Hello,
I am not sure whether this is expected behavior or a software issue, so I would appreciate some clarification.
Configuration
- USRP X440
- UHD 4.8.x
- FPGA image: X4_400
- RFNoC Replay
- Python
- TX0–TX3
- Center frequency: 1 GHz
- Sample rate: 368.64 MS/s
- 10 MHz complex tone
- Same waveform transmitted on all four TX channels
I removed timed tuning after reading the X440 documentation because timed tuning is not supported on X440.
Instead, I use a timed RFNoC Replay stream start (stream_now = False with a future time_spec) so that all Replay ports begin transmitting simultaneously.
Measurements
Using a Keysight oscilloscope with equal-length SMA cables, I consistently measure approximately:
- CH1–CH2 ≈ 4.4 ns
- CH1–CH3 ≈ 260–280 ps
- CH1–CH4 ≈ 975–980 ps
These delays appear to be repeatable across multiple application restarts.
Questions
- Is this expected behavior on a single X440?
- Does RFNoC Replay guarantee deterministic sample alignment only, while fixed RF path delays remain?
- Is there an official calibration procedure for TX phase alignment between channels?
If needed, I can provide my Python application, oscilloscope screenshots, and uhd_usrp_probe output.
Thank you.
Hello,
I am not sure whether this is expected behavior or a software issue, so I would appreciate some clarification.
Configuration
I removed timed tuning after reading the X440 documentation because timed tuning is not supported on X440.
Instead, I use a timed RFNoC Replay stream start (
stream_now = Falsewith a futuretime_spec) so that all Replay ports begin transmitting simultaneously.Measurements
Using a Keysight oscilloscope with equal-length SMA cables, I consistently measure approximately:
These delays appear to be repeatable across multiple application restarts.
Questions
If needed, I can provide my Python application, oscilloscope screenshots, and
uhd_usrp_probeoutput.Thank you.