On 2013-03-01, Richard B. Gilbert <rgilber...@comcast.net> wrote: > On 2/28/2013 6:03 PM, Chris Adams wrote: >> Once upon a time, Ralph Aichinger <ra...@pangea.at> said: >>> Has anybody *two* GPIO PPS devices on different GPIO >>> pins on their Raspberry Pi? Would a setup like this >>> enable the ntpd to check the devices against eachother >>> (or probably more likely give an idea of interrupt >>> handling capabilities)? >> >> Remember: a man with two clocks never knows what time it is. >> >> If they are the same, everything's great; if they are different, what do >> you do? >> > > This looks so simple!!! Is there a "catch" somewhere? > > Try adding a third clock. If NO TWO CLOCKS AGREE > you have a serious problem.
Yes, that was Adam's point. However, the problem is getting the time from those three into the computer. While the system is servicing the one interrupt, it cannot be servicing the other. This introduces a delay ( or something like 10us) into the time of the one with respect to the other. Three PPS clocks are even worst. And depending on exactly which one gets the pulse first that delay will not be consistant. So, having three clocks which you KNOW are badly out with respect to each other (10s of us vs the 10s of ns that the PPS are capable of) is that better than having just one, which you then "know" is within about 1us of the true time, but may of course break down and be way out? > > If you are relying on time from the Internet, that > may be your one of your problems!!! > > I use a Global Positioning System (GPS) Timing Receiver > and get time good enough for most purposes. He asked about having two PPS sourses (GPS clocks) vs having just one. > > It's specified +/- 50 nanoseconds. But you cannot get the time into the computer with that accuracy. More like +/- 1 microsecond > > If you need better, you can get it but you have some > studying to do and you had better have a "deep pocket"! > If your "deep Pocket" is not full of $1000 bills. . . . > _______________________________________________ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
