MPC5744P etimer frequency measurement with cascade counter

[YL005]

Hi

I implement frequency measurement function by using etimer module with cascade counter

The calculated frequency goes wrong sometime with the below CAPT1/CAPT2 counter
The second element counter in capture_ch0 is obviously full and starts counting from zero.
So the second element in capture_ch1 should be 1 greater than the others, but the picture below shows that the fourth element in capture_ch1 is also 1 greater than others.

From the reference manual of MPC5744P I think all related counters should be frozen in the Hold register.I'm wondering why the counter I read looks like this.

The SDK version I use is as follows. If I need to provide the SDK configuration or frequency calculation method, I can provide it.

** Processor : MPC5744P_144
** Component : etimer
** Version : Component SDK_S32_PA_11, Driver 01.00, CPU db: 3.00.000
** Repository : SDK_S32_PA_11

[PetrS]

Hi,

a meaning of your buffer should be below, I think, at least from the way CAPT registers are read out

So I think 32bit values for consecutive edges increments, if this gives correct result I cannot comment.

BR, Petr

[YL005]

Hi,

I think the value of 4.edge are weird.

If the counter in capture ch0 is less than 65535, why the 4th counter in capture ch1 is 1148 rather than 1147?

[PetrS]

Hi,

3rd edge value is also 1148 so 4th should not be lower, but seems 2nd is lower then 1st.
You can try to stop capture logic after 4 edges are get and start it again after reading/calculation is done.
What is the signal measured and thus expected values/differences between edges? 

BR, Petr

[YL005]

Hi Petr,

The signal under test is usually between 100Hz to 1kHz with 50% duty. Therefore, I'd expect the value between each eage should be the same. Most of time the captured values ​​are correct, but sometimes an exception occurs like the situation mentioned above.

[PetrS]

Hi,

ok, so if frequency is not changed during single measurement, then captured values should be correct.
Not sure what to comment more, I would recommend to have latest SDK. Maybe try to add DMA to read results and see difference.
NonSDK code is posted on 
https://community.nxp.com/docs/DOC-104285
https://community.nxp.com/docs/DOC-330571

BR, Petr

[YL005]

Hi,

Thanks for your suggestion, I will check if the latest SDK can solve this problem.

I would like to ask if there are any SDK-related examples, because I have already seen non-SDK examples, but I have some questions about configuring DMA in the SDK.

[PetrS]

Hi,

I am not aware of SDK example.

BR, Petr

[YL005]

The input signal is measurement from F[0] (eTimer_0 Channel 2)

I cascade the eTimer_0 channel 1 and channel 2, the register of eTimer_0 is as attached files.

The code to get the counter is as below.