oxco need to warm up?

[Awesome14]

I have a 10Khz oxco. I connected it to power and a DMM, but the frequency is all over the place. Do I need to let it warm up?

[suicidaleggroll]

OCXOs generally take at least a few minutes to warm up and settle down.  You'll need to wait hours for them to stabilize fully, but you should be within a ppm of the final frequency after <10 min, depending on the OCXO of course.

[Ian.M]

Of course.  The crystal oven has to thermally stabilise before the frequency will settle.  If its got a status output, monitor it.  Otherwise look at the supply current - it should either settle to a steady state or cycle with a regular period and duty cycle once the oven is up to temperature.  Once its reached steady state its going to need some time for all parts in the oven to reach thermal equilibrium, and it needs to be protected against drafts and changes in ambient temperature. Suicidaleggroll has suggested the timescales involved. If it isn't then stable and within its specified tolerance when checked with a calibrated frequency counter of at least ten times greater accuracy than the OXCO tolerance, its got problems.

However it shouldn't be grossly off frequency even shortly after startup - if you are seeing more than 0.1% error its certain that either its broken or your measurement setup is defective.  Most cheaper DMMs don't have high accuracy frequency ranges.  Even a good bench DMM is likely to have vastly inferior frequency accuracy than a comparably priced dedicated frequency counter so check your DMM's specs and try it on another stable frequency source in the KHz range

[Tom45]

What Ian.M said.

Plus, it might be that the DMM's frequency counter isn't able to reliably trigger on the oxco output.

[Awesome14]

OK, thanks.

[babysitter]

A 10 kHz OCXO sounds like a massive brick of silicon dioxide if it's oscillating at that frequency, if it isn't using a higher frequency crystal and divider. 10 MHz would sound (pun intended) more plausible. Turn your DMM to DC volt (should read a stable value which might be zero) and then AC VOLT (should read a stable number, different from DC V.) If it finds AC your DMM has at least the opinion that a alternating current is there, the manual tell you about its bandwith and so on. The correct tool to check it would be a oscilloscope which should show a somewhat periodic waveform.

But as it is Awesome14 asking... (stocking up popcorn)

[ogden]

But as it is Awesome14 asking... (stocking up popcorn)

Can't be him! He's an expert who made famous ultra-stable reference which does not oscillate - because of flat heatpipe, perfboard and very specific soldering technique used. Most likely his account was hijacked. So sad.

[james_s]

Either way I would expect that just about any OXCO is going to be orders of magnitude more stable and accurate than the frequency counter function in a DMM unless something is really wrong with it. In order to properly test the OXCO you need something significantly more accurate to compare it against. The shortwave station WWV and more recently GPS receivers are often used as an accurate frequency reference. A frequency counter with it's own internal OXCO should be good enough to tell you if something is horribly wrong.

[bson]

OXCO = Oxen delivered in two days or less, or it's free! 

[coppice]

A lot of the better ovenised oscillators have an output  signal to tell you if the chamber temperature has stabilised. Its bad that this is not a universal feature. The frequency can be a long way off when the chamber is cold. If you use this to feed a transmitter you could be pumping out lots of power on someone else's channel.

[BradC]

A lot of the better ovenised oscillators have an output  signal to tell you if the chamber temperature has stabilised. Its bad that this is not a universal feature. The frequency can be a long way off when the chamber is cold. If you use this to feed a transmitter you could be pumping out lots of power on someone else's channel.

I recently powered up an ex-Trimble 10MHz OCXO. It was within 3PPM in about half an hour but took the best part of 36 hours to really settle down.
Having said that it was probably removed from the PCB by an 8 year old with a propane torch, so who knows the stresses it has been exposed to.
It's now wandering about 0.01 Hz as the room temp and humidity moves around (not using any EFC on it at the moment). I can't measure it any more accurately than that. For the first 18 hours or so it was wandering over about 1Hz.

The Trimble has no monitoring output to let you know it's happy.

[thermistor-guy]

...
I recently powered up an ex-Trimble 10MHz OCXO. It was within 3PPM in about half an hour but took the best part of 36 hours to really settle down.
Having said that it was probably removed from the PCB by an 8 year old with a propane torch, so who knows the stresses it has been exposed to.
It's now wandering about 0.01 Hz as the room temp and humidity moves around (not using any EFC on it at the moment). I can't measure it any more accurately than that. For the first 18 hours or so it was wandering over about 1Hz...

The Agilent frequency counters that I've used typically need 24 hours or more to settle, after an extended power-down (weeks or months). I usually let test equipment run for 24-48 hours before using it for any important tests. That gives time for various factors to stabilize, like temperature gradients in the equipment, and thermally-induced mechanical stress in PCBs and components.

BTW, I've noticed something similar in precision analog circuits that I've constructed/soldered. The high temperatures from soldering shift the component values slightly, and it can take a day or more for the components to return to their pre-soldering values.