August 13, 2018
Differential devices and systems have been part of RF and microwave systems for decades. They are becoming more prominent today due to the rollout of 5G and advanced wireless applications. Because these current designs incorporate differential circuits and devices with improved sensitivity and low noise, the margin for error associated with verifying differential components has narrowed considerably.
For that reason, conventional methods of measuring noise figure do not provide the degree of accuracy so engineers have confidence in their designs. Engineers must efficiently characterize noise figure performance of 2-, 3-, and 4-port devices. Fortunately, there is a new process that utilizes dedicated software and leverages the advantages of the Anritsu Non-linear Transmission Line (NLTL) sampling technology found in the VectorStar™ vector network analyzer (VNA) to achieve that goal.
Importance of Noise Figure Measurements
Noise figure measurement accuracy is important for several reasons. With the increasing demand for greater performance being placed on so many systems, precisely knowing the noise performance is critical. At the R&D stage, higher accuracy leads to improved system/component models, which enables better correlation between design simulations and measurements. This can lead to better system optimization for greater product differentiation or more predictable developments for faster time-to-market. In manufacturing, better accuracy can reduce measurement guard bands, which can result in tighter specifications for greater product differentiation. It can also lead to higher yields, improved throughput, and lower overall test costs.
A VNA such as the VectorStar (figure 1) is the instrument of choice when conducting noise figure measurements. In addition to its NLTL design, the VectorStar incorporates the cold-source noise figure measurement method. Cold-source was developed to eliminate the need for a multi-state noise source, which results in a simpler, better controlled noise source (nominally a termination at room temperature). The cold-source method minimizes the mismatch errors commonly found in Y-factor noise source method.
For differential noise figure measurements, engineers can utilize the cold-source method in a very similar fashion as they do for a single-ended device. The only difference is in identifying the correlated noise power that can be present in differential measurements that may be contributing to uncertainties.
Conducting a Differential Noise Figure Measurements
When engineers are performing both single-ended and differential noise figure measurements using the cold source method, there are three initial steps:
- Measure the device under test (DUT) and obtain a sNp file
- Assemble a composite receiver for the DUT
- Configure the noise figure measurement
When conducting differential measurements, generally there are three choices for analysis – uncorrelated, combiner, and correlated. Below are brief outlines of each.
Uncorrelated – A traditional uncorrelated method is used when the DUT is believed to exhibit perfect noise isolation within the two paths. To achieve this, the engineer treats the differential device as two single-ended noise generators and performs single-ended noise figure measurements on both channels. Once the noise power of each has been acquired, the process would use the measured data with the DUT S-parameter and calibration data to compute the differential noise figure.
If there are little to no isolation issues within the DUT, the dual single-ended measurement method is probably acceptable to characterize the devices. If there are any cross-correlation and isolation issues, however, some uncertainties will be introduced into the measurement. Given the tight specifications associated with 5G and high-speed radar, these uncertainties are unacceptable for designs for emerging technologies.
Combiner – A single-ended measurement conducted using a balun allows engineers to acquire a measurement from the combined output of the balun. Of course, the balun will need to be de-embedded using the sNp file of the balun to account for the imbalance created by the combiner.
One potential problem that may arise with this approach is when the common mode crossover in the single-ended output is not accounted for. In this scenario, some uncertainties will be introduced due to the correlated aspects of the combiner, as well as the DUT.
Correlated – By using the correlated method, engineers can identify and characterize the correlated noise powers that are in the measurement system and remove them. A unique VectorStar correlated technique available through the differential noise figure software identifies and extracts the correlated noise present at the differential outputs to avoid both sources of error.
Here is an example of the correlated approach:
Integrate a pre-amp into the composite receiver that boosts the noise floor of the DUT above that of the VNA. Next, characterize the composite receiver with the VNA source. With the VectorStar VNA, there are two NLTL receivers that are coherent, which allows engineers to analyze coherent noise and identify the correlation aspects.
Once the correlation aspects are identified, terminate each receiver input to characterize the composite noise floor of each receiver on the respective channel. The final step is to attach the DUT and make the measurement. It is important to connect a load to the input of the DUT before making the measurement. The displayed noise power is the actual measurement.
Figure 2 shows the variations between the three methods. As the chart depicts, typically uncorrelated methods undercount differential noise power, while basic balun measurements have more uncertainty since they are not correcting for balun characteristics. The correlated method provides the most accurate measurement results.
Without the proper correlation, the measurement difference can range from a few tenths of a dB to a few dB. With high-speed, high-frequency designs, that variation can be monumental in terms of whether the device will perform within specification. Properly identifying and de-embedding the real and imaginary imbalance improves noise power measurements.
To learn more about conducting differential noise figure measurements using a VNA, visit this dedicated Noise Figure Measurement page.