From: Harlan Stenn <stenn@ntp.org>
Date: Tue, 30 Mar 2004 02:57:51 +0000 (-0500)
Subject: New irig notes from Dave Mills
X-Git-Tag: NTP_4_2_3~174^2~3
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=f89c1a4509773ced716a4e7cecc2a5f4169473cb;p=thirdparty%2Fntp.git

New irig notes from Dave Mills

bk: 4068e22fPW_To3XOB3gzytnOk4h9GQ
---

diff --git a/ntpd/refclock_irig.c b/ntpd/refclock_irig.c
index 0b3536863d..ad5ab41bc3 100644
--- a/ntpd/refclock_irig.c
+++ b/ntpd/refclock_irig.c
@@ -116,13 +116,14 @@
  * timestamp in NTP format.
  *
  * The fraction part of the on-time timestamp is a good indicator of how
- * well the driver is doing. With an UltrSPARC 30 and Solaris 2.7, this
- * thing can keep the clock within a few tens of microseconds relative
- * to the IRIG-B signal. Accuracy with IRIG-E is about ten times worse.
- * Unfortunately, Sun broke the 2.7 audio driver in 2.8, which has a
- * 10-ms sawtooth modulation. The driver attempts to remove the
- * modulation by some clever estimation techniques which mostly work.
- * Your experience may vary.
+ * well the driver is doing. Once upon a time, an UltrSPARC 30 and
+ * Solaris 2.7 kept the clock within a few tens of microseconds relative
+ * to the IRIG-B signal. Accuracy with IRIG-E was about ten times worse.
+ * Unfortunately, Sun broke the 2.7 audio driver in 2.8, which has a 10-
+ * ms sawtooth modulation. The driver attempts to remove the modulation
+ * by some clever estimation techniques which mostly work. With the
+ * "mixerctl -o" command before starting the daemon, the jitter is down
+ * to about 100 microseconds. Your experience may vary.
  *
  * Unlike other drivers, which can have multiple instantiations, this
  * one supports only one. It does not seem likely that more than one
@@ -132,13 +133,12 @@
  * Fudge factors
  *
  * Fudge flag4 causes the dubugging output described above to be
- * recorded in the clockstats file. When the audio driver is compiled,
- * fudge flag2 selects the audio input port, where 0 is the mike port
- * (default) and 1 is the line-in port. It does not seem useful to
- * select the compact disc player port. Fudge flag3 enables audio
- * monitoring of the input signal. For this purpose, the monitor gain is
- * set to a default value. Fudgetime2 is used as a frequency vernier for
- * broken codec sample frequency.
+ * recorded in the clockstats file. Fudge flag2 selects the audio input
+ * port, where 0 is the mike port (default) and 1 is the line-in port.
+ * It does not seem useful to select the compact disc player port. Fudge
+ * flag3 enables audio monitoring of the input signal. For this purpose,
+ * the monitor gain is set to a default value. Fudgetime2 is used as a
+ * frequency vernier for broken codec sample frequency.
  */
 /*
  * Interface definitions