[LEAPSECS] ATT drops time

[Chris O'Byrne]

I had an encounter with the phone company's clock many years ago that 
leads me to say "good riddance".

I was involved in a project that required very accurate time (1 
microsecond or better). We purchased a GPS with a 1pps output for this. 
The problem I had was that whereas I had these wonderfully sharp 
accurate pulses on each second, I wasn't quite sure WHICH second they 
were referring to!

The GPS manual wasn't that great, but I implemented what I thought was 
the correct solution. I then checked it against NTP time and - bingo - 
agreement! I then checked it against the phone company's clock and - 
hmmm - I was one second fast. So I checked it against a radio time 
signal and, once again, agreement.

So I called the phone company. Unfortunately, the person who maintained 
the clock was out, so I had to talk to his secretary. She was astonished 
- "why do you care that the clock is one second out!?". So I explained 
to her that one second was a big deal when I was trying to be accurate 
to a millionth of a second. After much scratching of heads, she finally 
said to me "Oh! Do you want me to change the time?!". The astonishment 
was now mine.

As one of my colleagues said - "you should have said yes - I'm late with 
my project!".

As an aside note, those of you involved in teaching physics might like 
to consider putting the output of a GPS 1pps and a radio time signal 
into an oscilloscope. This is what I did when I checked against the 
radio time signal. The radio time source I used was Radio Moscow (I was 
working in the west of Ireland at the time, and that was the strongest 
time signal I could receive with the radio set I had). I noticed that 
the time from Radio Moscow was 11 milliseconds "late". Of course the 
reason was because I was about 3,300km from Moscow - I had just 
inadvertently measured the speed of light!

Tom Van Baak wrote:

> Now one less service has to worry about leap seconds.

>  

> http://www.latimes.com/business/la-fi-lazarus29aug29,0,2466396,full.column

>  

> /tvb

> 

> 

> ------------------------------------------------------------------------

> 

> _______________________________________________

> LEAPSECS mailing list

> LEAPSECS at leapsecond.com

> http://six.pairlist.net/mailman/listinfo/leapsecs