DIY GPSDO with UBLOX M8N and VECTRON C4550A1-0213 OCXO 10MHz

In recent years, technology has made leaps forward in small handheld or laboratory GPSs such as those created for the NEO UBLOX series. I recently bought a NEO M8N, which itself mounts a programmable TCXO. It is indeed possible to connect the NEO module to a serial and program it with the U-Center program , which is free and can be downloaded from the Ublox website.

There are substantial differences between models, especially with regard to the internally mounted oscillator. Some mount XTL oscillators , others such as the 8N mount a TCXO . Please check the relevant datasheets for specifications. For a quick reference:

NEO-6T —> TCXO
NEO-7M –> XTL
NEO-7N –> TCXO
NEO-M8M > XTL
NEO-M8N > TCXO
NEO-M8Q > XTL

although some projects can be found online ,where the programmable lock frequency is used as a reference, the UBLOX output signal is not always clean, especially at high frequencies such as 10 MHz. In fact, the signal is characterized by jitter and overshoot . (To learn more about this please view this video).

But at low frequencies, convenient for an IC like the 4046 PLL , 10 kHz is more than enough to use this signal to discipline an external, well-made oscillator. For this test I chose the VECTRON C4550A1-0213 , 10 MHz square wave, which is readily available for few dollars in online markets. This oscillator has a voltage control pin , although it is not easy to find the datasheet online, I was able to retrieve some interesting features such as the stability announced by the manufacturer. Which makes it very interesting given the good long-term stability.

I don’t have any TIC with me at the moment so I cannot test this oscillator, but will do in the future.
The circuit diagram is as follow

The square wave from the Vectron OCXO is divided with CD4017BE ICs synchronous counters , first from 10 MHz to 1MHz, then 100 kHz and finally 10 kHz. The signal then arrives at the phase comparator CD74HC4046AE . The output of phase comparator 2 , through the loop filter, reaches the EFC control pin of the oscillator. I calculated for this OXCO the Kv (Tuning Law) is about 1E-5 MHz/V. And then it executes the loop. By using the phase comparator type 2 in fact it doesn’t matter the duty cycle because you’re going to lock in frequency, preventing it from hooking into harmonics. Finally, I added a low-pass filter to have a decent 10 MHz sine wave at the output .

The loop filter was specifically calculated for this oscillator with a PLL simulator. If you need the values for the filter, you can write a message.

In this video it’s possible to see the whole process till the OCXO is locked on the UBLOX GPS reference.
Please note that I configured the NEO M8N to release a 1Hz 50% Duty Cycle while NOT locked and 10 kHz 50% Duty Cycle while locked to GPS in 3D+ mode. (lower part is the GPS, upper part is the 10 kHz from the divider) .The video has been recorded in two different session so it is not perfectly synchronized as I had to use the probes on the IC directly , but the overall result is the same. The divider from the OCXO instead has a different Duty Cycle, about 10%.

I Tested the same circuit also with a CTI OSC5A2B02 OXCO and works pretty much the same as for the Vectron, but the Kv value for Voltage control is a bit lower hence will require few seconds more to lock , or a different loop filter configuration.

Overall using a programmable GPS such as the NEO GPS, it is not difficult to have a GPSDO ready to go, considering few years ago was absolutely mandatory to develop a similar circuit based on the GPS 1PPS output only, which requires at least a microcontroller or an FPGA to lock into and render the design exponentially more complicated to perform.