[time-nuts] Temperature controlled TCVCXO

David davidwhess at gmail.com
Sat May 14 17:52:33 EDT 2016


As you pointed out earlier, with a TCXO, any inflection points of the
crystal itself will be unavailable because the temperature coefficient
at any one point will be the product of the crystal and the
temperature compensation circuitry.

I am left to wonder if this will work better with a TCXO or just an
XO.  The later would allow for a better optimized voltage control but
at that point, I will have just reinvented the ovenized crystal
oscillator using AT cut crystals and operating at room temperature.

Do OCXOs use internal temperature compensation other than that
provided by the oven itself?

>Bottom line: A $10 eBay OCXO is likely to beat an ovenized or cooled TCXO. 

Designing something which relies on the availability of often
questionable Ebay items just bugs me.

On Sat, 14 May 2016 08:54:36 -0400, you wrote:

>Hi
>
>If the crystal has a “most stable point” it will be a point that the slope goes
>from positive to negative (or vice versa). These points are symmetrical about 
>the “center” of the crystal curve. For an AT cut, your center will be just a bit above
>25C. How far above depends on the geometry of the blank and a few other things.
>For simplicity you see 25 used a lot.
>
>If you look at the curves on the link posted earlier they follow a pattern. Anything 
>below “zero angle” never goes through a slope change The zero angle goes flat
>at the center. The “useful” part of the curves have a turn at 25 +/- X where X can 
>be anything from 5 to 100C in normal crystal. 
>
>You need to dig into the actual math to take a look at the slopes near turn. Obviously
>the turn at 50C is a better bet for your OCXO than the one at 125C.
>
>The enemy of any temperature stabilization system is waste heat. On an OCXO
>with a normal heater, there is some (small) current in the circuit even when the heater
>is turned off. With a TEC, there is a *lot* of waste heat when cooling. The systems
>I have seen using them get into stacked devices and water cooling pretty quickly. 
>
>Bob


More information about the time-nuts mailing list