Engineers working with cable, terrestrial or satellite TV applications are frequently required to make S-parameter measurements on these circuits. But how do you measure [S]-parameters of 75 Ω device under test (DUT) on a 50 Ω vector network analyzer (VNA)? If the situation warrants the cost, the answer is to buy lab equipment designed specifically for measuring 75 Ω circuits. Otherwise, using a minimum loss pad (MLP) to transform the conventional 50 Ω test port impedance to the 75 Ω DUT provides an inexpensive, easy way to get reasonable measurements.

When an IC manufacturer specifies the input return loss (|S11|) of a new cable TV LNA, the measurement is necessarily referred to 75 Ω. That is to say, if |S11| = -30 dB (reflected power is only one part in a thousand — essentially a perfect match), the idea is that when driven with a 75 Ω source impedance, the device input will allow virtually all of the power to be transferred to the LNA.

The same tuner input will not offer good return loss when driven from a 50 Ω source impedance, like those of the typical VNA, signal generator, NF meter, etc. Directly connecting this perfectly matched TV tuner input to a 50 Ω VNA will yield an |S11| measurement something close to -14 dB — with reflected power now one part in 25. So with this same 50 Ω VNA, how can we verify that the TV tuner input is as good as we say it is?

A matching circuit is required. It should have flat frequency response and the lowest insertion loss possible. The industry-standard answer to this is the MLP, a simple resistive network of Figure 1. The key feature of this network is that it transforms a 75 Ω DUT load impedance into 50 Ω for the measurement instrument and transforms the 50 Ω source impedance of the instrument to the native 75 Ω impedance of the DUT. In this way, reflections are removed, the response is flat, and the loss of the network is easily backed out of the measurement to get to the DUT. The math required to transform Z_LOAD into Z_LOAD' is straightforward. The resulting expression for Z_LOAD' describes the cascaded impedance of the MLP and the DUT, as seen from the measurement port (R_SOURCE). Turning the equation around and solving for Z_LOAD in terms of Z_LOAD' offers a way to back out the effects of the MLP and to determine the true Z_LOAD from the measurement data taken at Z_LOAD'. The result is provided below:

A sanity-check calculation proves sound. Assume we just made an impedance measurement of a 75 Ω resistor through the MLP. The assumption is that the VNA will measure R_LOAD'=50 Ω (infinite return loss), and we expect the math to tell us that this result came from a load resistor of 75 Ω. Let R_LOAD'=50 Ω, and we see that with R₁=43.3 Ω and R₂=86.6 Ω, we get Z_LOAD=75 Ω as expected.

This simple expression could be made more useful by breaking up the real and imaginary components and using a spreadsheet to do the calculations on the bench.

In a practical example, let's say we want to measure S21 of a cable/terrestrial TV LNA like Maxim's MAX3558 Quad LNA. The DUT is inserted in the test setup with MLPs at the input and output, as in Figure 2. Calibrate the VNA as usual, not including the MLPs in the cal. Connect port 1 to the 50 Ω side of one MLP, and connect the 75 Ω side to the LNA input. Do the same for one of the outputs and port 2 on the VNA.

Make the S21 (forward gain) measurement. The VNA will indicate a gain at 500 MHz near -5 dB. Back out the 12.0 dB insertion loss from the two MLPs and their connectors/adapters, and we see the LNA is providing a 75 Ω power gain of about 7 dB.

At frequencies above several hundred MHz, a PCB-mounted MLP built from 0402 resistors brings measurement accuracy into question. Parasitic effects break the assumption that this network is purely resistive — cases like this require a more complicated approach to the problem. One method would be to fully characterize the MLP, and use a Smith Chart to more accurately back out the effects of the matching circuit. Another solution is to use an inductor-based transformer to do the impedance transformation with much lower loss. RF transformers are described in terms of their impedance transformation ratio, not the turns ratio, so find one described as “75:50.”

For most general lab applications below 1 GHz, a PCB-mounted minimum loss pad built from 1 percent 0402 or similar resistors offers a quick and easy means to test a 75 Ω circuit with 50 Ω lab equipment. In most cases, the only correction factor required is the insertion loss of the MLP — 5.7 dB plus any addition connectors. Difficult calculations or Smith Chart work is often not required to make basic [S]-Parameter measurements.

James Liu and Brian Whitaker are wireless strategic applications engineers at Maxim Integrated Products, Sunnyvale, Calif.

RSS    Save to Del.icio.us  Digg This