• Resolved

LMK03318: PSNR Figure For LMK03318

Part Number: LMK03318


The LMK03318 manual states figures for rejection of noise on the power supply in terms of Power Supply Noise Rejection (PSNR), with units of dBc. I'm not familiar with this figure and haven't had much success finding an equation. From context, however, it appears to be along the lines of PSNR = 10log(Pspur/Pcenter) dBc, where Pspur is the power of a particular spur caused by power supply injected ripple? Please correct me if I am mistaken.

Thank you

  • Power supply noise rejection, or PSNR, is defined as the single-sideband phase spur level (in dBc) modulated onto the clock output when a single-tone sinusoidal noise signal (ripple) is injected onto the power supply. Assuming no amplitude modulation effects and small index modulation, the peak-to-peak deterministic jitter (Dj) resulting from the supply noise-induced phase spur can be calculated as follows:

    Dj (ps p-p) = [ (2 x 10(dBc / 20)) / (π x Fout) ] x 1e12
    - dBc is the measured single-sideband phase spur level (in dBc)
    - Fout is the output clock frequency (in Hz)

  • In reply to Alan O:

    Thank for the reply, Alan. If I understand correctly then, under the given assumptions, the single-sideband phase spur level at output is compared to the power of the single tone sinusoidal noise signal injected at the power supply to obtain the figure in dBc?
  • In reply to MOSEL:

    The sideband spur level measured on output clock's power spectrum is relative to the power level of the output clock (carrier signal).

    Sideband spur power level (relative to carrier, or dBc) = sideband spur power level (dBm) - carrier power level (dBm)

    If the measured carrier power level is +5 dBm and measured sideband spur level is -80 dBm with 50-mVpp noise on VDD, then the sideband spur level relative to the carrier is -85 dBc.

    The sideband spur level should change dB-for-dB with the change in power level of the single-tone sinusoidal noise signal.

    Continuing from the example above, if the noise level on VDD is reduced from 50 mVpp to 25 mVpp (20*log(25/50) = 6 dB lower), then we can expect the sideband spur level to reduce by 6 dB (from -85 dBc to -91 dBc).

  • In reply to Alan O:

    Very helpful. Thank you for the detailed explanation.