Noise Figure Measurement Using A Spectrum Analyzer And Signal Generator

Procedure, General: Tune Signal generator and Spectrum analyzer to frequency where measurement is to be performed; adjust Sig Gen for a signal approximately 3 dB above the noise as displayed on the S/A and input this Sig Gen reading into equation below. Spectrum Analyzer Set Up (Rohde & Schwarz FSEB30 used here): RBW = 1 MHz VBW = 1 KHz or 500 Hz SPAN = 10 MHz Trace(1-4)
Clear/Write Trace(1-4)
Detector
RMS Trace(1-4)
Sweep Count
100 Trace(1-4)
Average Marker
Search
Peak Marker
Delta
Dial +- 4.5 MHz

Gives us plenty of noise power to work with Smooth out some of the detected noise Noise cursor will be placed at +- 4.5 MHz Write fast initially; see the peak easier RMS seems to be slightly more pessimistic Set count for random variation averaging Begin to average the readings Look for the signal peak Dial second cursor away from peak in center

Procedure, Detailed: With two markers set, one at the signal peak and the other riding the output noise at approximately +- 4.5 MHz or so and adjust the Sig Gen for a 3 dB ratio between the two cursors. This yields a 3 db S/N ratio and Sig Gen power = noise power. Note: One marker indicates the DUT’s output noise power level, NO = FGkTOB and the other marker indicates So + No for double the power (+ 3dB) indication or 2*NO of the noise-only marker. (Equation derivation is from ref. 1 slide #36: G*Si = NF*G*k*TO*BW and NF = Si / k*TO*BW)

Procedure, Calculation: Using this equation input the Sig Gen reading to calculate DUT NF: NF (in dB) = Si

+ 114 - 10*Log (BW) (where Si is in dBm and BW = S/A RBW in MHz)

Example calculation (538 product): BW = 1.0 MHz Si = -110.12 dBm

(3 dB bandwidth of the selected spectrum analyzer’s RBW) (Sig Gen power corrected for any line loss. This is important!)

Calculated NF (in dB) = -110.12 + 114 - 0 = 3.88 dB

Detailed equipment set up, High-Gain (~40 dB) DUT. This set up is suitable for measuring the higher gain DUTs like the 538 product that exhibit on the order of 40 dB of gain.

Test Equipment: • Sig Gen - Agilent E4437B • Spec Ana - R&S FSEB 30 • Attenuator - MiniCircuits VAT-6 The S/A RF Attenuation level may be left at 10 dB in this set up.

Detailed equipment set up, Low-Gain (~20 dB) DUT. This setup is suitable for measuring the lower gain DUTs like the 438 product that exhibit on the order of 20 dB of gain. The addition of an amplifier after the DUT assures us that the noise of the DUT will sufficiently override the inherent noise floor of the Spectrum Analyzer.

Test Equipment: • As above, plus • Test amplifier - MiniCircuits model ZHL-42, 36 dB 700 – 4200 MHz, NF < 10 dB The S/A RF Attenuation may be left at 10 dB in this set up.

Measurement of 438 Product Gain: 19 – 23 dB NF: < 6 dB Freq: 2400 – 2483 MHz Test measurement set up for measuring a low gain DUT as described earlier. The input loss to the DUT has been compensated for via an offset input into the Sig Gen; a power meter was used to obtain an absolute power value out of the Sig Gen then an offset value was input to the Agilent ESG Sig Gen 4437. The value was -7.86 dB includes 6 dB pad and cable loss and also offsets any Sig Gen output power level error at this freq. Also notice the Test Amp with the black heat sink fins fed by the DC power supply in the lower right corner.

The power from the generator is adjusted until a 3 dB ratio is had between the noise (marker on the right) and the Signal plus Noise (marker in the left/center). Notice the RBW (1 MHz), VBW (1 KHz), RF Att (10 dB) and delta cursor reading (-3.00 dB).

The amplitude value is then read from the signal generator and used in the equation described earlier resulting in this calculation: NF (in dB) = -109.20 + 114 - 0 = 4.72 dB

Full size view of S/A screen Screen shows as 3 dB S/N ratio (The bump in center is radio’d against the flat noise level on the right.)

Full size view of Sig Gen screen This screen shows the absolute power value level (of the CW signal) that is required to achieve the same power level as the noise in a 1 MHz bandwidth. This is the equivalent ‘strength’ of the noise present in our 1 MHz wide sample (our IF bandwidth, RBW, is set for 1 MHz).

References: 1. Mohr on Receiver Noise Characterization, Insights & Surprises Richard J. Mohr, PE www.ieee.li/pdf/viewgraphs_mohr_noise.pdf