dad@cs.brown.edu (David A. Durfee) (01/18/91)
I have recently gotten involved in the specification of a thin ethernet installation. A lose end still remains which I would like to clear up. When I was involved in thick ethernet installations a few years back, I recall a recommendation to keep cable lengths in multiples of one wavelength (approx. 76ft) for best results. In fact, I notice that thick ethernet premade cables continue to be sold only in multiples of a wavelength. A. Since thin ethernet is effectively the same technology, wouldn't the same recommendation apply? B. Where does this recommendation come from? I understand so-called "microwave theory" concerning reflected waves due to impedence mismatches to some degree. My question is what problem is being solved here. It was my understanding that even multiples of quarter wavelengths would reduce problems do to the insertion of a cable which has a different characteristic impedence than the rest of the net. (that is because the reflections from the two end of the cable would cancel). BUT reflections due to poor connectors/tees would have less effect if multiples of odd quarter wavelengths are used (I don't recall why). These requirements appear to conflict. Any of you out there really know the answer? I've called the tech support of several companies involved in making ethernet equipment but nobody was able to shed light on my questions.