KEEPING BEACON PROGRAM TIME ACCURATE. Page last altered 09 June 2011 GPS notes

While Peter's program seems capable of maintaining it's time lock indefinitely from the received signals it may be considered desirable to keep the computer locked to a time standard. This seems to be the only way to help guarantee an unattended restart due to a power cut etc. (See below, there may be pitfalls!) 

This can be done it a variety of ways, eg GPS, internet time server or by radio. A radio unit enables the computer to be corrected without an external connection and is considerably cheaper than the GPS method. It is also possibly  more accurate than GPS! The NMEA output of a GPS is usually one second behind so a program must compensate. A local internet Time Server can often be better than 100mS. But see below, using a GPS with 1 PPS output very accurate setting is possible.

Nick, G4IRX, suggested using a MSF or DCF receiver and "Tardis". In the time it took me to find my module,  rig up a prototype and make this page Nick produced a nice page describing his unit.  

I include this, my unit is slightly different and I've made a few notes about the implementation.

The program:- is "Shareware".  Some suggestions here:- Maplin do not seem to stock modules any more but many LCD Radio Controlled Clocks have separate modules and are seen cheaply at times..


Setting the time on a computer can be accomplished by using time servers. The program I use is "About Time" at It compensates for the "ping" time. Click here for my .ini file containing servers in the UK. (Replace the .ini in the program.) This can be used on its own if you have a reliable permanent internet connection. I often disconnect overnight, I find no conflicts using MSFWIN at the same time so each backs up the other.

I have only used this with a computer connected to the internet, it is possible to configure "About Time" so the "Viewer" computer is a "Server" so computers on the network are synchronised by that computer, even if the viewer computer has not been set by the internet. (The program is run on each computer) (I tried this without success, so...)

GPS.  GPS units are available quite cheaply now, appearing on ebay regularly. Motorola Oncore units seem more suitable, most software reads the NMEA data to set time, the Rockwell Jupiter has a very useful precise 10KHz output but I have not yet found out how to get the NMEA  GPRMC output. Ah! The Rockwell LabMon does not seem to do it but CsiGps (Zip File) is able to configure the Jupiter to send GPRMC sentences, as well as others.

The information for these modules is easily found on the internet, connections to them need to be made carefully! The output from them is at TTL levels but I have not yet found a serial port that is not TTL compatible. A single transistor may be used as a buffer as in the clock interface below, reversed for the GPS input. Many hand held units can be configured to give a NMEA output, eg Garmin Etrek.

The COAA software below may be used. A command line utility is NMEAsync, this can be executed at suitable intervals with cron or pycron If anyone asks I will produce some details. Settime is a more comprehensive Windows shareware application.

"Tardis" does not seem suitable for use with GPS, I have not found a way to compensate for the inherent delay of the NMEA output, it sets the computer two seconds slow. I understand it is possible to correct it by altering an ini file. 

June 2011 One Pulse Per second (1PPS) from a GPS. Sets time very accurately.

The above units are not so easily available now but do show up occasionally. For computer timekeeping one that is now available cheaply (Ebay etc) is the discontinued Garmin GPS 18 LVC. It is an integrated unit with a 1PPS output. Look on the Garmin site for an Operation Manual and configuration software. Software for time setting:- ToyNTP is free and was designed for this unit. Satk (Satellite Assisted Time Keeper) is a Japanese software, the English version is here This works with the Rockwell and can be configured to work with the Garmin GPS 18 LVC. Select different GPS models until the seconds display is accurate.

The 1PPS may be used to produce a frequency standard, various module information on the web .

I have some more notes about GPS on this page CLICK

Richard Russell has produced a Windows version of his DOS MSF program This has various input choices, the default being pin 14 of the games port, TTL level.  The same interface will work, 5V is on pins 8&9, ground pin 5. This program is freeware so no need to register "Tardis". 

This program works well using the circuit below.  On my test computer I just used an adapter lead, 15pin to 9pin. The 5 Volt line will most likely provide much more power for a different receiver, it seems to go directly to the computer's 5V line, a fuse and smoothing would be desirable. My circuit can be used with a serial or parallel port input as well, the connections and settings are detailed in the MSFWIN text file.

The 15 pin "D" on my monitor computer is not a standard game port so I have looked at the COM and LPT ports. The serial com ports have no pins set high to power anything. The printer LPT port  pin 4 is set high in all input settings and will feed my  low power module.

I have used a module from a cheap clock. (It was a 4.99 Maplin radio controlled clock, I changed it twice but none of them ever set itself. However the radio module in the third one worked and was extracted after I had tried everything to get it to synchronise. The clock continues to function as a surprisingly accurate crystal controlled timepiece!) It is possible other clocks contain a similar module, Nick suggests one. Try Google for inspiration! Test and note the connections before removing. The Maplin LCD wall clock, sometimes sells at 9.99, has a similar separate module. 

It  strikes me that the pulse could be extracted from a clock without removing the module, it would have to be wired to receive continuously with, maybe, a larger battery or external power.


My crude "prototype" interface is shown. A piece of copper clad board with lands cut with a small hacksaw blade. As so often happens this board will most likely be used as the finished product.

The module actually runs on down to 1.0 volts, but was fed from 3 volts in the clock, with a current of less than 100uA. The simple interface enables it to be powered from a serial RS232 "Com" port. 

If you use "Tardis" and have no com port a USB/com port converter may not be able to supply sufficient current to power a unit. a list of possible converters is in here.  (These are reckoned to be able to power a CAT level converter.)

Other radio modules could be used, but might need external power. This could be derived from the computer 12 volt line with adequate fusing, smoothing and regulation. I have found a filter choke from an old car radio to be effective smoothing with the addition of an electrolytic. I have powered audio amplifiers this way.

There are probably quite a few receivers from the "Rubichron" MSF clock around, there is a design for a home brew receiver in "Radcom" April 1994 issue. I used this for a MSF frequency standard. Using MSFWIN on the game port will power anything that needs 5V. I have tried a "Rubichron" unit, a serial port is able to power it. 

I have retuned a "Rubichron", with a different antenna, to DCF (LW rod with 1100pF, but not all produce a good enough signal)  Software  used  (This ExpertMouseCLOCK(BNC) software is commercial,  but there is no warning regarding its usage. There is also a version for  the  Swiss HBG system, 75KHz, the transmitted data is the same) Serial port, pin 4 power, pin 2 pulse. 

I have checked a DCF LCD alarm (The first radio controlled clock I had). It has two ICs, on for the radio and one for the clock. There are two tracks between them, one carries the pulse, the other 1V5 to power the radio. If the power track is cut the radio chip can be fed with 1V5 and the pulse is produced continuously with no apparent effect to the clock's operation. Most "Radio controlled" clocks do not receive continuously, just synchronise regularly, eg. at 1 or 4 hour intervals, watches may just synchronise overnight when interference is lower. The pulse line on this clock is high impedance, a transistor or two is needed to isolate it, similar to the circuit below.

Receivers need to be kept away from sources of interference, notably monitors, (Including TFTs!) televisions and "economy" lamp units.

"Tardis" requires a negative-going pulse on  serial port, pin 1. "MSFWIN" adapts to either + or -ve pulses.

My interface for low power Radio. Radio connections are shown as in the picture of the module. Diodes 1,2,3,4  1N914 type. D1,2,4 prevent supply going negative when "Tardis" is not active.  D6, LED, verifies reception, flashing regularly in time with radio pulses. Q1 inverts/level shifts  pulse, any small medium gain device. Resistor values may need adjustment. Would probably be OK with lower value capacitors. R1 just reduces the voltage across LED, so it almost extinguishes in the "off" state. (Circuit produced by "Circuit Maker Student" - no longer available.).  Nick uses less components, decide which are necessary!

For use with MSFWIN a 5V supply is adequate coming from pin 8 of the game port, (or pin 4 of the LPT port.)  Pulse from the collector of Q1 to pin 14, ground pin 5.    (See MSFWIN.txt for other configurations.)

To set "Tardis" you may "Clear All" the servers, then "Add". Set protocol as a "Serial Radio Clock" Nick's page gives more details.

Sometimes when I tried "Tardis" it did not setup the com port correctly, I set it as "Galleon Radio Clock" which did, then reset to "Serial" and the correct configuration was maintained. This configuration is pin 4 high (around +9V), pin 1 zero .

Note that MSF is not absolutely 100% reliable, it MAY be off for a while on Tuesdays and possibly for longer periods during the  maintenance week in October. May/June 2005 Major work on the masts meant prolonged interruptions.  I am now using "About Time" as well, so far there has been no apparent conflict. 

Beacon Program:- If "Track Time" is not ticked the program may lose synchronisation until the signal returns.

If "Track Time" is ticked the program will follow the received beacons but LOOSE synchronisation when the computer time is corrected as the MSF signal returns. To correct this the "Track Time" will have to be reset to zero.

When setting the computer this way set "Track Time" and "Track Rate" to zero (except if a GPS is 1 second out! The COAA software below compensates.)

Keeping the time locked will enable a monitoring computer to resume providing it will restart after a power cut. The radio used must also keep it's settings if it looses power. There could also be problems if it restarts too far from the frequency used the last time the program was closed normally.

Note that an abnormal shut down of the program will not save the frequency and time settings in use. If these, or filenames, are reset  close the program and restart to save them, these settings will then be used after an abnormal shutdown, but not frequency and time after another normal shutdown. (If the boxes are not ticked the values will not alter.)

Index / Archives Page 

Some links to time setting ideas. DCF/joystick, W9x only. Will not run on XP. Use Expert Mouse Clock GPS freeware.  SpecLab decodes time signals, MSF DCF77, HBG. Will set PC, but only manually. A little off topic but it will act as a VLF receiver with the antenna connected to a soundcard. As well as being a spectrum analyser and recorder!