Generating QAM baseband I/Q file for ADRV9025 — normalization and RRC filter question

the transmitted spectrum looks flat (see attached image).

1A) The IQ file must be non-normalized and needs to be IQ tabbed

2A) ADRV does not have internal channel filtering, so, user needs to apply filtering on the waveform.

Example waveform files below:

ez.analog.com/.../lte-waveform-files

Thank you , 

I already applied an RRC filter in MATLAB before exporting the I/Q file, but the transmitted spectrum still looks flat (see image).
Could you please tell what could be the reason for this?

The image below shows the spectrum in MATLAB (after applying the RRC filter)

Thanks.

Try using the waveform that i shared in my previous reply and confirm that the setup is fine. What is the sample rate and waveform length- is it 10ms?

 I tried the waveform you shared and it gives a good result ,my waveform uses 983.04 MSPS with 307,200 symbols

That's the problem. You would have to have the sample rate of the waveform that is the same as that of the IQ rate of a given Use case/ profile and the length of the waveform will typically be 10ms which is one radio frame.

Thank you for your reply.

The profile I am using on the ADRV9025 has a sample rate of 122.880 MHz.
However, in my MATLAB code below, I am not specifying any sample rate. I simply generate the 16-QAM symbols and create the I/Q text file.

How should I apply this sample rate in MATLAB when creating the I/Q file?
Do I need to upsample/interpolate the symbols to 122.880 MHz, and if yes, what is the correct method?

Here is the code I am using:

In the filter , right ?

Please follow the below link for generating waveforms as needed.

 Test Waveform Generation 

Hello,

I generated a 16-QAM baseband waveform in MATLAB and loaded it into ADRV9025.

My understanding is that the symbol rate can be determined using:

Rs=\frac{Fs}{\text{samples per symbol}}

​where Fs is the IQ sample rate of the Tx profile.

Could you please confirm if this is the correct formula, and clarify which parameters it depends on (IQ rate, samples per symbol, filtering, Tx profile, etc.)?

Thank you.

Best regards,