A third-order intercept point (IP3) measurement technique using a network analyzer is proposed in this work. Using network analyzers to measure IP3 is becoming a trend for modern measurement instruments.1-3 Generally speaking, extra sources and controlling programs are needed to fulfill this measurement task. The novelty of the proposed measurement method rests upon its ability to perform IP3 measurements, employing a two-tone excitation with a wide frequency spacing in the range of a few hundred megahertz, which, for example, is used in power amplifier memory effects related research and development.4,5 Plus, the industry bandwidth drive potentially demands an IP3 measurement with wide two-tone separations, for future wideband devices. The two-tone amplitude imbalance related to this method is calibrated out based upon arithmetic average. The detailed multiple external sources and network analyzer receiver programming techniques are presented in this article. The agreement between the measured results of an amplifier using this method and the conventional method using a spectrum analyzer confirms the functionality and effectiveness of the proposed method. With extra power level calibrations at different frequency bands, this programming technique also can be extended to mixer IP3 points measurement.

Recent research on power amplifiers and their memory effects requires the measurement of intermodulation distortion (IMD) products as a function of the modulation bandwidth. Consequently, a common measurement adopted in this field is to measure the amplifier outputs, including fundamentals and IMDs, with different tone spacings for a two-tone input excitation.4,5 Using a conventional measurement approach, with a spectrum analyzer plus two sources, can provide the required measurement,6 but it is inefficient and tedious to get the third-order intercept point (IP3) information over a band of frequencies. This is because the lack of certain automation of the two external sources, which should sweep frequency synchronously. Using network analyzers can provide the control of external sources without extra controlling computers. Previously, a third-order IMD measurement method using a network analyzer has been available. This uses a network analyzer as a controller to control two external sources, providing a certain amount of automation.2 The reported method makes use of a network analyzer to control two external sources as well as a tuned receiver of the network analyzer to measure IMD products, therefore offering a scheme for IP3 measurement over a band of frequency. This method performs best, when the two tones are spaced not very far away from each other, that is in the range of a few megahertz.2 This is because amplitude imbalance will be inevitable, if the two tones are widely spaced. Therefore, inaccuracy could be generated if this method is applied to the measurement of IMD with a wide two-tone spacing. In this work, a novel measurement method for IP3 is presented, which provides the ability to measure the third-order IMD with two-tone frequency widely separated in the range of a few hundred megahertz. The introduced amplitude imbalance is calibrated out based on an arithmetic average. Plus, compared with the conventional method,2 the proposed method can obtain IP3 information with only one frequency sweep instead of two sweeps, which could potentially reduce measurement time and be of interest to manufacturers with a large volume of measurement tasks.

Figure 1 An IP3 measurement setup for an amplifier, using a network analyzer.

Network Analyzer Calibration and Multiple Source Console

The measurement instruments hardware setup is similar to the one published in Anritsu Application Note,2 where a power combiner is used to combine the two input tones, while minimizing the interference between them, as shown in Figure 1.

Two external sources are controlled by the network analyzer through a general-purpose-interface-bus (GPIB). The detailed network analyzer calibration and multiple source console programming techniques are explained step by step as follows.

The first step is to establish a 0 dBm power calibration line. This is essential because while the two sources are swept simultaneously, the network analyzer receiver response shows a different absolute power value at the same frequency. Therefore, the calibration is aimed to establish an interpretable power reading based upon the input of a known source power level. To do this, the two sources are set to sweep simultaneously with a frequency span of 2ΔF and a power level of 0 dBm, while the network analyzer receiver is programmed to receive the input signal consecutively, where ΔF is the desired two-tone excitation frequency spacing. By doing so, the network analyzer receiver real response to a known power level at different frequencies, including the two upper and lower IMD bands and two fundamental output bands, is accurately recorded. A 0 dBm power calibration line can be established by normalizing the recorded data to themselves.

The second step is to program the network analyzer into four different, consecutive frequency bands. This way, the network analyzer is programmed as a receiver, which is operated on four consecutive bands, F1 to F4. Each band is used to measure a different frequency component at the amplifier output, such as IMD1, OUT 1 (tone 1 output), OUT 2 (tone 2 output), IMD2. The complete multiple source programming details are listed in Table 1, where Bandi is a frequency band of interest, with a width of two interested products; Fi is a frequency band of interest, with a width of one interested product; ΔF is the frequency span separated by two tones.

In order to measure two fundamental output signals and two IMD products through one sweep, two external sources are programmed to sweep across the desired bandwidth synchronously, while the network analyzer receiver is programmed to receive four equal bandwidth output products (equal Fi), namely IMD1 (with frequency 2f1-f2), OUT1 (f1), OUT2 (f2), and IMD2 (2f2-f1) consecutively. This is significantly different from the measurement method published by Anritsu.2 This way, the amplifier gain, IMD level as well as the signal-to-IMD ratio are all measured and can be displayed at the same time.

Figure 2 Four consecutive bands used to measure four products of an amplifier output.

The third step is to extract the related data segment, which represents the amplifier's different outputs and performs post processing. Because the two tones are widely separated in frequency, their amplitudes will inevitably have differences as shown in Figure 2. As suggested by P. Vizmuller,7 the amplitude imbalance can be calibrated out, based on arithmetic average. Therefore, the two different intermodulation distortion ratios are averaged and the amplifier input IP3 is calculated based on this averaged number.

Measurement Results

Anritsu network analyzer 37397C is used in this work. In fact, the above mentioned method applies both to Anritsu 37×××C and 37×××D series analyzers. The two external sources are Anritsu 68177C and 68037C. The amplifier used in this work is a packaged microwave integrated circuit (IC) HMC482 from Hittite Microwave Corp., which is biased at +7.0 V, 83 mA. The selection of input excitation tones power level should be high enough to be able to excite the third-order IMD products, and low enough not to reach the amplifier input P1dB point. This is because the arithmetic average used assumes that the outputs of the fundamental tones and IMD products are increased linearly with the inputs (when evaluated in dB). Because the datasheet suggests that the amplifier has an input 1 dB saturation point at approximately 2.5 dBm, the two input fundamental excitations are set to -5 dBm and 100 MHz spaced apart. The insertion loss of the power combining network is calibrated out during the 0 dBm absolute power calibration. It can also be measured in advance with the network analyzer following a standard S-parameter measurement and deducted in data post processing.

Figure 3 Measured results of an amplifier using a network analyzer and a spectrum analyzer.

The individual power levels of the output signals and IMD products are extracted to perform the amplitude imbalance calibration.7 The measured amplifier gain, third-order IMD and the extrapolated input IP3 (IIP3) are plotted in Figure 3.

In order to verify the functionality of the proposed measurement technique, a spectrum analyzer is used to make the measurement with the same amplifier and power combining network setup. It is observed from Figure 3 that the results obtained from these two measurement methods generally agree, which confirms the functionality of the proposed method in this work.

Meanwhile, the network analyzer measurement setup can also be used to perform a swept power measurement at a fixed frequency point. For the same amplifier, the measured results are shown in Figure 4 with the input power swept from -10 dBm to 0 dBm. The extracted IP3 number is approximately 15 dBm, which generally agrees with the results of the sweep frequency measurement and verifies the robustness of the proposed method.

Figure 4 Measured results for swept power at a fixed frequency.

It is worth pointing out that if extra 0 dBm absolute power line calibrations are done both at a radio-frequency band and an intermediate-frequency band, the same multiple source programming technique can also be used in a three-port mixer input IP3 measurement.

Conclusion

This article presents a network analyzer programming technique for IP3 measurement, using Anritsu network analyzer 37×××C, which is featured with the ability to perform a two-tone measurement with a wide frequency spacing in the range of a few hundred megahertz. The detailed network analyzer calibration and multiple source console programming procedures are given. To verify the functionality of the proposed method, the classic spectrum analyzer approach is used as a comparison. The agreement between these two measurement methods confirms the effectiveness of the proposed method.

Acknowledgments

The authors would like to thank Mr. J. Gauthier and Professor Cevdet Akyel, all with the École Polytechnique de Montreal, Montreal, QC, Canada, for helpful discussions. Financial support from the Natural Science and Engineering Research Council of Canada (NSERC) and Le Fonds Québécois de la Recherche sur la Nature et les Technologies (FQRNT) are gratefully acknowledged.

References

  1. "Agilent 2-port and 4-port PNA-X Network Analyzer N5241A Data Sheet." Available at http://cp.literature.agilent.com/litweb/pdf/N5242-90007.pdf.
  2. "Intermodulation Distortion Measurements Using the 37300 Series Vector Network Analyzer," Anritsu, Morgan Hill, CA, Application Note, May 2008. Available at: http://www.us.anritsu.com/downloads/files/ 11410-00257a.pdf.
  3. M. Demmler, B. Hughes and A. Cognata, "A 0.5-50 GHz On-wafer, Intermodulation, Load-pull and Power Measurement System," 1995 IEEE MTT-S International Microwave Symposium Digest, pp. 1041-1044.
  4. J.S. Kenney, W. Woo, L. Ding, R. Raich, H. Ku and G.T. Zhou, "The Impact of Memory Effects on Predistortion Linearization of RF Power Amplifiers," Proceedings of the 8th International Symposium on Microwave and Optical Technology, Montreal, QC, Canada, 2001, pp. 189-193.
  5. H. Ku, M.D. McKinley and J.S. Kenney, "Quantifying Memory Effects in RF Power Amplifiers," IEEE Transactions on Microwave Theory and Techniques, Vol. 50, No. 12, December 2002, pp. 2843-2849.
  6. A.E. Bailey, "Microwave Measurements," Peter Peregrinus Ltd., London, 1985, pp. 300-311.
  7. P. Vizmuller, "RF Design Guide: Systems, Circuits and Equations," Artech House, Norwood, MA, 1995, pp. 186-190.