ad9371 ORX noise raised at frequency edges

I bypass the RFIR. The bulges become smaller. What I am guessing, these bulges are caused by the delta-sigma ADC noise shaping  effect.  But I still expect a flat noise level.

Yes these bulges are caused because of the noise shaping of the sigma delta ADC. You can flatten the noise floor by increasing the RX path gain or by running the ADC at a higher sampling rate.

Refer below links:

https://ez.analog.com/wide-band-rf-transceivers/design-support/f/q-a/80236/ad9361-iio-oscilloscope-noise-floor/166092#166092

https://ez.analog.com/wide-band-rf-transceivers/design-support/f/q-a/79015/ad9361-wide-bandwidth-noise-floor-shape-issue/106756#106756

Thanks Srimoyi 

According to your suggestion, I increase the Orx gain to 255. But the noise shape looks similar to the case with 10dB less gain as show in following figure.

I use the standard profile of 245.76MHz IQrate to capture 200MHz signal. I am wondering, the bulges are almost 20dB higher than the noise closed to the signal. 

Furthermore, i check the ad9371model in matlab with same ORX profile and get a different pic. the bulges are not higher. 

Maybe I should set the ADC bandwidth，but I do not find any example related to it.

cheers

Sorry for the delay in response. Somehow missed this post 

It is quiet strange that you are seeing these noise edges at the ORX output. We tried this on our evaluation platform and saw that the ORX output noise floor is flat.

You need not change manually anything in the config files. The correct method would be to 

generate profiles with required BW and sampling rate from the filter wizard GUI and then load them into the TES GUI and generate config files from the GUI. Use the same config files in your system. You can download the tooI from the below link:

https://wiki.analog.com/resources/eval/user-guides/mykonos/software/filters 

Dear Srimoyi

I am still struggling on this issue. 

A little bit background of my problem maybe helps for better understanding.

Old project is developed on zcu102 and ad9371. Only Tx and Rx with 122.88MHz IQ rate are tested. Orx also works well, I saw flat noise floor and high SNR. Right now I am carrying on this project for developing transceiver algorithm targeting 307.2MHz IQ rate for Tx and Orx IQ rate for maximal 250MHz bandwidth.  

I did generate the config file by TES standalone software.

According to my understanding I adapt the RF related settings as following:

1. obsrxfirCoefs  

2.orxProfile 

3 mykonosClock

4. obsRxSetting, 

At this point generated config sets the ORS_RXOFF? Why? The Orx does not work with it. The output of Orx is a DC signal.

Other settings which are AGC, framer related are not changed. 

Furthermore only the config.c is used to modify the corresponding file in old project. 

If help I can post the config.c in my project.

thanks  

Furthermore, I try following your suggestion. At first config the ad9371 in Matlab and generate a profile.txt.

But by loading it to TES GUI I get the error of error parsing clock settings

these are first few lines of the profile.txt

The TES GUI does not support programming with all the DEV_CLK frequencies.  Generate the same profile with DEV_CLK frequency of say 122.88MHz or say 153.6MHz, then change the JESD or the ORX settings as required in the GUI, and then generate config files from the GUI.

If the ORX path is OFF after power up, try running the below API to enable the Obsrx path

I tried with different settings, but still got same error by loading profile into TES.

can you please provide a snapshot of valid settings, or send me a valid profile. I am not sure, if it is a bug of the software

Please use the latest version 1.10 of the filter wizard and then generate the profiles. Download it from the below link:

https://wiki.analog.com/resources/eval/user-guides/ad-fmcomms2-ebz/software/filters 

Attaching the profile  below with the same sampling rate and BW that you had used.

8662.Ezone.zip 

Hi breadjyy ,

We have also observed same thing in our design. Tx and ORx at 245.76MSPS. Would like to know if your Tx output is clean or are you observing any spurious at offset of 122.88MHz or 245.76MHz from LO. If possible share the spectrum snapshot with 1GHz span. How does the ORX plot look if sine wave is given as input?

Thanks