[time-nuts] PC Clock adjustments

David Ackrill dave.g0dja at tiscali.co.uk
Sun Mar 9 04:50:05 EDT 2008


Hello,

I'm new to this forum, so please excuse me if I use the wrong 
terminology, or show up my lack of knowledge in the subject of keeping 
time.  Please also excuse the length of this message, but I thought it 
would be best to explain, as much as I can, the set up here and the 
problems that I am hoping someone might be able to help me out with.

Up until recently I just used programs like Dimension 4 to periodically 
check and, if necessary, adjust my PC Clock to bring it back in line 
with the publically available SNTP or NTP time servers.  This seemed to 
work OK for general Amateur Radio use, such as making sure that I was 
transmitting and receiving the WSJT modes at about the right time each 
minute or 30 seconds.  It was also good enough for logging programs as 
well, of course.

However, a friend of mine developed an SSTV program which calibrated the 
soundcard output against the PC clock, and had a routine built in which 
would check the clock against one or other of the (S)NTP time servers 
and adjust the clock if it was out of step.

I sould say, upfront, that neither of the following PCs is 'overclocked' 
and, to the best of my knowledge, have not had any modifications made to 
the motherboards.

I have two PCs permanently connected to the internet (via two different 
ISPs, telephone lines, WiFi routers etc.) and the older one, which has 
an ABIT motherboard in it, I think, seems to keep time reasonably well. 
  It's not brilliant, but Tardis says that it has a daily drift of about 
0.054 seconds per day and NTPmonitor shows that the maximum error, 
usually against time.apple.com, is something like +500 milliseconds 
every couple of hours.  Against the stratum 2 clock that John (N8UR) 
runs, it is much better and there are not the 'spikes' of +500 
milliseconds against his time server, nor are the 'spikes' seen in 
comparison with a couple of other publically available servers.  This PC 
has a 'hard life' being the one that I install any program that takes my 
fancy, is used for email and general internet surfing, is connected to 
an older WiFi unit and to the slower of the two ISP connections.

The other PC has recently been reformatted, Windows XP reinstalled (I 
had tried to use it with an SDR radio, but gave up on that about a year 
ago and I had changed a few things in XP that I could not remember how 
to undo again) it has a newer ABIT motherboard, a KN8, has a faster 
processor than the older machine and is only used for programs connected 
with either running a radio using CAT control, or programs which run 
SSTV/WSJT modes through a soundcard.  I recently removed an M-Audio D44 
soundcard board, in case that was part of the problem, but there's been 
no difference, that I can see, with the time keeping problems on this 
newer PC.

This second PC is a real pain when it comes to keeping it on time. 
About every hour any program which is set up to correct the clock (such 
as Tardis, D4 or the SSTV program mentioned before) detects that the 
clock is up to +1.5 seconds out, however, in the next period it is 
almost the same value out in the other direction!  If I leave NTPmonitor 
running, without any correction, I see the +1.5 second 'glitch' but then 
the PC clock settles back onto reasonably good time keeping, which I 
think is why the automatic adjustment systems carry out a plus followed 
by a minus correction.  Tardis calculates the estimated daily drift at 
about -83 seconds per day and NTPmonitor shows a deviation, from many of 
the various time servers, of up to 2 seconds at times, and quite often 
1.5 seconds.  However, if I turn on the adjustment system, I get the 
+1.5/-1.5 second ajustment effect.

Other than scrapping the newer PC, or fitting a very expensive add on 
unit, can anyone think of a way of tameing this errant PC clock please?

Thanks for any help.  In fact, having this PC clock problem has made me 
read up on the subject of time and the PC Clock. :-)

Dave Ackrill (G0DJA)



More information about the time-nuts mailing list