Calibration of FRO in LPC546xx

[carstengroen]

On the LPC546xx devices, we can select between 48MHz and 96MHz FRO (and 12MHz etc).

In the datasheet, it is mentioned that "the FRO is trimmed to +/- 1%" but in the Users Manual it is outlined as if it is only the 96 MHz that is factory trimmed. However, when selecting between the 48 and 96 MHz FRO I see that the FROCTRL trim values changes, so I guess that both FRO's are indeed factory trimmed. Can you confirm this ?

The background is we see deviations of up to -4% on the 48 MHz FRO which we use, but so far, only on devices where the PCB has been potted. I wonder if the mechanical "changes" somehow can influence the FRO calibration (if the 48MHz FRO is calibrated from factory)

Datasheet:

Users Manual:

[carstengroen]

Thanks @Harry_Zhang , so all the FRO oscillators are calibrated from factory. We see some deviation (3 to 4%) on devices (that are potted), we will figure something out with regards to calibration (using the FROCTRL register).

Thanks for your input!

[Harry_Zhang · Accepted solution]

Hi @carstengroen 

"However, when selecting between the 48 and 96 MHz FRO I see that the FROCTRL trim values changes, so I guess that both FRO's are indeed factory trimmed. Can you confirm this ?"

Yes, you are right.

The FRO (12/48/96 MHz) of LPC546xx is factory calibrated (± 1%).

You can refer to these chapters.

"Why you can still measure −4% on some boards?"

Even if the factory has been adjusted by ± 1%,
Environmental factors, especially encapsulation, can still cause significant additional deviations in internal RC/ring oscillators.

Hope this will help you.

BR

Harry

[carstengroen]

@Harry_Zhang , do you have some more information on how the encapsulation can influence the FRO frequency?

Has it to do with mechanical stress on the CPU or something else ?

We see that using potting compound, the FRO will change (go lower in frequency) with about 3 to 4% compared to "non potted" devices.

[Harry_Zhang]

Hi @carstengroen 

It is normal for the FRO frequency of LPC546xx to shift by 3-4% due to encapsulation, and it is not a chip or calibration issue.

main cause:
1. Mechanical stress: Curing shrinkage of sealing material  Delay change of ring oscillator Frequency decrease
2. Changes in thermal environment: After encapsulation, the chip becomes hotter, temperature can decrease by 2-5% with a few degrees of temperature increase
3. Parasitic capacitance and changes in power environment: can also cause small offsets
The impact of encapsulation on internal RC/ring oscillators is a common phenomenon in the industry.

If high precision is required:

I think you can use an external crystal oscillator
or measure and calibrate FRO Trim with software after encapsulation.

Hope this will help you.

BR

Harry

[carstengroen]

Thanks @Harry_Zhang 

Yes, understood. We can easily overcome this in our application by trimming, it seems that it is not (at all) temperature dependent what we see, so thats a good thing

We will do some more testing here.

Thanks for your input!