Path: utzoo!attcan!uunet!know!zaphod.mps.ohio-state.edu!wuarchive!decwrl!csus.edu!ucdavis!uop!nsayer From: nsayer@uop.edu (Nick Sayer) Newsgroups: comp.protocols.time.ntp Subject: Heath clock timing, and building a WWV clock Message-ID: <271228ea.69ef@uop.uop.edu> Date: 9 Oct 90 19:45:46 GMT Organization: University of the Pacific, Stockton, CA [138.9.200.1] Lines: 42 Heath's "hi-spec" accuracy is +/- 10 ms, and its resolution is 100 ms. That comes from the manual's definition of hi-spec, and the fact that the clock's least significant digit is tenth-of-a-seconds. That's not so good, but the picture gets even bleaker. I just got off the phone with Heath, and they say there is no specification synchronizing the RS-232 output to real time! That is, you're not sure WHEN the data being transmitted is right. It could be right at the begining of the transmission, it could be right at the carriage-return. Given that a bit-time at 9600 baud is about a microsecond, and that it takes about 300 of them to get the message across, that's +/- another 3 or 4 msec. Besides which you're not sure WHERE within the 100 msec resolution you are. The net result is a resolution AND accuracy of 100 msec. Never mind the fact that the time you get is exactly 1 second slow! Well, that makes it a nice conversation piece, I suppose... If one is not happy with that, how tough is it to build a propper WWV clock? If one wanted to keep it simple, would this do: for GEE-173 (uProc sys design -- I'm a college student) we built an 8088 based small system with some ram, some rom, 2 8 bit parallel ports and a serial port (actually an 8256, so it's got some timers and stuff too). If one had the 100 Hz and 1 kHz decode outputs (i.e. a logic 1 when 100 Hz is heard and 0 otherwise), the 8088 can figure out what time it is, probably using the same sort of heuristics the heath clock uses. The difference is that you could provide a sensible RS-232 output. Using the sort of circuitry that heath used (567's for the tone decoders and a normal double-het AM receiver), what kind of accuracy could I get? How about if I keep it REALLY simple and just grab the two decodes straight from my existing heath clock? Let it worry about which signal (5/10/15 MHz) to use, and just piggyback the data. the 8256 internal timers have a basic granularity of 62.5 usec (or 1 msec if you want), but this can be halved with some simple handwaving to 32.25 usec. If this isn't the right group to talk about this stuff, just let me know and I'll be on my merry way. -- Nick Sayer | "Disclaimer? I don't think so! mrapple@quack.sac.ca.us | Homey don't play dat." N6QQQ [44.2.1.17] | 209-952-5347 (Telebit) | -- Homey the Clown