LA1224 FR1/FR2 Test Tool (v5.1): Using Option8 time-domain injection for single-carrier waveform

[SakuraAhua]

LA1224 FR1/FR2 Test Tool (v5.1): Using Option8 time-domain injection for single-carrier waveform – is a dedicated image/kernel configuration required?

Hi NXP team,

I’m using LA1224-RDB-BHS with FR1/FR2 Test Tool v5.1 and VSPA images in this tool（e.g., MEvspa_images_HS.2T2R_400M_120K_1966_1966_TDDFDD） to validate FR2 single-carrier (SC) waveform transmission in Option8 time-domain mode. I’m using SMA cables with the LA1224-FR2-BRK to loop back the TX to RX on a single LA1224-RDB-BHS device.

Steps I performed:
1. Generate a 10 ms complex IQ waveform at 491.52 Msps (int16 IQ interleaved), file size 19,660,800 bytes.
2. Load it via:
./boot_vspa.sh MEvspa_images_HS.2T2R_400M_120K_1966_1966_TDDFDD
./channels_start.sh 4
./set_dfe_state.sh 4 normal
./update_test_vector.sh my_waveform_10ms.bin
./dump_time_domain_tx.sh 4
./dump_time_domain_rx.sh 4

Dumps obtained:
- TX dump: tx_timedomain_20ms_1966080ksps_dump_ant4.bin (Fs=1.96608 Gsps)
- RX dump: rx_timedomain_20ms_983040ksps_dump_ant4.bin (Fs=983.04 Msps)

I performed MATLAB correlation-based alignment and best-window comparison.

Observations:
- Correlation peak scores are very high (e.g., ~0.998), so the injected waveform is definitely being transmitted and received.
- However, the measured EVM between my reference waveform (A) and TX dump (B) is not negligible:
- For PN/BPSK + RRC test waveform: A→B EVM ≈ 6.75% (NMSE ≈ -23.4 dB)
- For another SC waveform: A→B EVM ≈ 22.8%
- Disabling some kernels does not change EVM significantly:
./kernels_disable_tx.sh 4 qec cfr dpd
./update_fine_cfo_nco_freq.sh 4 0
(EVM/NMSE nearly unchanged)

Questions:
1. For single-carrier FR2 validation, is the provided VSPA image (with OFDM SCS-based naming) still appropriate, or is there a dedicated single-carrier optimized image?
2. In Option8 time-domain injection, are upsampling/downsampling/CFO/QEC kernels always in the processing chain? Is there a supported way to bypass them (or guarantee “transparent pass-through”) for SC waveform testing?
3. What is the expected EVM floor between (A) injected waveform and (B) TX dump in this setup? Is ~6–7% reasonable due to fixed-point/filtering, or does it indicate a misconfiguration?

I can provide exact scripts, config, and waveforms if needed.

Thanks!

[flappy]

Hello! If you replace the file generated in the time domain with Fs 491.52 MHz, will the DUC x4 work up to 1966.08 MHz on the VSPA?
I believe you need to generate a time domain signal with the same Fs frequency as the DAC, i.e., 1966.08 MHz.

[June_Lu]

LA1224 support is not public.

If you need assistance, please create a case at the link below and attach your NDA:

https://support.nxp.com/s/?language=en_US

Thanks.