home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.time.ntp
- Path: sparky!uunet!UB.com!quack!mrapple
- From: mrapple@quack.sac.ca.us (Nick Sayer)
- Subject: Re: Accuracy problems on the Heathkit clock
- Message-ID: <fR8Pqm#@quack.sac.ca.us>
- Organization: The Duck Pond public unix: +1 408 249 9630, log in as 'guest'.
- References: <Sep.2.11.39.26.1992.26046@turbo.bio.net>
- Date: Thu, 3 Sep 1992 19:54:24 GMT
- Lines: 52
-
- shibumi@turbo.bio.net (Kenton A. Hoover) writes:
-
- >At some point in time, there was some discussion about how Heathkit clocks
- >made poor sources for chimers. I just recieved the technical manuals for the
- >new Motorola 6-channel GPS core receiver. The interface on it maxes out
- >at 9600 baud, and I was trying to figure out if it would suffer from the
- >same problems as the Heathkit in terms of accuracy. Would someone care to
- >summarize the reason for the Heathkit's problems?
-
- They pretty much boil down to two things:
-
- 1. There is no point in the RS-232 timecode that is an official
- "on-time" mark. At 9600 bps, the 20 characters (or so) in the
- timecode take 20/960 of a second, or about 20 msec.
-
- 2. The timecode lacks sufficient granularity. The timecode itself
- contains time down to the nearest 100 msec. This wouldn't be so
- bad if it was guaranteed to start or end on time, but that is
- not the case. The timecode starts whenever it likes within that
- 100 msec window. This is the truly damning fault.
-
- #define FLAME
- Both of these faults were confirmed by technical support personel
- at Heath hot-line. They acknowledged that it was not possible
- to get 5 ms timestamps from the clock. They said that the clock
- was not supposed to be considered a serious tool. I suspect
- this sort of attitude is what did Heath in.
- #undef FLAME
-
- The moral: The GC-100 can be trusted to about +/- 120 msec. Hardly
- the 5 ms they claim. To be fair, it is possible that the clock
- internally gets +/- 5 ms, but that isn't terribly helpful.
-
- >BTW, the Motorola 6-channel GPS core reciever eval kit will cost $1200.00
- >and will be available in November.
-
- Most reasonable clocks define some point within the timecode at which
- the time given by the timecode is accurate. For example, CHU timestamps
- are on-time at the end of the last stop-bit in the code. Some other
- systems send a garbage character and use the begining or end of a
- particular bit in that garbage character to signify the point at
- which the next (or previous) timestamp is in effect. You can then take
- a local clock reading of what time your system THINKS it is when that
- bit has arrived, then compare that stamp with the time that the
- bit in question represents and thus get a good difference between
- the standard and your local clock.
-
- --
- Nick Sayer <mrapple@quack.sac.ca.us> |
- N6QQQ @ N0ARY.#NOCAL.CA.USA.NA | "Leave everything to me!"
- 37 19 49 N / 121 57 36 W |
- +1 408 249 9630, log in as 'guest' | -- Powdered Toast Man
-