CCS/MSP430F6779A: Frequency Deviation for Crystal Frequency Temperature Compensation

Hi,

looking at the table K is B, means is the parabolic coefficient and T0 would be in this case the Ti, means turnover temperature. This is the temperature of the maximum frequency.

B cannot be given by the crystal datasheet, as this is a setup and unit dependent value. This means you have to measure it in your setup. This is on one hand the ppm deviation at room temperature resulting from crystal sample tolerance, and on the other hand from the actually applied effective load. Keep in mind the nominal frequency is only reached with 0ppm tolerance of the sample and applying exactly the nominal load capacitance. Any deviation from the load capacitance results in more or less bigger frequency deviation. e.g. order of magnitude 9ppm/pF. Though if you want to know ti accurately you need the so called pullability of the crystal, which is exhibiting approximately hyperbolic characteristic. This can be only obtained by measurement or directly from crystal manufacturer.

Keep in mind the effective capacitance is consisting of capacitors applied in form of components, but also driving device pin capacitance and other stray capacitance values.

When measuring the crystal frequency, please don't measure the analog signals, as you would shift the frequency by the applied probe capacitance, but measure the resulting digital clock, e.g. ACLK.

Best regards

Peter

Hello,

please let us know, whether there is something else we can do on this for you. Many thanks in advance.

Best regards

Peter

Hi Peter,

Sorry about lately reply.

I will try to contact crystal manufacturer for frequency deviation value. However I have some thing wants to ask you. It is about frequency measurement as you mentioned.

The point is I can't figure out how to measure this. Do you have any design guide or material about frequency deviation measurement?

And from your replied as above. Can I confirm with you that Frequency deviation is in unit of PPM?

Thank you,

Nattapol

Hi Nattapol,

the frequency measurement would be pretty simple. Switch the ACLK being sourced from the 32.768 crystal to a GPIO. Connect the signal to a frequency counter and read the frequency value from the display. Of course, you need to measure at the same time the temperature. Thus, if you have a temperature chamber, you could also measure the characteristic of the used crystal yourself, instead of calculating it. The characteristic should be for the crystal devices of the same type pretty much the same.

You should not measure the signal directly at the crystal, as this would influence the frequency.

Not sure what you're aiming for with your last question. The frequency deviation of a crystal is usually given in ppm, and the values of a 32kHz tuning fork crystal over industrial range are in the ball park of ~180ppm.

Best regards

Peter