How to write AD9361 FIR filter coeffs on no-OS driver?

Please refer to the below of FIR Filter wizard and guidelines.

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

Thank you for your comment.

I set FIRcoeffs below.

In the FIRcoeffs, Assuming Fpass is 1 MHz, Fstop is 2 MHz and input-data-rate is 20MHz.

Is there a problem with this coefficient?
Moreover, is there anything else syntax required on the no-OS driver.?

【FIR coeffs】

AD9361_TXFIRConfig tx_fir_config = {
3, // tx
-6, // tx_gain
1, // tx_int
{-77,-39,-189,-185,-321,-325,-375,-287,
-172,78,362,709,1001,1210,1219,1001,
504,-224,-1117,-2016,-2748,-3094,-2873,-1934,
-232,2194,5171,8444,11674,14508,16611,17732,
17732,16611,14508,11674,8444,5171,2194,-232,
-1934,-2873,-3094,-2748,-2016,-1117,-224,504,
1001,1219,1210,1001,709,362,78,-172,
-287,-375,-325,-321,-185,-189,-39,-77,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0}, // tx_coef[128]
64, // tx_coef_size
{1280000000,160000000,80000000,40000000,20000000,20000000}, // tx_path_clks[6]
4011543 // tx_bandwidth
};

Are you generating RX filter coefficients and loading them into AD9361? are you keeping receiver FIR coefficients in the Rx A filter structure AD9361_RXFIRConfig rx_fir_config?

kazucd57 said:

However, the FIR filter does not work correctly on the chip.

can you please give more details on this?

RR4 said:

Are you generating RX filter coefficients and loading them into AD9361?

>>Yes. setting RXFIRcoeffs below.

==================================================================

AD9361_RXFIRConfig rx_fir_config = {
3, // rx
-12, // rx_gain
1, // rx_dec
{-91,-62,-187,-281,-309,-477,-384,-461,
-197,-78,348,606,1025,1172,1277,1000,
553,-251,-1142,-2128,-2881,-3300,-3083,-2167,
-419,2052,5127,8493,11837,14762,16942,18101,
18101,16942,14762,11837,8493,5127,2052,-419,
-2167,-3083,-3300,-2881,-2128,-1142,-251,553,
1000,1277,1172,1025,606,348,-78,-197,
-461,-384,-477,-309,-281,-187,-62,-91,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0}, // rx_coef[128]
64, // rx_coef_size
{1280000000,160000000,80000000,40000000,20000000,20000000}, // rx_path_clks[6]
4034482 // rx_bandwidth
};

==================================================================

RR4 said:

can you please give more details on this?

>>

Problem is that the bandwidth of the RF signal is not narrowed.

Please look at the graph below.

In this graph, red-line(reference) shows the target power spectrum using the MATLAB/Simulink simulation.

However, the output of the AD9361 in the current configuration is wideband like black-line.

Please let me know if you need any other information.

what is the signal bandwidth used in the above image?

kazucd57 said:

However, the output of the AD9361 in the current configuration is wideband like black-line.

where did you measure the AD9361 output, is the Tx output plotted against MATLAB FIR response?

RR4 said:

what is the signal bandwidth used in the above image?

>> The signal bandwidth means occupied-band-width(OBW) for each line of the image. 

RR4 said:

where did you measure the AD9361 output, is the Tx output plotted against MATLAB FIR response?

>>The measurement point is the output of an RF circuit, not the AD9361 output. In reality, the RF circuit (including BPF and amplifier etc.) is connected after the AD9361.

can you please confirm the sampling rate of the FIR filter and the waveform sampling rate, form the above graph it appears that the sampling rate is mismatched between the FIR filter and the waveform.

Can you please share the FIR filter coefficients in .ftr format?

RR4 said:

form the above graph it appears that the sampling rate is mismatched between the FIR filter and the waveform.

>>OK, I will check it soon.

RR4 said:

Can you please share the FIR filter coefficients in .ftr format?

>>The description below shows the output of .ftr file.

==========================================================

# Generated with AD9361 Filter Design Wizard 16.1.3
# MATLAB 9.12.0.1884302 (R2022a), 15-Sep-2022 08:48:16
# Inputs:
# Data Sample Frequency = 20000000 Hz
TX 3 GAIN -6 INT 1
RX 3 GAIN 6 DEC 1
RTX 1280000000 160000000 80000000 40000000 20000000 20000000
RRX 1280000000 160000000 80000000 40000000 20000000 20000000
BWTX 4011543
BWRX 20172411
-77,12
-39,-138
-189,318
-185,-435
-321,398
-325,-173
-375,-250
-287,830
-172,-1450
78,1924
362,-2034
709,1577
1001,-438
1210,-1341
1219,3526
1001,-5646
504,7062
-224,-7031
-1117,4702
-2016,340
-2748,-8529
-3094,19968
-2873,-26769
-1934,15636
-232,15636
2194,-26769
5171,19968
8444,-8529
11674,340
14508,4702
16611,-7031
17732,7062
17732,-5646
16611,3526
14508,-1341
11674,-438
8444,1577
5171,-2034
2194,1924
-232,-1450
-1934,830
-2873,-250
-3094,-173
-2748,398
-2016,-435
-1117,318
-224,-138
504,12

==========================================================

Please check the sampling rate differences between FIR and waveform, meanwhile, I will check the filter file.

Thank you for your help.

I will check it.