Hi One thing to watch out for on any OCXO, but more on ones with crazy large tuning ranges:
As the OCXO changes temperature, it’s current draw changes. It’s amazing just how much voltage drop you get in some seemingly well laid out boards when this occurs. The gotcha is that a “bias it up” with a couple resistors approach may or may not help this. It depends on just where the resistors are on the board. On some specs, they go to the extreme of specifying a point on the ground lead (after soldering in a board) that is the “official” ground for the oscillator. Good luck finding that on a data sheet out in the wild … If you have a 0-5V tuning range and a ~2:1 sensitivity variation over that range and a 0.25 ppm EFC: 0.5 ppm / 5 = 0.1 ppm / V. A 10 mv shift (which is not unusual on a 5V OCXO) gets you 1x10^-9 of frequency change. If that’s over 0-50C it’s contributing 2x10^-11 / C. Small changes in air flow quickly change the oven current so the loop may be having fun …. Bob > On Apr 5, 2019, at 10:45 PM, Jim Harman <[email protected]> wrote: > > Thanks for your ideas, Ed. In my actual implementation I am using > reasonably low tempco 1% fixed resistors. In the photo they are the blue > resistors between the DAC and the OCXO. > > I have noticed a significant temperature sensitivity in the system, > especially if I run it with the cover off.so the ambient temperature can > change quickly. Moving air has a short term effect on the OCXO and to a > lesser extent the GPS. Sustained ambient temperature changes cause an > offset in the DAC value when the loop is closed, so the loop is doing its > job to compensate for them. Putting the system in a small picnic cooler > with several water bottles makes any temperature changes slower than the > loop time constant, which should minimize temperature effects on the > overall system. > > I calculate the open loop temperature sensitivity at about 7e-11 / deg C. > The C-MAC oscillator spec for what it's worth is "-10 C to +70 C / ref.at > +30 C < +/- 3.0 e-9". If the variation was linear, which I realize it is > probably not, that would be 7.5e-11 / deg C so I am in the same ballpark. > > On Fri, Apr 5, 2019 at 8:21 PM ed breya <[email protected]> wrote: > >> I'd recommend that once you get things figured out and tuned up to >> nominal running conditions, you should narrow the offset pot range, and >> use good low-TC resistors to make up most of the network R, with the pot >> having as small an effect as practical. BTW I don't see the pot in the >> pictures, but I do in the schematic. >> >> You may want to consider lowering the entire network resistance by >> scaling everything down, say ten times lower or more. This would reduce >> noise, and lessen effects from the varicap bias (leakage) current in the >> OCXO. Also, I have seen a number of OCXOs with an internal termination >> resistor (like 50-100 k) on the tuning line - that has spoiled a lot of >> fun for me, having to worry about the characteristics of that resistor, >> and including it in the deal. With an unknown or unspecified OCXO, it's >> good to check for any unwanted extra parts. >> >> It may help the stability to put some insulation around the tuning >> resistor network and maybe the DAC too, especially if the waste heat >> from the OCXO is significant. >> >> Ed >> >> _______________________________________________ >> time-nuts mailing list -- [email protected] >> To unsubscribe, go to >> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com >> and follow the instructions there. >> > > > -- > > --Jim Harman > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
