TCXO/VCTCXO - frequency stability in intermittent sunlight (thermal shock?)

[wb0gaz]

I'm using a VCTCXO for the first time in a radio project - the VCTCXO provides frequency reference for a PLL synthesizer IC. The VCTCXO is around 20 MHz; the synthesizer outputs about 1 GHz (so VCTCXO output is multiplied by 50). The VCTCXO is a small SMD device attached to a PC board.

Here's my question/problem - I took my (unpackaged) circuit board outside (it's part of a receiver and I wanted to see how it would receive an indoor signal generator's signal) and discovered whenever the sun directly hit the VCTCXO part that it's output frequency suddenly shifted (took <1 second to shift); when the sun was shaded from the device, it went back to the original frequency (again took <1 second to shift.)

Here's my question - is a TCXO/VCTCXO susceptible to thermal shock (which I presume is what's happening when the sun hits it directly?) I didn't expect this behavior - does it sound normal?

[iMo]

What else do you have on the pcb?
How much does it shift?

[wb0gaz]

Nothing near the part that has appreciable power/current command, and nothing within maybe 8mm of the part in any direction. The part is powered by a 78L33 regulator (which has almost nothing else to do - just a microcontroller with sub-milliamp current demand.) The shift I've seen is about 5 kHz at the PLL output (which implies about 100 Hz at the TCVCXO output.) The only other culprit I can think of would be abrupt temperature change affecting the voltage control input (perhaps poor solder connection). My main reason to post was to get a sense of whether TCXO in general has a time characteristic when faced with temperature change (perhaps gradient between the crystal and the time compensating circuitry?) so I can get a sense of whether this is something I should deal with from a design perspective (obviously I won't use the device with sun exposure like this, but it did raise a general point of curiosity), or whether this is unexpected behavior, in which case I'll chase down construction blunders and the like.

Thanks for the quick follow-up question!

[SiliconWizard]

I would expect the compensation loop of a common TCXO to have limited abilities in reacting to very fast temperature changes, but as long as the temperature stabilizes within the rated range of the TCXO, I'd expect the frequency to get back to its nominal value. Many TCXOs are rated for up to +85C and I doubt the temperature would raise above that in like 1s in direct sunlight, so I'd expect the shift to at most be only very temporary.

Are you sure it's a temperature problem? Is the TCXO you're using hermetic or not (some TCXOs are not)? It may also be just reacting to UV light. To check that, you could just cover the TCXO with a piece of black paper or black tape. Being black, it would heat up but reasonably block UV light. If the frequency doesn't drift, then you know it's not because of the temperature. Just a thought.

[wb0gaz]

Thanks - this is really helpful - much appreciated.

The device appears hermetic (it's SMD in a 5x7mm class package and visually there's no sign of a way in.)

The device definitely returns to intended frequency once it's in the shade (it takes only a fraction of a second to get to the displaced frequency, then a fraction of a second to recover once sunlight removed.)

Thinking about cumulative duration of exposure - it does stay displaced (with no further change) while it's in sunlight (that is, it changes frequency when entering sunlight, stays there for the duration of the sunlight, then changes back when removed from sunlight.)

I hadn't thought about UV vs IR - that's a cool idea and going to go on my to-do list (I'm in Colorado - plenty of UV!)

I'm sure it's not getting to 85C (given the fraction-of-a-second timeframe for the displacement). Also, the displacement and recovery durations are both comparable (say 1/4 of a second, just rough estimate).

Fascinating stuff!

[iMo]

Touch the chip with your finger - that increases the temperature of the package to aprox 32degC. You will see the freq shift if caused by the temperature.
I asked whether are there any other components on the board, like diodes in glass packages - those are sensitive to sun light.
Btw - doublecheck the datasheet of your TCVCXO. TCXO does not mean it will be rock stable..

[wb0gaz]

Thanks - I'll try the finger test next time I'm set up to do testing.

There are no light sensitive components I can think of in the circuit (sorry I mis-understood your question initially) - certainly nothing intentionally light sensitive.

The only possible spec that I could see applicable in the (very terse) datasheet is the following:

Frequency/Temperature characteristics Max. ±0.28×10–6

There is no mention of units of measure for this figure. (perhaps % change in frequency / % change in temperature, would be unit-less?)

In any event, as you point out, "does not mean it will be rock stable" is what I'm learning here!

[iMo]

In case it is the temperature characteristics - it means 0.28ppm/degC, in your case 20MHz * 0.28e-6 = 5.6Hz/degC

[mino-fm]

Sorry, but ±0,28 ppm/degC would be too bad for TCXO/VCTCXO. So I'm sure it means ±0,28 ppm over the whole temperature range.