Mixers are one of the fundamental components of every superheterodyne receiver (Rx), but evaluating them has never been easy, even with powerful vector-network analyzers (VNAs). The techniques employed to characterize the phase and group-delay performance of mixers are especially cumbersome, lengthy, and prone to error. To solve this problem, Agilent Technologies (Santa Rosa, CA) has created new calibration techniques for its PNA Series of VNAs that reduces or eliminates traditional problems inherent in characterization of mixers and converters. It is available for the E8362B, E8363B, and E8364B VNAs, with coverage from 10 MHz to 20, 40, and 50 GHz, respectively, as well as for the new E8361A network analyzer with coverage from 10 MHz to 67 GHz.

The 67-GHz E8361A (Fig. 1) should find a following in manufacturers of passive components and subsystems designed for satellite communications, point-to-point digital radio, broadband wireless access, and OC-768 (40 Gb/s) optical-communications systems. The instrument has all of the features and capabilities of previous PNA Series analyzers, including trace noise of less than 0.03 dB at a 1-kHz bandwidth, dynamic range greater than 90 dB at 67 GHz, and measurement speed of less than 26 µs per point. Similar to all of the PNA Series instruments, the E8361A is based on the Windows 2000 operating system, which provides the operator with a familiar operating environment, provides a multitude of connectivity choices, and allows programs to be run inside or outside the analyzer.

The frequency-offset measurement capability is implemented as a hardware and firmware solution in the analyzers. The hardware provides the ability to make basic offset-frequency measurements, including mixer-conversion loss, intermodulation distortion (IMD), and harmonic and spurious responses. The firmware automates the mixer-measurement process, making it possible for users without extensive knowledge of mixer measurement to set up, calibrate, and characterize their devices accurately and quickly. The firmware's advanced calibration choices include vector correction of conversion loss, phase, and group delay, and match-corrected absolute-power measurements, both of which increase the overall accuracy of the process compared to other methods in use today.

Any mixer-based superheterodyne receiving system requires that the mixers within it have well-controlled amplitude phase, and group-delay responses. Characterizing the amplitude response (conversion gain or loss) is the easiest measurement. Conversion phase and group delay, however, continue to be difficult to measure with high accuracy and repeatability, and the test set-up employed in the process usually requires multiple external components, with many connections and reconnections. This process creates mismatch and connector repeatability errors, and increases the chance that operator error will occur, creating a high level of uncertainty in the measurement results.

Agilent's new vector mixer calibration accommodates conversion loss as well as phase and group delay, resulting in a far more accurate and comparatively simple technique that requires fewer external components and connections. This is best understood by comparing it with two other methods that are commonly employed to characterize mixer phase or group-delay responses.

The first method requires the designer to make three measurements on three pairs of mixers. The amplitude and phase responses of each mixer are calculated by solving the three linear equations created by the three measurements. The technique uses upconversion and downconversion and employs an intermediate-frequency (IF) filter between the mixer pairs to keep the unwanted mixing product from being reconverted. The method also assumes that at least one of the mixers is reciprocal (it has the same conversion loss and group delay in upconversion as downconversion). Its most obvious drawback is the fact that three sets of measurements must be made and the mixer pairs must be reconnected with the filter. Errors can creep into the process due to connector repeatability and the mismatch effects between the filter and mixer pairs, as well as between the mixers and test equipment.

Page Title

The other popular method used for characterizing mixer group delay involves measuring mixer return loss with an air line that is terminated with a short, and taking the time-domain transform of the response. The time delay of the response with the short is subtracted from the delay due to the length of the airline by itself, which yields the two-way delay of the mixer. This technique has proven useful only for broadband mixers, and delay resolution is limited by the time-domain resolution. In addition, the delay response is a combination of the response from the sum and difference products, and the measurement requires a reciprocal mixer. On the positive side, the technique does not require additional mixers or locking the local oscillator (LO) to any of the test signals.

Agilent's technique for the PNA Series analyzers is a vector-corrected mixer calibration approach that employs reflection measurements to fully characterize a reciprocal calibration mixer/filter pair, without any additional mixers. This calibration mixer/filter pair is then used in conjunction with short, open, load, and through standards to calibrate a test system that can then be used to measure the conversion loss, conversion phase, and absolute group delay of any mixer or converter under test. The method can be used to measure both reciprocal or nonreciprocal mixers and converters, and the calibration process and control of external test equipment is highly automated.

Previously, in addition to the tedious nature of the mixer-characterization process, the instrument interface from which the designer must work has typically added to the confusion of an already-difficult task. To remedy this, Agilent's new frequency-conversion firmware presents a clear picture without requiring the user to enter obscure and confusing values. All values are set up on a single screen. By entering values into a single dialog box (shown in Fig. 2a for single-conversion devices and Fig. 2 for dual-conversion devices), all the values are presented in a single place. The firmware ensures the values are within acceptable ranges and provides help when requested.

The vector-mixer-calibration technique is conducted in two steps. The user characterizes a mixer/filter pair with reciprocal properties first, and this mixer then becomes an additional through standard with which to calibrate the test system. With this step completed, the test system can characterize nearly any reciprocal or nonreciprocal mixer or converter without the need for reconnection of the calibration mixer. Information is also provided about the input and output match of the calibration mixer, which can be used to remove mismatch errors at the input and output of the test system. Since there is no need for multiple mixer connections during the mixer-calibration process, connector repeatability is eliminated as a source of measurement error.

Measurement systems that use traditional techniques for measuring group delay are inherently not well-matched, requiring generous use of attenuators to reduce mismatches of the test system. These attenuators cause serious degradation of the test system's dynamic range and calibration stability. Until the introduction of Agilent's vector-mixer-calibration technique, no method had been proposed to correct for the calibration or reference mixer's input and output mismatch, and thus they have been impossible to determine. The vector-calibration technique is currently one of the only commercially available methods that corrects for input- and output-mismatch effects of the entire test system, providing accurate transmission and reflection measurements of the mixer or converter under test.

To show the viability of the technique, a classic network-analyzer procedure can be used. In this process, a mixer is measured by itself and then with an air line, which is a low-loss, well-matched delay line. In an ideal measurement, the test system should show the conversion loss of the mixer reduced by exactly the loss of the air line, and mismatch effects should introduce minimal ripple on the measurements. The results of measurements performed with an air line scalar and vector calibrations show that ripple in the scalar measurement is nearly an order of magnitude greater than that of the vector-calibrated measurement. P&A: $139,000.00 (E8361A 67-GHz PNA Series VNA), $19,500.00 (frequency-offset measurement capability, typical mixer option configuration); now available for order. Agilent Technologies, Test and Measurement Organization, 5301 Stevens Creek Blvd., MS 54LAK, Santa Clara, CA 95052; (800) 452-4844, Internet: www.agilent.com/ find/PNA.