1

Introduction to Noise Parameters

2Noise Parameters

Noise Figure:

• Noise Performance at one Impedance

• Typical – 50Ω Noise Figure

Noise Parameters:

• Noise Performance at any Source Impedance

3

( )

−+

−+=

22

2

min

114

sopt

opts

nrFF

n

optoptopt

r

F

→ |,|

min

Four Scalar Values:

nr

Noise Parameters

Imag s

Real s

opt

NF

minF

4Noise & Gain Circles

5Noise Parameters Measurement

• Set 4 Values of s

• For Each s, Measure F

• Solve 4 Simultaneous Equations for the 4 Values

𝐹 = 𝐹𝑚𝑖𝑛 + 4𝑟𝑛Γ𝑠 − Γ𝑜𝑝𝑡

2

1 + Γ𝑜𝑝𝑡21 − Γ𝑠

2

.

General Method

6Noise Parameters Measurement

Practical Method

Use Over-Determined Data:

• Measurement is Sensitive to Small Errors

• Measure at more than 4 s Values

• Use Least-Mean-Squares to Reduce Data

Use Noise Power Equation:

• Rigorous Solution

• Account for hot and cold of Noise Source

• Allows Hot/Cold or Cold Only Approaches

𝑃 = 𝑘𝐵 𝑡𝑛𝑠 + 𝑡0(𝐹1 − 1) 𝐺𝐴1 + 𝑡0(𝐹2 − 1) 𝐺𝑇2

Data Collection Methods 7

Traditional Method

Ultra Fast Method*

Very Time Consuming• Can Have Drift Issues • Data Scatter

• One Frequency at a Time• Allows Ideal Impedance Pattern• Based on 1969 Paper• Used by everyone for almost 40 years

• Characterize One Set Of Tuner States

• Sweep Frequency at Each State

• Take Advantage of Fast Sweep of Modern Instruments

*Invented by Maury Microwave

US Patent 8,892,380

8Noise Parameters System Block Diagram

Typical Maury on-wafer noise parameters system

PNAX

NS

MT7553B03

DUT VNA

Bias

Power Supply

DUT

Switch Driver

GPIB

28V

NSMNoise

Source

VNA

Tuner

Bias

MT984AL01

9Noise Parameters System Photo

10

On-Wafer, 8-50 GHz, Fmin= 0.35 – 0.9 dB

Measured Data, No Smoothing Applied

Noise Parameters Measurement Results

Data Outliers and Noise Data Processing 11

Outliers

• Transistor Data should be continuous• Noise Measurements are Over-Determined• Noise Data results should always come from one set of measured data.

Processing: • Data Outlier causes change from good data• Calculate Noise Parameters from Subsets to Remove Outliers• Errors add or cancel at different frequencies• Data Scatter vs Frequency is an Indicator of the measurement quality• Removing outliers tends to reduce scatter

12

Noise Parameter Validation & Confidence in Measurements

Measurement Confidence 13

System Accuracy is established using known devices:• Passive Verification Devices

• Active Device as Golden Standard

• Cascade Verification

System Repeatability must be a requirement• Measurement Repeatability

• Multiple calibrations

• Overlap bands with different tuners or receivers

• Comparison with other Labs

• Benchmark with legacy ATN systems

Passive Device Measurement 14

A passive device tests accuracy of:• S-Parameter Measurements

• Noise Calibration

• Noise Measurement

Passive DUT

Passive Device Measurement 15

Two Options:

• Matched Passive Device → Some Errors are Masked

• Mismatched Passive Device → Better Test

Known Limitation: System is tested at a different Gain Range than Active DUT

Passive Device Measurement 16

0

1

2

3

4

5

6

7

8

9

10

7 12 17 22 27 32 37 42 47

Fmin

(d

B)

Freq (GHz)

6dB Mismatch comparison: NFmin

ATS_with NRM Passive verification

Active Device as Golden Standard 17

• Challenges: • Selecting a stable bias condition for repeatable measurements

• Device wears out over multiple touch-downs

• Device performance can change over time

• Benefits:• System verification is established using similar DUT conditions

• Recommendations: • Multiple known devices based on reference measurements

• Limit the usage of these golden standards to system verification only

18

DUT

1. Measure Noise

2. Measure S-Para

3. Measure Noise

4. Calculate Noise

DUT

Passive

Passive

DUTPassive +Compare

Cascade Verification

Advantage: System is tested at same Gain Range as Active DUT

System Repeatability 19

• Measurement Repeatability

• S-parameters

• Multiple calibrations

• Overlapping bands with different tuners (same or different setups)

• Comparison between measurements at different sites

• Benchmark with legacy ATN system

20Noise Measurement Repeatability

• On wafer Calibration measured over multiple days • Excellent repeatability

• Complete tear down and rebuilding setup• Multiple coaxial and on-wafer calibrations

0

0.3

0.6

0.9

1.2

1.5

1.8

2.1

2.4

2.7

3

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

Nfm

in (

dB

)

Frequency (GHz)

6

7

8

9

10

11

12

13

14

15

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

Nfm

in (

dB

)

Freq (GHz)

Noise Calibration DUTRAW DATA

Measurements at Different Sites 21

0

0.5

1

1.5

2

2.5

3

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

NFm

in (

dB

)

Freq (GHz)

Modelithics (FL)Maury (CA)

0

0.5

1

1.5

2

2.5

3

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

NFm

in (

dB

)

Freq (GHz)

Raw Data Processed Data

Benchmark With Legacy ATN System 22

0.5

1

1.5

2

2.5

3

3.5

4

0 5 10 15 20 25

NFm

in(d

B)

Frequency (GHz)

ATS

ATN/NP5

23

Critical Variables that Affect

Noise Parameters Measurements

24With NRM

Without NRM

PNAX

NS

MT7553B03

DUT VNA

Bias

Power Supply

DUT

Switch Driver

GPIB

28V

NSMNoise

Source

VNA

Tuner

Bias

MT984AL01

Noise Receiver Module

0

0.3

0.6

0.9

1.2

1.5

1.8

2.1

2.4

2.7

3

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50N

fmin

(d

B)

Frequency (GHz)

Without NRM

With NRM

NRM To increase the sensitivity of DUT noise measurement a LNA stage is required.

Maury Microwave Noise Receiver 25

Model Freq range Noise FigureMT7553B03 0.1-50GHz 10dB Typical

16dB Max

• Hold your solutions providers accountable!• Published vs actual data comparison

23456789

10111213141516

8 18 28 38 48

Nfm

in (

dB

)

Frequency (GHz)

On wafer calibration

26

• Maury SW has the ability to calibrate at multiple attenuation levels • Dynamically chooses the right attenuation during DUT

measurements

Measurement Problems• Too much attenuation• Too low attenuation• Receiver overload

High Attenuation

Low Attenuation

Noise Receiver Attenuation

0

0.3

0.6

0.9

1.2

1.5

1.8

2.1

2.4

2.7

3

2 4 6 8 10 12 14 16 18

Nfm

in (

dB

)Frequency (GHz)

High Attenuation

Low Attenuation

27

Shields cell phone and wireless signals that can interfere with noise measurements.

0

1

2

3

4

5

6

7

8

0.8 3.2 5.6 8

NFm

in(d

B)

Freq (GHz)

Noise calibration (0.8 to 8 GHz)Screenroom No screenroom

interference point at 1.9 GHz

Screened Room (Faraday Cage)

28

• Need a good Γ𝑆 spread at every frequency• Need to have the Γ𝑆 spread include high gamma values.

Bad Γ𝑆 spread Good Γ𝑆 spread

Typical Γ𝑂𝑃𝑇 DUT

Low Γ𝑆 spread

Gamma Source Selection

Oscillation 29

DUT Γ𝑂𝑃𝑇

• At some high-gamma states, oscillation could occur

• MMC Software detects oscillation states and

removes them

• Points in the stability circle may not be unstable,

they are only Potential Unstable, stability also

depends on the output load

• But most high-gamma states are useful –

Γ𝑂𝑃𝑇 can be within stability circles