Noise Figure Measurement Technique for 3mm Monolithic Integrated Circuit

0 Introduction
3 mm has many advantages in military applications due to its short wavelength, so it is widely used in precision guidance and point-to-point communication. As a key technical indicator of various military electronic equipment, the sensitivity of the receiver is the key technical indicator, and the receiver sensitivity mainly depends on the noise level of the receiver. Therefore, the noise factor of the measurement system is one of the key parameters for evaluating electronic equipment systems. The 3 mm low-noise monolithic amplifier circuit pre-researched by the military needs to measure its noise factor. It is imperative to establish a 3 mm noise factor measurement system, study its measurement method, and achieve accurate measurement. For this purpose, this paper establishes a 92-97 GHz on-chip noise factor measurement system.

1 Noise figure measurement principle
The principle block diagram of the system designed in this paper is shown in Figure 1.

(2) Reducing the influence of the phase noise of the local oscillator signal generator
The phase noise of the 3 mm signal generator is measured using an Agilent 8563E spectrum analyzer, a 3 mm harmonic mixer, and the phase noise measurement software 85671A. The maximum offset frequency that can be measured is 300 MHz. The measurement results of the phase noise of the local oscillator signal generator are shown in Figure 4.

The requirements for the local oscillator phase noise in noise figure measurement should meet any of the following statements:
a. The phase noise level at an intermediate frequency that deviates from the carrier does not exceed -130 dBm/Hz;
b. The local oscillator phase noise level does not exceed [-174 dBm/Hz+NFdut+Gdut].
The measured phase noise of the local oscillator signal generator AV1482A is -110 dBc/Hz when the deviation from the carrier is greater than 50MHz. Due to the use of a balanced mixer, it has a 20 dB suppression degree on the local oscillator noise, and the isolation from the local oscillator to the input end is 20 dB. Therefore, the noise level caused by the local oscillator phase noise at the mixer input end is:

Where: Pt (dBm/Hz) is the power of the local oscillator phase noise leaking to the mixer input; Pc (dBm) is the local oscillator carrier power; L (dBc/Hz) is the local oscillator phase noise; Im (dB) is the isolation between the mixer local oscillator input and the RF input; Sm (dB) is the mixer's suppression of the local oscillator phase noise; NFdut (dB) is the noise factor of the DUT; Gdut (dB) is the gain of the DUT.
Under the worst conditions, NFdut = 3 dB, Gdut = 0 dB, NFsys = 5 dB, Gsys = 30 dB.
The noise power generated by the DUT when the input impedance is 50Ω and the noise power generated by its own noise and the noise of the system low noise amplifier at the mixer input is:
Pn=KT0+NFdut+GdutNFsys+Gsys=-174 dBm+3 dB+0 dB+5 dB+30 dB=-136 dBm／Hz
In the formula: NFsys(dB) is the noise factor of the low noise amplifier; Gsys(dB) is the gain of the low noise amplifier; B(Hz) is the noise bandwidth; T0(K) is the standard temperature (290 K); K is the Boltzmann constant (1.38×10-23).
​​Conclusion: The noise level generated by the local oscillator phase noise of this system at the mixer input does not exceed the requirements:
-147 dBm／Hz<<-130 dBm／Hz meets the requirement of item a;
-147dBm／Hz<<-136 dBm／Hz meets the requirement of item b.
Since the system calibration is required to correct the secondary noise of the system when measuring the noise figure, the above conditions will not affect the measurement uncertainty of the noise figure.
(3) Adding a 3 mm low-noise amplifier to the system
In the 3 mm frequency band, the conversion loss of the balanced mixer is >10 dB, and the noise
figure is also at this level. If a low-noise amplifier is added to the system, it will not only reduce the contribution of the secondary noise of the system, but also make the system work very stable, and the repeatability of the measurement data is very good. At the same time, it reduces the impact of the system local oscillator phase noise on the system measurement.
(4) Calculate the dynamic range of the measurement system
① Estimation of the dynamic range of the amplifier:
Taking into account the fluctuations of the amplifier gain and noise figure, take its noise figure as 5 dB, then:

The input signal of the amplifier P-1dB compression point is -40 dBm, so the dynamic range of the amplifier is 23.6 dBm.
② Estimation of system dynamic range
Estimation of noise source output power:
First, calculate the average excess noise ratio (ENR) of the noise source:

The output noise power is:

This estimates that the dynamic range of the system is about 15dB. Therefore, amplifiers with a gain greater than 15 dB need to be tested together with an attenuator after the amplifier.

3 Analysis of measurement results
3.1 Measurement data
16 bare PHEMT circuit chips developed by our institute were measured. Figure 5 shows the measured noise figure and gain curve of one of them. The bias conditions are Vds=1.0 V, Ids=22 mA.

3.2 Analysis of measurement uncertainty
The noise figure measurement uncertainty depends not only on the accuracy of the noise figure analyzer, but also on the noise figure and gain of the device under test, as shown in Figure 6.

Considering the mismatch factor at the same time, the following calculation formula is used
:

According to the above formula, taking the 94 GHz MMIC amplifier as an example, calculate UB.
Noise figure NF1 (dB) = 3.43 dB, F1 = 2.203,
gain G1 (dB) = 13.46 (dB), G1 = 22.182,

3 mm receiver noise figure NF2 (dB) = 4.85 dB, F2 = 3.054 9,
standing wave ratio is 1.12, ρ = 0.056 6,
noise source output standing wave ratio is 1.13, ρ = 0.061 0,
F12 = F1 + (F2-1) / G1 = 3.608 9.
Calculate the following quantities:

From the technical indicators of the noise factor analyzer, we know that: δNF = 0.1 dB, δG = 0.15 dB.
According to the mismatch uncertainty formula: ±20log(1+ρsρl), the mismatch uncertainty is calculated:

According to formula (7), the noise figure measurement uncertainty is calculated to be 0.28 dB.

4 Conclusion
This paper only introduces the measurement of low noise monolithic integrated circuit bare chip noise figure in the frequency range of 92 to 97 GHz. In fact, this system can be used to measure the noise figure in the frequency range of 75 to 110 GHz. Currently, the 3 mm noise source calibration technology is being studied on this system.