ADRF6850 long term frequency stability test

After yesterday's slightly discouraging frequency stability results using on-board 13.5 MHz reference and a SRS DS345 function generator as REFCLK sources, I decided to try it with one of my eBay OCXOs.

OCXO connected to dual power supply. Left voltage is connected to OCXO's Vtune pin.

ADRF6850 evaluation board is connected to HP E4433B signal generator (148 MHz RF source) and one DC-coupled end of the baseband output is connected to the 54624A oscilloscope via a low pass filter (DC coupled).

I first 'tuned' the OCXO by beating it against a 10 MHz frequency generated by the HP E4433B using a MCL ZAD-6 mixer. The OCXO and signal generator were turned on for one hour and then the OCXO was set by carefully adjusting Vtune until no perceptible frequency difference was detected over ten seconds.

The ADRF was set to:

LO = 148 MHz Frequency doubler = on Frequency division = off

The signal generator was then set to 148 MHz and the OCXO connected to the ADRF6850's REFIN reference clock input. Due to rounding errors, 148 MHz can't be set precisely, so the signal generator was adjusted and a stable baseband output was observed for ten seconds at 147.99999996 MHz:

Note that the trace is horizontal. This is very encouraging!


  • Assuming that the signal generator is perfectly stable at 10 MHz, continuous Vtune adjustment is required to keep the OCXO at the same frequency.. though the actual difference in frequency is very, very low.
  • The OCXO responds surprisingly quickly to changes in Vtune: pretty much immediately.
  • Almost perfect frequency stability over 20-30 seconds is very easy to achieve, even through after an excursion through the PLL and VCO.
  • The DS345 function generator's frequency stability is considerably worse than the OCXO, but due to loud fan noise I haven't let it warm up for a long time, though if it is OCXO controlled, it's possible that the oven is always left on because lit LEDs are visible inside the instrument even when the front panel power is off.. Ah, apparently it has a 5ppm reference oscillator. It would be better to use it with an external 10 MHz reference.
  • I've ordered a cheap ($67) rubidium frequency standard from eBay and am looking forward to being able to do a similar mixer beat test with the GPS-disciplined OCXO to ensure that the PIC firmware is working correctly!