KEA64 IRC and FLL accuracy

[emanuelwahlqvis]

In order to decide whether the IRC of the KEA64 MCU is good enough for my application I have tried to find the accuracy of the bus clock (derived from the system clock) which drives the UARTs. I have a requirement to communicate at 115200bps with an allowed inaccuracy of +/-2%.

Using the IRC->FLL with IRC @ factory trimmed value 31250Hz is not an option as the baudrate will be far off even att room temperature.

So, questions is if the accuracy of the FLL stated in the datasheet (min:-2.3% max:0.8%) is relevant when using an external crystal? OR is the stated accuracy including the IRC inaccuracy of min-0.8% max: 0.8% (min: -1% max:0.8% when trimmed @ 25°C)?

I would also want to know, when changing the IRC frequency through the trim value, is the FLL accuracy affected?

I found the app note https://www.nxp.com/docs/en/application-note/AN5300.pdf  which suggest that the FLL output has a guaranteed accuracy of +/-4%. Where does this number come from?

Best regards

Emanuel Wahlqvist

[jorge_a_vazquez]

Hi

So, questions is if the accuracy of the FLL stated in the datasheet (min:-2.3% max:0.8%) is relevant when using an external crystal?

Do you refer to the "Frequency accuracy of DCO output using factory trim value"? if so, as it is named, this value was taken using the factory trim value.

is the stated accuracy including the IRC inaccuracy of min-0.8% max: 0.8% (min: -1% max:0.8% when trimmed @ 25°C)?

The accuracy of -0.8% and 0.8% is the accuracy of the IRC, it doesn't take into account that it will pass through the FLL

I would also want to know, when changing the IRC frequency through the trim value, is the FLL accuracy affected?

No, it should remain the same.

Hope this helps

Best regards

[emanuelwahlqvis]

Hi,

Thank you for looking into this issue. Please let me clarify my question:

So, questions is if the accuracy of the FLL stated in the datasheet (min:-2.3% max:0.8%) is relevant when using an external crystal?

Do you refer to the "Frequency accuracy of DCO output using factory trim value"? if so, as it is named, this value was taken using the factory trim value.

- Yes, I was referring to that value. Then what is the accuracy of the FLL when using an external crystal to feed it? And do I interpret it correctly that if using the IRC to feed the FLL and the IRC frequency is changed with the trim value, the total accuracy of the FLL output will not be changed due to the lower IRC accuracy?

Best regards

Emanuel W

[jorge_a_vazquez]

Hi Emanuel

Sorry for the late reply.

Then what is the accuracy of the FLL when using an external crystal to feed it?

Unfortunately, this value wasn't measured and so, it wasn't included in the datasheet specification, in those cases, you have to check crystal characteristics.

if using the IRC to feed the FLL and the IRC frequency is changed with the trim value, the total accuracy of the FLL output will not be changed due to the lower IRC accuracy?

No, FLL accuracy shouldn't change.

Best regards

[emanuelwahlqvis]

Hi,

Thank you for the clarifications. I do have another question which I cannot find an exact answer to:

If one reads the factory setting for the SCTRIM as from ICS_C3 and adjusts this with a fixed offset as show in this code snippet:

// Adjust the internal oscillator frequency with this fixed offset
#define FREQ_OFFSET_ADJUSTMENT 10U
ICS_C3 = ICS_C3 - (ICS_C3_SCTRIM(FREQ_OFFSET_ADJUSTMENT));‍‍‍‍‍‍

Would the resulting frequency be the same for every device (as they all are calibrated for the same IRC frequency and the offset is fixed) or will it vary due to the non-linear behavior of the SCTRIM adjustments?

Best regards

Emanuel W

[jorge_a_vazquez]

Hi Emanuel Wahlqvist

No, this value is not the same for every device, all devices are calibrated to the same frequency, but you cannot apply the same formula, because every device varies in different ways.

Best regards

Jorge Alcala