NRF52840 32MHz crystal offset parameter recommendation

[Frank Yang]

Hello, Nordic!

We are old customers of Nordic NRF52840. We have been using NRF52840 since the promotion of NRF52840. We use NRF52840 to make some IoT modules for our customers, mainly for some Tracker and industrial monitoring markets.

But we have some troubles in production. The passing standard of the 32MHz crystal test we made is that the frequency deviation is ±10ppm. But it is difficult to meet, because the crystal is greatly affected by the temperature and the test equipment, and it is easy to exceed the standard, resulting in many defective products (4%).

The NRF52840 datasheet reference design recommended 32MHz crystal requirement is total ±40ppm. Bluetooth low energy，applications, packet length > 200 bytes，the specification recommended crystal  total ±30ppm.

I would like to know:

1. Under what circumstances is it necessary the 32MHz crystal to achieve ±30ppm? Which applications scenarios require packet length > 200 bytes?

2. For general customers, what is the frequency offset requirement for the crystal? Because we also need to consider the influence of the aging of the crystal and the service life of the product.

3. Can we set the crystal frequency at ±15ppm? What are the risks in using it?

Thanks！

Frank Yang

[Frank Yang]

In addition, this is the specification of the 32MHz crystal oscillator we use, how to calculate its total tolerance？

[Frank Yang]

The bill of material recommand crystal specification

[ketiljo]

1. Under what circumstances is it necessary the 32MHz crystal to achieve ±30ppm? Which applications scenarios require packet length > 200 bytes?

This is set by the Bluetooth test spec to make sure interoperability between radios from different manufactures are handled well. If you want to increase the datarate by sending longer packets, you need to be able to send long packets.

2. For general customers, what is the frequency offset requirement for the crystal? Because we also need to consider the influence of the aging of the crystal and the service life of the product.

You have to include tolerance, stability and aging. The three parameters needs be be 30 ppm, or 40 ppm in case you only use short packets, or better. 

3. Can we set the crystal frequency at ±15ppm? What are the risks in using it?

Unless the stability over temperature is more than 10 ppm, you should be fine with this. 

Frank Yang said:

In addition, this is the specification of the 32MHz crystal oscillator we use, how to calculate its total tolerance？

For this crystal, you have tolerance 7 ppm, stability 12 ppm and aging 1 ppm, meaning a total of 19 ppm plus 1 ppm / year. More than good enough!

[Frank Yang]

Hi Ketiljo

Thanks for your reply！

3. Can we set the crystal frequency at ±15ppm? What are the risks in using it?

Unless the stability over temperature is more than 10 ppm, you should be fine with this. 

---------------------------------------------------------------------------------------------------------------------

How to understand this? The crystal we are using now has a frequency stability of ±12ppm. Is the condition to set the test pass parameter to exceed ±10ppm not met?

[Frank Yang]

In addition, this is the specification of the 32MHz crystal oscillator we use, how to calculate its total tolerance？

For this crystal, you have tolerance 7 ppm, stability 12 ppm and aging 1 ppm, meaning a total of 19 ppm plus 1 ppm / year. More than good enough!

--------------------------------------------------------------------------------------------------------------------------------------------------

Considering that the product has been used for 10 years, the cumulative frequency deviation of 10 years will reach 10ppm. The sum is 29ppm, which is just ±30ppm to meet the needs of long packages. Is my understanding correct?