On Thu, Jul 18, 2013 at 7:23 PM, Denny Esterline wrot= e: > It seems there are several issues here. Given that this is commercial rai= l > and human lives are at stake (at least here in the USA) the dominant fact= or > should/would be safety and liability. The cost of the inevitable lawsuit = is > much, MUCH greater than the cost of this system. Given that there are > available, existing, accepted mechanisms for detecting trains (e.g. > crossing light controls), I'd be very reluctant to touch this. Assuming > that either India is different or that this is an intellectual exercise..= ... > I am pretty sure that current solutions also will malfunction at some cases/exceptions as well. Direct remote sensing or local sensing and sending the data to a remote location ; both are based on physical phenomena that we know very well and can describe fully by equations. Thus I see no reason why one solution can be considered safer than another. Because your approach leads to this result ; none of the human-life-involved safety systems should be upgraded, like never. > There's seems to be two options here, remote sensor 2km away or a _really= _ > sensitive detector that can recognize a train 2km away. Granted a remote > sensor has issues - power and communication mainly. Those seem readily > surmountable with off the shelf product. Solar and batteries, RF link, la= y > a cable, induce the signal into the track. > It shouldn't be really sensitive, you can mock it up with an off the shelf detector these days. You can even design the detector (antenna), 2km's is not that far actually. > > I think the killer problem for the really sensitive detector concept is > people. Several sensors have been suggested that _could_ sense a train fr= om > a distance in a controlled environment. But inverse square law and common > sense says that any detector that can sense a train at 2km is going to be > swamped by a bicycle at 1m. Multiple sensors and complex filtering _may_ = be > able to discern signal from the noise (rhythm of the train going over the > track joints?) but complexity is rising as reliability is falling... > It will not be affected from bike, you can estimate the size of the object as well, no problem. Totally implementable under a few k$. --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .