performance even at highest frequencies by providing an entirely closed Faraday cage, and at the same time removing the eventual interference resulting from ground loops caused by the double grounding of shield. By the way, signal grounds at each end of bus cable are soft grounded, by connecting signal ground to chassis ground via a cap. And, signal is symmetrical, is RS485! Kai |
