Crystal temp range: is there any meaning in -40C versus say -10C?

[peter-h]

Assuming a sealed ceramic crystal, is there anything physical of relevance?

Or is it just a number in a data sheet, allowing the -40C specced one to be sold for more money?

This one is one of many -10C examples
https://www.mouser.co.uk/ProductDetail/Murata-Electronics/XRCGB25M000FAN00R0?qs=r3UZegS7eeqcM%2FlpH6ntCA%3D%3D

[ajb]

"Physical of relevance" to what?   If you need a part that you can trust to operate down to -40C, then you need to use a part spec'ed for that, if not, then you don't. 

The part with the wider temp range may or may not be substantially different from the other one, but the only guarantees are what the datasheet says.  It may be that the parts have no real differences except that one is tested and qualified to a higher standard (which actually is more expensive, hence the higher price), and the lower spec will work fine at -40C.  It may even be that the manufacturer will bin parts based on performance/demand, and you may order the lower spec part and get a higher spec part if the manufacturer overproduced those and down-binned them, but there's no guarantee that will happen every time you order them. 

Using a part outside of the limits given in the datasheet may cause immediate failure, or it may work fine, or it may work fine with *this* batch of parts but not *that* batch of parts, or 99.9% of parts may work but that last 0.1% may fail at the worst possible time.  If it's a one-off project that lives on your bench, sure, give it a try, not a big risk if it breaks.  If you're shipping a million units out all over the world then you better do your homework to make sure that you're choosing parts that will be reliable, and you go outside the datasheet limits exclusively at your own risk. 

[jimmc]

In general the temperature coefficient of a quartz crystal is not linear. For instance an AT cut crystal has a cubic frequency deviation vs temperature curve.
By using different cutting angles it is possible to minimise the error over a specified temperature range. A crystal cut for operation over -10 to 70'C will have a large error at -40'C.
Conversely a crystal specified to -40'C will have a worse tempco around room temperature. Hopefully fig 4 of https://coloradocrystal.com/applications/ will make this clearer.

Jim

[TimFox]

Sealed packages (especially with metal-ceramic joints) can have difficulty with the difference in thermal expansion co-efficients stressing the joints between dissimilar materials with temperature change.  Therefore, if one really needs -40 C rating, the part must be tested down to that temperature for safe use.  This is not an absurd temperature:  the winter temperature at the US-Canada border (49 deg N) can go down that far.

[peter-h]

Problem is, I am seeing what looks like exactly the same part, with a different spec.

It looks like a marketing differentiation exercise, only.

[ajb]

Sure, it could just be marketing, but is that a bet you wanna make?

It may even be the exact same part (design/materials/manufacturing) except with some adjustments for higher grade raw materials or more tightly controlled processes, which would certainly make them more expensive to manufacture.

The point is, you don't know, and "well, I thought the datasheet specs were just marketing hype" isn't gonna stop them falling if that's what they're going to do.

[TimFox]

If the same unit off the line, in order to be sold as a premium part, needs to be measured or certified over a wider temperature range, then it will probably cost more than the part number for the same thing guaranteed over an easier temperature range.  Why is that the fault of marketing?