1. Hot-Cold Source Measurement	-	The hot-cold source measurement was performed between -77C and 100C using an Airborne Instrument Laboratory Type 70 Hot-Cold Body Standard Noise Generator with liquid nitrogen. One after the other, the output from each port of the noise generator was input to the LNA under testing. The output from that LNA was fed through a Lark XMC-690-1020-10AB (ANITA-2/ANITA-3 LNA) bandpass filter and carried to a HP437 Power Meter by a 12″ Rosenberger (which has low loss) cable. The calibration of the power meter was checked against the power setting of the Rohde & Schwarz SMIQ-03B Vector Signal Generator using sine waves of various power and of various frequency; the agreement was within 0.1 dBm. The Y-factor method was applied to the power measurement measurement to extract the gain and noise temperature of the LNA under the assumption that the adapter between the noise generator and the LNA had an insertion loss of 0.02 dB insertion loss and attenuation of 0.02 dB. These estimates for the adapters were based on by-eye averages of the S11 and S21 measurements made with the PNA-L 5230C network analyser.
2. 8970A Noise Meter Measurement	-	The 346A (May 2012) Noise Source was attached to the input of the LNA via a Fairview Microwave SM4252 N-to-SMA adapter and the output of that LNA was carried back to the 8970A Noise Meter purchased in August, 2012 using a 12″ Rosenberger (which has low loss) cable. The meter then scanned from 10 MHz to 1600 MHz in steps of 10 MHz. If the gain of the LNA exceeded 40 dB, a 5 dB pad was placed on the output of the LNA in order to limit the input power within the specification for the noise meter.
3. 8970B Noise Meter Measurement	-	The 346A (May 2012) Noise Source was attached to the input of the LNA via a Fairview Microwave SM4252 N-to-SMA adapter and the output of that LNA was carried back to the 8970B Noise Meter purchased in August, 2013 using a 12″ Rosenberger (which has low loss) cable. The meter then scanned from 10 MHz to 1600 MHz in steps of 10 MHz. If the gain of the LNA exceeded 40 dB, a 5 dB pad was placed on the output of the LNA to limit the input power within the specification for the noise meter.

In all three measurements, the LNAs were powered by a lab-supply. To minimize RF pickup on the power wires, a snap-on ferrite was placed on the power and ground wires and, as possible, the two wires were twisted together.

Although the relative agreement between the band-averaged, noise meter and the hot-cold source results for the four LNAs is good, there is an absolute disagreement between the two measurements of 15 K with the noise meter reporting values that are higher. Since the results from the hot-cold source method are in better agreement with the LNA manufacturer specifications, we presently believe that the hot-cold source provides a better absolute measurement. The magnitude and frequency dependence of the measurements from the two noise meters are in very good agreement, although, at some frequencies, small peaks are observed in one meter that are not seen by the other meter. None of these peaks are observed on all of the LNAs. In contrast to these peaks, a common peak is often observed at ∼850 MHz and is believed to be pick-up from the cell phone tower near the lab in Watanabe.