PSN-L Email List Message

Subject: Re: Timing Accuracy Within WinSDR
From: Larry Cochrane lcochrane@..............
Date: Fri, 04 Jul 2008 16:02:10 -0700

Chris and Tom,

A/D data is time stamped at the digitizer board 
(, when a GPS receiver is connected to it. 
WinSDR does not use the computer's time when GPS time is used as a reference. If 
everything is working correctly the accuracy should be under +-4 milliseconds, will 
within one sample time at 100 samples per second. The GPS receiver connected to the 
board must have a 1 pulse per second (PPS) signal, that is why the board only 
supports the Garmin GPS 16/18 sensor and the now obsolete Motorola ONCORE receiver.

The firmware on the board maintains an accumulator, incremented by a 1 millisecond 
interrupt, that contains the time of day (UTC time) down to the millisecond. This 1 
ms interrupt also monitors the 1 PPS GPS signal. At the top of the second the 
accumulator time is saved and sent to the host computer (running WinSDR) along with 
the ADC data. It's up to the host computer to adjust the ADC time accumulator, by 
sending commands to the board, to stay within a few milliseconds of UTC time. The 
time of day from the GPS NMEA messages is also parsed and sent to the host computer 
so it knows what second the pulse happened at.


You should not be trying to use earthquakes to check your station timing! There are 
just too many variables. The accuracy of the travel time tables being one of them. If 
you suspect that your timing is not correct you should inject a signal at a known 
time into the A/D board and viewing the results in WinQuake. One way of doing this is 
to disturb the sensor at the top of the minute by listening to one of the WWV 
stations and jumping next to the sensor right at the top of the minute mark. Another 
way would be to connect a 1.5 volt battery to the input of the A/D channel, again 
right at the top of the minute mark. With a little practice you should be able able 
to get you test pulse to within +-100 ms of UTC time. While a test like this will not 
give you millisecond accuracy it should be good enough to make sure you are within +- 
1/2 second of UTC time.

Larry Cochrane
Redwood City, PSN

ChrisAtUpw@....... wrote:
>        I am not sure how WinSDR actually does the timing, even when corrected 
> by GPS. The display program will presumably use the software clock on your 
> computer. Have you asked Larry? The software clocks fitted to PC computers are 
> often crap. They can loose or gain 10s of minutes per day. 

Public Seismic Network Mailing List (PSN-L)

[ Top ] [ Back ] [ Home Page ]