How to make synchronized atomic clock?

[Romcik Molnea]

Hello, I ‘m a student and I need your help. I wanna make a synchronized atomic clock. What should I start from? I am gonna make it from zero.

 

[schmitt trigger]

GPS receiver modules have frequency outputs which are fully synchronized to atomic clocks. So right there is your first building block.
If I remember properly, the output is 1 Hz, which can be used as a timebase for your clock.

Now, if you want higher frequencies, you can always use a PLL frequency multiplier. Ratios of 100:1 or even 1000:1 are easily achievable.
For higher ratios phase jitter may become an issue.
In those instances you require a VCXO (voltage controlled crystal oscillator) module, whose frequency is tempered by the GPS output.

Silicon Chip magazine had such a project about 5 years ago.

 

[D.A.(Tony)Stewart]

Atomic clocks are very frequency stable but noisy so an SC cut ovenizd VCXO is PLL is sync'd to the atomic clock.

Since you are not likely to want to make a rubidium oscillator , what you mean is you want a Stratum 1,2 or 3 level clock.

GPs clock with 1MHz and 1pps outputs are available but , try to understand the limitations and requirements from the scope of Stratum level clocks. https://en.m.wikipedia.org/wiki/Network_Time_Protocol

 

[Tunelabguy]

If you are making your atomic clock "from zero", a GPS receiver might not qualify as "from zero". In that case you might want to consider a much older atomic clock technology - VLF radio from station WWVB in Colorado. This is only feasible in the United States since VLF radio does not travel very well. But there are DIY projects where the signal from WWVB (60 kHz) is used to sync a clock. Just google WWVB receiver. It's not as accurate as GPS, but it is low tech (comparatively), and it is synchronized to the same atomic clock that syncs GPS.

 

[betwixt]

To add to that - there are lots of other reference clocks on or around 60KHz in other parts of the World although they all use different coding methods. My 'local' one is about 120Km away but a strong signal here and is accurate to (I think) one second in about 11,000 years. Receivers are very easy to build but some thought has to be be given to locking your oscillator to 1Hz pulses. They do have the advantage of not only being very precise freqency standards but also telling you the time and date as well. MSF (my local one) even tells you if the rotation of Earth is fast or slow compared to the standard day length!

Brian.

 

[D.A.(Tony)Stewart]

My 1st design in 1975 was a multi-channel VLF receiver . The US has MegaWatt transmitters around the world with the carrier Freq. all multiples of 100Hz in the 20kHz range. Antenna is just a long wire. Front end is a low noise FET. I used 3 stage ceramic filters but you can use Active Bandpass Filters. The CMOS PLL has a limiter built in, and recovers the carrier clock and an other PLL or simple down counter generates the reference clock you want. Such as 1pps or 10MHz. If VCXO error is high, eg 100ppm you may need dual bandwidth filter on type I detector to go from fast to slow <<1 sec rate of change. Otherwise get a 1 ppm TCXO chip with voltage control.

A TCXO can be used for the PLL or improved by adding a stable heater resistor and thermistor to regulate the temp and thus the error to 0.01 error for short term LOS or loss of signal.

Similar can be done with 60kHz WWVB which also includes binary coded data for corrections to UTC.