I did a project on this very idea (using automobiles on a road rather than trains) in University. I did not have any success tracking the Doppler shift of the noise from a car (a train might be easier but would run into many of the same problems). What I ultimately did was use a stereo microphone and time the difference between the sound peak in the left and right channels. That produced very good results. You have to calibrate it at first to determine the effective positions of the two points where the microphone pattern maxima intersect the road centerline, but it works well after that. Since a train is longer, you would have to look for some feature other than the peak because the "peak" would be much longer. Perhaps you could look at when the sound exceeds some threshold in each channel, or you could convolve the signals from the L and R channels to determine the effective delay between them. Sean On Wed, Oct 27, 2010 at 4:47 AM, Richard Prosser wrot= e: > On 27 October 2010 14:47, Youda He wrote: >> Magnet could work. Will try that, we can not mount anything on the >> train, so the magnet needs to be about a foot away from the train. >> >> Sound wave is also an interesting idea how accurate can that be? >> >> -- Youda >> > > I think that the accuracy of the FFT/sound method may depend on what > noises the train makes. If it was a constant pure tone and travelling > at a constant speed , I think you would be limited by the sample rate > and possibly the geometry of the situation. > But it won't be. There may be motor noise however which could have > components that are useful enough. I guess it comes down to what you > need =A0- =A0What is the accuracy as a percentage? > > It the train is slowing or accelerating it might be a bit more difficult. > > > > RP > > > >> >> On Tue, Oct 26, 2010 at 6:08 PM, Richard Prosser w= rote: >>> Could you record the noise it makes, do an FFT on it and get the speed >>> from the doppler shift? - Eg compare frequency spectra =A0coming & >>> going? >>> >>> RP >>> >>> On 27 October 2010 13:56, Mark Rages wrote: >>>> On Tue, Oct 26, 2010 at 7:32 PM, Youda He = wrote: >>>>> Camera will be problematic, sub tunnel is dusty and moist, that will >>>>> cause the maintainance problem, we will need tgo send people into >>>>> tunnel from 2am to 5am to clean things up. >>>>> >>>>> I have found some radar detector (like the ones on the street telling >>>>> you how fast are you travelling), cost < $500. Still want to find >>>>> alternate solutions just to see what are there. >>>>> >>>> >>>> An inductive proximity sensor can probably sense the presence of a >>>> subway train. >>>> >>>> Two sensors placed a known distance apart, and measured with accurate >>>> timer, will give the average speed between them. >>>> >>>> Regards, >>>> Mark >>>> markrages@gmail >>>> -- >>>> Mark Rages, Engineer >>>> Midwest Telecine LLC >>>> markrages@midwesttelecine.com >>>> -- >>>> http://www.piclist.com PIC/SX FAQ & list archive >>>> View/change your membership options at >>>> http://mailman.mit.edu/mailman/listinfo/piclist >>>> >>> -- >>> http://www.piclist.com PIC/SX FAQ & list archive >>> View/change your membership options at >>> http://mailman.mit.edu/mailman/listinfo/piclist >>> >> >> -- >> http://www.piclist.com PIC/SX FAQ & list archive >> View/change your membership options at >> http://mailman.mit.edu/mailman/listinfo/piclist >> > > -- > http://www.piclist.com PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .