I have a no-name OCXO with pins marked by seller (VCC, GND, RF OUT, Vadj). I have checked its output frequency against rubidium oscillator using the following method:

I connected them both to different channels of the oscilloscope, set triggering on the rubidium oscillator signal and adjusted the frequency of an OCXO by connecting negative terminals of the lab power supply together (to get a common ground) and changing one channel voltage between zero and +4V. At about 2.5V (exact value doesn't matter for my question) frequencies seem to match.

Now, I'm looking for a way to provide an adjustable voltage reference to control its frequency when this OCXO is placed in some device. My first guess was to add some resistance in series, but I failed - current consumption of Vadj pin is way too low. I was actually unable to measure it with my Fluke 289 in microamps range.

My second guess was to use an adjustable voltage regulator, for example LM317 rated at 100 mA. My concern is - is this LM317 precise enough for the purpose I described (assuming I use precise and temperature-stable resistors for adjustment)? It has minimum load current of 3.5 mA which is far more than I'm gonna have, relatively high temperature drift and 0.1%/V line regulation.

If using LM317 for that purpose is not a good idea, I would be grateful for any better solution.

  • 2
    \$\begingroup\$ You need to define your stability goals, if you have any. Unless you use a frequency counter, this sounds like a bench project, not something specific. \$\endgroup\$ – Sean Houlihane May 13 '16 at 21:40
  • \$\begingroup\$ It sounds like you just need a simple trimmer potentiometer, but you need to be clear about what your performance goals are. \$\endgroup\$ – Dave Tweed May 13 '16 at 21:46
  • \$\begingroup\$ I would like my performance goals to actually match my OCXO characteristics in terms of short- and long-time drift. Can't fully match its temperature stability as it is heated, of course but the potentiometer together with all the rest of the circuit for frequency adjust is going to be physically close and thus heated a little bit. \$\endgroup\$ – Alexey Malev May 13 '16 at 21:50
  • \$\begingroup\$ @SeanHoulihane I am going to use OCXO in a homemade frequency counter in the future. In that case I'm gonna feed in rubidium oscillator output and adjust OCXO frequency to match 10 mhz. For now, I just try to make both waveforms do not move relativeley to each other on oscilloscope screen. \$\endgroup\$ – Alexey Malev May 13 '16 at 22:00
  • \$\begingroup\$ There are chips you can buy that are called "precision voltage references". They work somewhat like an LM317, but with better accuracy (and less output current capability). \$\endgroup\$ – The Photon May 13 '16 at 22:21

Don't use an adjustable reference, use a static 5V reference and a resistive divider (using an adjustable ~4k7 trimmer). This should be good enough for what I think you describe. If phase noise matters, put a capacitor on the Vc input.

  • \$\begingroup\$ Static 5V reference or a linear regulator like LM1117 or 78L05? \$\endgroup\$ – Alexey Malev May 13 '16 at 23:14
  • \$\begingroup\$ @AlexeyMalev choosing a specific reference-type device is usually just a high stability, low power linear regulator. For initial testing, you would be just fine using a 7805 as your voltage reference, but remember that you may not have the long-term stability you want. \$\endgroup\$ – user2943160 May 14 '16 at 4:46
  • \$\begingroup\$ Tried that today. Works exactly as I wanted, thanks. \$\endgroup\$ – Alexey Malev May 18 '16 at 20:21

Most of those OCXOs also have a VREF pin which supplies a reasonable reference voltage of around 3.9 V (zener-based) for you to connect a 10-20k pot to. If yours doesn't have that, then use your LM317 -- load it with 100 ohm or so to keep it > 3.5 mA; connect a 20k pot across it (10-turn is better) to VADJ.

Or just use a 100 ohm and a 3.9 V zener, powered from the 5 V input. If you thermally connect the zener to the OCXO, its temperature will be reasonably regulated and therefore the V slightly more stable.

For reference, 1/4 turn on a 10-turn pot changes my OCXO by less than 1 ppb. With a 4 V reference, this represents about 100 mV, so stability of the reference isn't particularly critical.

  • \$\begingroup\$ +1 particularly for multi-turn trim potentiometer. 10 to 25 turns makes initial tuning of the OCXO quite nice to get close enough to the desired frequency, then it'll mostly stay there. Also useful: a thermal connection between the OCXO can and the potentiometer. \$\endgroup\$ – user2943160 May 14 '16 at 4:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.