[time-nuts] 1PPS for the beginner
Tom Van Baak
tvb at LeapSecond.com
Tue Aug 14 15:37:25 EDT 2018
Typically all you want is the leading edge. So a 10 or 20 us wide pulse is enough for most purposes. The duty cycle is essentially zero in this case.
A noted exception is when the pulse is detected using vintage RS232 modem signal inputs, for example, when NTP uses DCD to catch a 1PPS. In that case a much wider pulse, many milliseconds, is required. See also FatPPS . Modern computers which have GPIO don't have this problem since IOC interrupts or time capture registers will detect the leading edge and pulse width doesn't matter.
I used to like 1 Hz square waves. But over the years I've learned that having racks of gear exchanging 1PPS (5V into 50R) along with lots of flashing LEDs, all aligned to UTC, created noticeable periodic jumps in my 120 VIC mains, which invites injection locking.
----- Original Message -----
From: "Bernd Neubig" <BNeubig at t-online.de>
To: "'Discussion of precise time and frequency measurement'" <time-nuts at lists.febo.com>
Sent: Tuesday, August 14, 2018 11:52 AM
Subject: Re: [time-nuts] 1PPS for the beginner
> Another question, a bit off the road of this thread:
> Is there any common practice for the duty cycle of the 1 PPS pulse?
> I have seen devices having extremely short 1 PPS pulses with duty cycles of a few percent or less and others with close to 50% duty cycle.
> Best regards
More information about the time-nuts