PSN-L Email List Message

Subject: Re: Computer Timing Problems / Solutions
From: ian ian@...........
Date: Sat, 16 Apr 2005 11:34:51 +0100


Regarding the standard a/d board and without knowing its details, I note 
that there is a crystal onboard and that the spec of the a/d is 100 
samples/second.  What would be the accuracy obtained for the time 
between the P and S wave by simply counting samples between them and 
multiplying that by the sample period?  Would this suffice?  The flip 
side of this question is: if that crystal isn't very good, then is all 
this for naught!?

Thinking more about this board, is the sample data buffered (other than 
that provided by the uart)?  Are samples lost/overwritten if the pc is 
temporarily busy?

Also, after last week's discussion, was the conclusion that the accuracy 
which can be achieved by a good NTP service sufficient for absolute 
timing and that the emphasis should be on the relative time between the 
2 waves.  Again, what is a reasonable spec?

Ian Smith

ChrisAtUpw@....... wrote:

> Hi there,
>  
>     I have inspected six new computer motherboards. They all appeared 
> to be using the miniature cylindrical 32,768 Hz watch crystals. You 
> can get AT cut crystals in these cases but watch crystals are the 
> commonest. 32 KHz Watch crystals have a parabolic error plot with 
> temperature, peaking at about 25 or 30 C, +/-5 C. The coefficient is ~ 
> 0.04 x (Cdiff)^2 ppm. If it is a 30 C crystal and the temperature 
> falls to 5 C, you can expect to get 25 ppm drift - about 2 sec per 
> day.     
>     However, the time loss errors shown on my newish computers are 
> very considerably above this, so I suspect that this must be due to 
> changes / errors in the counted interrupt rate. On older computers the 
> disk drive R/Ws could override the regular interrupt timing. The time 
> update options given in Windows do NOT seem to "discipline" the rate 
> loss of the time system. Some other programs can do this.
>  
>     Most of the USA can receive the WWVB 60 KHz timing signals. The 
> coverage is shown on 
> http://www.boulder.nist.gov/timefreq/stations/wwvbcoverage.htm Galleon 
> Systems at
> http://www.ntp-time-server.com/  produce a range of timing products, 
> from complete time server units to separate module boards.
>     The receivers use a 60 KHz tuned ferrite aerial, costing $2.68.
>     The EM2S receivers cost $10.89. They have a normal operating range 
> of 3 to 12 V DC. They use a crystal filter to get a high stability 
> narrow band response and have an AVC circuit to optimise reception for 
> changes in signal strength.
>     The MCM-RS232 Microcontroller Module costs $15.89. It operates off 
> 3 V DC. When combined with the EM2 Receiver Module and the Antenna, it 
> provides time information in standard RS232 data format via a serial 
> interface. External buffering to full RS232 levels is required using a 
> proprietary RS232 level shifter IC. The advantage of the module 
> compared to direct host decoding of the receiver output is the 
> continuous availability of exact decoded time information with no host 
> processing overhead. The module has a real time clock that 
> recalibrates itself against the Atomic Radio Time Signal.
>  
>     I have used these modules 1,800 miles from the transmitter and 
> they worked reliably. At ranges over about 500 miles, the ferrite 
> aerial should be horizontal, mounted perpendicular to the direction of 
> the transmitter and it can be an advantage to mount the modules high 
> up and external to the building, facing in the direction of the 
> transmitter. I tried a variety of situations to try to make the system 
> fail. Failures were all due to the presence of strong interfering 
> radio signals and not to signal strength / internal noise problems. 
> The decoder modules need about 4 mins of clear signal to synchronise 
> initially, but can update hourly in less than 90 sec. I connected a 
> piezo earpiece to the receiver output and listened for any interfering 
> signals or changes in the output pulse rate. I also measured the 
> voltage on the AVC capacitor, min 0.7 V up to the normal range of 0.9 
> to 1 V, as a logarithmic indication of the signal strength when 
> selecting reception sites.
>  
>     The receiver + aerial need to be >6 ft away from CRT computer 
> monitors and TV sets. It didn't work well very close to radio and TV 
> transmitters. Steel framed buildings and Al foil / corrugated iron 
> covered roof spaces showed low signal levels. Roof spaces can suffer 
> from very large temperature variations. The radio signal was still 
> received OK, but the presence of utility power wiring allowed the 
> pickup of RFI. Local lightning can prevent clear reception. Local 
> electric arc welding may also give problems.
>  
>     The EM2S receiver would work without problems in poor signal 
> locations where my LW AM radio was seriously effected by 
> internal noise. You can buy (borrow?) battery clocks which give 
> 'Atomic Time' eg Oregon Scientific. These work off the WWVB signal 
> and those with LCD displays have a radiating aerial mast display with 
> four levels indicating the signal strength at the last update. My 
> clock updates every hour. I have no financial or other connection to 
> Galleon Systems.
>  
>     If you look at 
> http://www.boulder.nist.gov/timefreq/stations/wwvbtimecode.htm you 
> will see that the 59th and the 1st second of WWVB signal both start 
> with 800 mS low pulses. It should be quite easy to set up a dual 
> retriggerable monostable multivibrator to detect this and to give high 
> precision minute timing and / or second timing pulses.  
>  
>     Larry, I would like to suggest that you give consideration to 
> providing the 16-bit Serial Output A/D Board for WinSDR with an option 
> to receive and fully decode WWVB signals, or the ability to read the 
> output of a MCM microcontroller module. I have read your note, dated 
> 1998, on WWVB signal reception.
>     I note that the serial board can currently use WWV minute tone 
> decode signals, but you state that 'You will not get 24 hour reception 
> on any one channel, as long as you can get 4 to 6 hours per day will 
> be fine. At my location I get best reception on 5.0Mhz at night and 
> during the day 10.0Mhz or 15.0Mhz.'
>     Does the timing for SDR originate on the board, or is it dependant 
> on the computer clock, please? My computers can NOT keep time over a 
> timing break of maybe 12 hours, to better than about 2 sec. This is 
> certainly NOT good enough for seismic work! Unlike WWV, WWVB has the 
> potential to provide accurate timing signals over the full 24 hrs for 
> most places in the USA. The extreme range daytime WWVB signal is 
> certainly a lot weaker than the nightime signal, but with the possible 
> exception of Maine, it should be satisfactory
>     The ability to fully decode the WWVB signal could cope with the 
> situation of a power outage. It wasn't until I bought a 60 KHz radio 
> corrected digital clock that I realised just how bad my computer 
> timing systems were! Another reason for doing this is to provide a 
> timing system totally independant of the www. There are already 
> predictions of future communications problems on the web.
>  
>     The total cost could well be about that of just a GPS ANTENNA !
>     
>     Regards,
>  
>     Chris Chapman




  
  


Regarding the standard a/d board and without knowing its details, I
note that there is a crystal onboard and that the spec of the a/d is
100 samples/second.  What would be the accuracy obtained for the time
between the P and S wave by simply counting samples between them and
multiplying that by the sample period?  Would this suffice?  The flip
side of this question is: if that crystal isn't very good, then is all
this for naught!?

Thinking more about this board, is the sample data buffered (other than that provided by the uart)?  Are samples lost/overwritten if the pc is temporarily busy?

Also, after last week's discussion, was the conclusion that the accuracy which can be achieved by a good NTP service sufficient for absolute timing and that the emphasis should be on the relative time between the 2 waves.  Again, what is a reasonable spec?

Ian Smith

ChrisAtUpw@....... wrote:
Hi there,
 
    I have inspected six new computer motherboards. They all appeared to be using the miniature cylindrical 32,768 Hz watch crystals. You can get AT cut crystals in these cases but watch crystals are the commonest. 32 KHz Watch crystals have a parabolic error plot with temperature, peaking at about 25 or 30 C, +/-5 C. The coefficient is ~ 0.04 x (Cdiff)^2 ppm. If it is a 30 C crystal and the temperature falls to 5 C, you can expect to get 25 ppm drift - about 2 sec per day.     
    However, the time loss errors shown on my newish computers are very considerably above this, so I suspect that this must be due to changes / errors in the counted interrupt rate. On older computers the disk drive R/Ws could override the regular interrupt timing. The time update options given in Windows do NOT seem to "discipline" the rate loss of the time system. Some other programs can do this.
 
    Most of the USA can receive the WWVB 60 KHz timing signals. The coverage is shown on http://www.boulder.nist.gov/timefreq/stations/wwvbcoverage.htm Galleon Systems at
http://www.ntp-time-server.com/  produce a range of timing products, from complete time server units to separate module boards.
    The receivers use a 60 KHz tuned ferrite aerial, costing $2.68.
    The EM2S receivers cost $10.89. They have a normal operating range of 3 to 12 V DC. They use a crystal filter to get a high stability narrow band response and have an AVC circuit to optimise reception for changes in signal strength.
    The MCM-RS232 Microcontroller Module costs $15.89. It operates off 3 V DC. When combined with the EM2 Receiver Module and the Antenna, it provides time information in standard RS232 data format via a serial interface. External buffering to full RS232 levels is required using a proprietary RS232 level shifter IC. The advantage of the module compared to direct host decoding of the receiver output is the continuous availability of exact decoded time information with no host processing overhead. The module has a real time clock that recalibrates itself against the Atomic Radio Time Signal.
 
    I have used these modules 1,800 miles from the transmitter and they worked reliably. At ranges over about 500 miles, the ferrite aerial should be horizontal, mounted perpendicular to the direction of the transmitter and it can be an advantage to mount the modules high up and external to the building, facing in the direction of the transmitter. I tried a variety of situations to try to make the system fail. Failures were all due to the presence of strong interfering radio signals and not to signal strength / internal noise problems. The decoder modules need about 4 mins of clear signal to synchronise initially, but can update hourly in less than 90 sec. I connected a piezo earpiece to the receiver output and listened for any interfering signals or changes in the output pulse rate. I also measured the voltage on the AVC capacitor, min 0.7 V up to the normal range of 0.9 to 1 V, as a logarithmic indication of the signal strength when selecting reception sites.
 
    The receiver + aerial need to be >6 ft away from CRT computer monitors and TV sets. It didn't work well very close to radio and TV transmitters. Steel framed buildings and Al foil / corrugated iron covered roof spaces showed low signal levels. Roof spaces can suffer from very large temperature variations. The radio signal was still received OK, but the presence of utility power wiring allowed the pickup of RFI. Local lightning can prevent clear reception. Local electric arc welding may also give problems.
 
    The EM2S receiver would work without problems in poor signal locations where my LW AM radio was seriously effected by internal noise. You can buy (borrow?) battery clocks which give 'Atomic Time' eg Oregon Scientific. These work off the WWVB signal and those with LCD displays have a radiating aerial mast display with four levels indicating the signal strength at the last update. My clock updates every hour. I have no financial or other connection to Galleon Systems.
 
    If you look at http://www.boulder.nist.gov/timefreq/stations/wwvbtimecode.htm you will see that the 59th and the 1st second of WWVB signal both start with 800 mS low pulses. It should be quite easy to set up a dual retriggerable monostable multivibrator to detect this and to give high precision minute timing and / or second timing pulses.  
 
    Larry, I would like to suggest that you give consideration to providing the 16-bit Serial Output A/D Board for WinSDR with an option to receive and fully decode WWVB signals, or the ability to read the output of a MCM microcontroller module. I have read your note, dated 1998, on WWVB signal reception.
    I note that the serial board can currently use WWV minute tone decode signals, but you state that 'You will not get 24 hour reception on any one channel, as long as you can get 4 to 6 hours per day will be fine. At my location I get best reception on 5.0Mhz at night and during the day 10.0Mhz or 15.0Mhz.'
    Does the timing for SDR originate on the board, or is it dependant on the computer clock, please? My computers can NOT keep time over a timing break of maybe 12 hours, to better than about 2 sec. This is certainly NOT good enough for seismic work! Unlike WWV, WWVB has the potential to provide accurate timing signals over the full 24 hrs for most places in the USA. The extreme range daytime WWVB signal is certainly a lot weaker than the nightime signal, but with the possible exception of Maine, it should be satisfactory
    The ability to fully decode the WWVB signal could cope with the situation of a power outage. It wasn't until I bought a 60 KHz radio corrected digital clock that I realised just how bad my computer timing systems were! Another reason for doing this is to provide a timing system totally independant of the www. There are already predictions of future communications problems on the web.
 
    The total cost could well be about that of just a GPS ANTENNA !
    
    Regards,
 
    Chris Chapman

[ Top ] [ Back ] [ Home Page ]