Reduce FM Noise Floor Artifacts on NI USRP-2954

Updated Aug 26, 2026

Issue Details

I am using an NI USRP-2954 to generate RF signals and have observed FM noise or ripple-like modulation near the noise floor during radar test set evaluations. The issue is visible even when transmitting a pure sinusoidal tone and after reducing RF gain to avoid clipping. The behavior appears consistently across test conditions and impacts FM noise measurements. I would like to improve the noise performance and determine whether the observed effects originate from the radio configuration, synthesizer behavior, digital signal chain, or another portion of the transmit path.

Solution

FM noise floor artifacts on an NI USRP-2954 can be influenced by calibration state, tuning configuration, digital signal headroom, clocking, and measurement setup. Perform the following troubleshooting steps, starting with the lowest-effort actions:
  1. Re-run transmitter calibration utilities.
    • Run uhd_cal_tx_dc_offset.
    • Run uhd_cal_tx_iq_balance.
    • Retune the radio after calibration and repeat measurements.
    • Compare results before and after calibration.
  2. Configure a non-zero local oscillator (LO) offset.
    • Tune the RF LO away from the desired carrier frequency.
    • Apply a digital frequency shift to move the signal back to the desired RF frequency.
    • Keep the desired signal within the central portion of the Nyquist bandwidth to avoid filter roll-off effects.
  3. Increase the sample rate.
    • Increase the sample rate to provide additional digital tuning range.
    • Maintain the desired signal away from Nyquist-band edges.
    • Compare FM noise measurements across multiple sample-rate configurations.
  4. Evaluate Integer-N tuning.
    • Configure the LO using Integer-N mode when supported by the software environment.
    • Compare spectral measurements between Integer-N and fractional-N operation.
    • Evaluate whether close-in spurs or FM-like ripple are reduced.
  5. Verify digital signal headroom.
    • Confirm that the complex waveform amplitude remains below full scale.
    • Ensure that DAC and interpolation stages are not being overdriven.
    • Repeat measurements using reduced waveform amplitudes if necessary.
  6. Verify the reference clock source.
    • Confirm proper operation of the internal reference or any external 10 MHz reference source.
    • Repeat measurements using an alternate qualified reference when available.
  7. Analyze the FM-demodulated signal.
    • Capture the FM-demodulated waveform.
    • Perform an FFT analysis.
    • Determine whether the noise appears as discrete spurs or broadband phase-noise-related energy.
    • Use the spectral location of the noise to guide further troubleshooting.
  8. Isolate the measurement path.
    • Compare results with and without external filtering where practical.
    • Use a narrow band-pass filter as a diagnostic tool to determine whether artifacts originate in the transmitter chain or downstream measurement equipment.
  9. Compare test conditions.
    • Record and compare:
      • LO frequency
      • Sample rate
      • Calibration status
      • Transmit amplitude
      • Reference source configuration
    • Use controlled comparisons to identify which parameter most strongly influences the observed noise floor.

Related Links

Other Support Options

Ask the NI Community

Collaborate with other users in our discussion forums



Request Support from an Engineer

A valid service agreement or active software subscription may be required, and support options vary by country.