NetBSD/usr.sbin/xntp/html/driver26.html

101 lines
3.4 KiB
HTML

<!-- $NetBSD: driver26.html,v 1.1 1998/12/30 20:20:34 mcr Exp $ -->
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML Strict//EN">
<html><head><title>
Hewlett Packard 58503A GPS Receiver
</title></head><body><h3>
Hewlett Packard 58503A GPS Receiver
</h3><hr>
<p><h4>Synopsis</h4>
<p>Address: 127.127.26.<var>u</var>
<br>Reference ID: GPS
<br>Driver ID: GPS-HP
<br>Serial Port: <code>/dev/hpgps<var>u</var></code>; 9600 baud, 8-bits,
no parity
<br>Features: none
<p><h4>Description</h4>
<p>This driver supports the HP 58503A Time and Frequency Reference
Receiver. It uses HP SmartClock (TM) to implement an Enhanced GPS
receiver. The receiver accuracy when locked to GPS in normal operation
is better than 1 usec. The accuracy when operating in holdover is
typically better than 10 us per day. It receiver should be operated with
factory default settings. Initial driver operation: expects the receiver
to be already locked to GPS, configured and able to output timecode
format 2 messages.
<p>The driver uses the poll sequence <code>:PTIME:TCODE?</code> to get a
response from the receiver. The receiver responds with a timecode string
of ASCII printing characters, followed by a &lt;cr&gt;&lt;lf&gt;,
followed by a prompt string issued by the receiver, in the following
format:
<pre>
T#yyyymmddhhmmssMFLRVcc&lt;cr&gt;&lt;lf&gt;
</pre>
<p>The driver processes the response at the &lt;cr&gt; and
&lt;lf&gt;&lt;cr&gt; and &lt;lf&gt;, so what the driver sees is the
prompt from the previous poll, followed by this timecode. The prompt
from the current poll is (usually) left unread until the next poll. So
(except on the very first poll) the driver sees this:
<pre>
T#yyyymmddhhmmssMFLRVcc&lt;cr&gt;&lt;lf&gt;
</pre>
<p>The T is the on-time character, at 980 msec. before the next 1PPS
edge. The # is the timecode format type. We look for format 2. Without
any of the CLK or PPS stuff, then, the receiver buffer timestamp at the
&lt;cr&gt;y is 24 characters later, which is about 25 msec. at 9600
bps, so the first approximation for fudge time1 is nominally -0.955
seconds. This number probably needs adjusting for each machine / OS
type, so far:
-0.955000 on an HP 9000 Model 712/80 HP-UX 9.05
-0.953175 on an HP 9000 Model 370 HP-UX 9.10
<p>This receiver also provides a 1-PPS signal, but I haven't figured out
how to deal with any of the CLK or PPS stuff yet. Stay tuned.
<p><h4>Monitor Data</h4>
<p>When enabled by the <code>flag4</code> fudge flag, every received
timecode is written as-is to the <code>clockstats</code> file.
<p><h4>Fudge Factors</h4>
<dl>
<dt><code>time1 <i>time</i></code>
<dd>Specifies the time offset calibration factor, in seconds and
fraction, with default 0.0.
<p><dt><code>time2 <i>time</i></code>
<dd>Not used by this driver.
<p><dt><code>stratum <i>number</i></code>
<dd>Specifies the driver stratum, in decimal from 0 to 15, with default
0.
<p><dt><code>refid <i>string</i></code>
<dd>Specifies the driver reference identifier, an ASCII string from one
to four characters, with default <code>GPS</code>.
<p><dt><code>flag1 0 | 1</code>
<dd>Not used by this driver.
<p><dt><code>flag2 0 | 1</code>
<dd>Not used by this driver.
<p><dt><code>flag3 0 | 1</code>
<dd>Enable <code>ppsclock</code> line discipline/streams module if set.
<p><dt><code>flag4 0 | 1</code>
<dd>Enable <code>clockstats</code> recording if set.
</dl>
<hr><address>David L. Mills (mills@udel.edu)</address></body></html>