Temperature Dependent Microwave Attenuator Models

Parent Category: 2020 HFE

By Hetvi Patel, Eric Valentino, Hugo Morales and Larry Dunleavy

Accurate attenuator models are described that provide designers with the flexibility to predict temperature-dependent and substrate-dependent behavior, and that conveniently scale with the nominal attenuation value. Example models are demonstrated and validated for API Inmet TCAF-N05 series passive temperature-variable surface mount chip attenuators from DC through 18 GHz over a temperature range of -40o to 150oC and a wide range of test boards. Additionally, an amplifier temperature stabilization application example illustrates the full utility of the broadband and temperature-dependent aspects of these innovative models.

Introduction

Attenuators are extensively used in high-frequency applications to control signal levels within circuits such as receivers or transmitters. Attenuators are usually comprised of either a T- or Pi-network of resistive elements (e.g. resistive layers with temperature-compensating characteristics). These resistors are selected for the network based on the attenuator’s desired frequency, temperature and power-handling capabilities [1]. Since the gain of microwave amplifiers tends to decrease as the temperature increases, temperature-compensating attenuators are designed to have the opposite trend: their “gain” increases (attenuation decreases) as the temperature increases. As a result, an attenuator with inverse behavior over temperature in comparison to an amplifier can be cascaded with that amplifier, cancelling the two responses and thereby producing constant output power over temperature [2]. For this type of signal-leveling application and others, a temperature-dependent substrate-scalable attenuator model could greatly aid designers, if only to understand and assess whether the amount of anticipated variation in attenuation with temperature over a given frequency range can be tolerated.

To develop such a model, reliable data must be obtained over a range of temperatures. For additional flexibility, data can also be acquired using various board types. This data, in cooperation with a physically motivated equivalent circuit and associated extraction techniques, can produce a model that demonstrates both good measured-to-modeled agreement and accurate scalability to conditions other than those in which it was modeled. An example of such a model will be explored in this article.

TEMPERATURE DEPENDENT ATTENUATOR MODEL

The Modelithics simulation model for the API Inmet TCAF-N05 attenuator series, for which the image of model symbol is included in Figure 1, offers the designer remarkable flexibility thanks to its part value, substrate, pad and temperature scalability. This paper focuses on the prediction of temperature dependence of the attenuator model, but also presents the additional features such as substrate for a specific PCB mounting layout and attenuation value scalability in the first section. The PCB mounting configuration and layout diagram used for this full-wrap termination chip are shown in Figure 2. A thru-reflect-line (TRL) VNA calibration places the reference planes at the outer edge of the solder mounting pads for the characterization on each of the three substrates used: 6.6-mil Rogers 4350B (measured up to 18 GHz), 30-mil Rogers 4350B (up to 12 GHz), and 60-mil Rogers 4003C (up to 6 GHz). Using these different board types enables Modelithics’ application of proprietary techniques to extract a substrate-scalable model that is valid for an H/Er ratio (in mils) ranging from 1 to 17 mils, where H is the substrate thickness and Er is the dielectric constant. Figure 3 demonstrates that the model, 6 dB in this example, shows good agreement with measured data when the component is mounted on various substrates. Figure 4 shows a comparison of the performance of multiple attenuator values (2, 4 and 6-dB) on 30-mil Rogers 4350B at 25°C.

 2009 Attenuators fg01

Figure 1 • Schematic view of the Modelithics TCAF-N05 attenuator model.

 2009 Attenuators fg02

Figure 2 • TCAF-N05 Attenuator mounted in 2-port series configuration. DUT image on the measurement test fixture is shown on the left. The dimensioned layout, used to measure the parts on all substrates, is shown on the right.

2009 Attenuators fg03 

Figure 3 • Typical model performance for a 6 dB attenuator vs. substrate thickness. Legend: RED 4-mil Rogers 4350B, BLUE 30-mil Rogers 4350B, PINK 60-mil Rogers 4003C, Lines - Model, Symbols - Measured data.

 2009 Attenuators fg04

Figure 4 • Attenuation level (S21 dB) vs. frequency for attenuator values of 2, 4 and 6 dB on 30-mil Rogers 4350B substrate at 25°C. Legend: RED 2 dB, BLUE 4 dB, PINk 6 dB, Lines - Model, Symbols - Measured data. 

Pad scalability is a useful feature that allows the user to set the pad dimensions to any value within a specified range when using the built-in pad model. This feature also gives users the option to disable pads and use an EM pad representation via co-simulation. The particular response of the pads with the specified dimensions can then be accounted for separately. Similarly, substrate scalability allows the model to be used on any substrate within the validated range for the ratio of height to relative permittivity. Ranges for any Modelithics model with substrate or pad scalability can be found in the Modelithics model information datasheet, as well as other useful information describing the models’ validations and pertinent measurement details. The model also interpolates performance between part values, offering users even more flexibility. Figure 5 shows part value interpolation on 30-mil Rogers 4350B from 2 to -6 dB in -0.5 dB steps, interpolating between the whole-number attenuation values used to fit the model.

2009 Attenuators fg06

Figure 5 • The TCAF-N05 attenuator model interpolation results on 30-mil Rogers 4350B, 2 to 6 dB in 0.5 dB steps.

The model and measurement results for the 6 dB attenuator at -40°C, 25°C and 150°C are plotted in Figure 6, showing good agreement over both temperature and frequency.

2009 Attenuators fg06

Figure 6 • Attenuation level (S21 dB) vs. temperature for the 6-dB attenuator model. Measurements on 30-mil Rogers 4350B. Legend: RED -40 ˚C, BLUE 25 ˚C, GREEN 150 ˚C, Lines - Model, Symbols - Measured data.

Plated via holes and grounding effects from Figure 2 are included in the model and measurements.

The TCAF-N05 attenuator series operates over a frequency range of DC to 12 GHz at temperatures from -55°C to 150°C. Included in the series are 2, 3, 4, 5 and 6 dB attenuators in which nominal attenuation is relatively flat versus frequency up to 12 GHz when directly mounted onto a metal ground carrier for the ground wrap terminal and the RF mounting pads are well matched to 50 ohms. This family has a nominal temperature coefficient of attenuation (TCA) of -0.005 dB/dB/°C. TCA can be used to predict how much a part’s attenuation deviates from the nominal value due to temperature, with the nominal attenuation value specified at 25°C. Given a temperature T (in °C) and a nominal part attenuation Anom (in dB), we can find the predicted attenuation Apred (in dB) as affected by the temperature change as follows:

Apred = Anom + (T - 25)(Anom * TCA)

Similarly, one can define an effective TCA. TCAeff can be calculated by replacing Apred with an observed attenuation Aobs given an Anom at temperature T.

Plotted in Figure 7 are comparisons between attenuation and effective TCA over temperature for measurements performed versus corresponding Modelithics models for the entire TCAF-N05 family.

2009 Attenuators eq01

2009 Attenuators fg07

Figure 7 • TCAF-N05 family (2 to 6 dB) at 2 GHz for measured and modeled part values on 30-mil Rogers 4350B. Left plot shows attenuation vs temperature, right plot shows effective TCA vs temperature.

Design Example: Temperature Stabilization of Qorvo TGA8344-SCC Monolithic Amplifier

To demonstrate the utility of the temperature scalability feature of this component model, a simulation using temperature compensation of an amplifier was performed. Simulations are performed on 30-mil Rogers 4350B substrate. The goal is to present a circuit with the minimum variation in gain over the temperature range of -10 to +60°C. This could be an important design requirement for communications equipment in outdoor or industrial settings. While adding an attenuator does mean that some gain must be sacrificed, it also yields benefits beyond temperature insensitivity. Doing so additionally improves stability and input return loss.

The Qorvo TGA8344-SCC is a monolithic amplifier that has about 19 dB of gain over an operating frequency range of 2 to 18 GHz. The Modelithics model of this device is unique as it offers temperature scaling of S-parameters. True to the physical amplifier, the model reveals decreasing gain with increasing temperature. Figure 8 shows S11 and S21 of the amplifier over the operating frequency range at a temperature of 25°C.

2009 Attenuators fg08

Figure 8 • S11 and S21 of TGA8344-SCC MMIC Amplifier at 25°C and 60°C on a 30-mil thick substrate.

The effect of temperature on the gain of the amplifier at 2 GHz is shown below in Figure 9. The temperature varies from -10 to +60°C, and, as indicated by the markers, the gain drops 1.46 dB over the 70-degree increase. This corresponds with an average slope of -0.021 dB/°C over the temperature range.

2009 Attenuators fg09

Figure 9 • S21 of TGA8344 Amplifier vs. Temperature at 2 GHz.

Assuming the nominal TCA value of -0.005 dB/dB/°C for the entire family (typically specified at DC), to compensate for a -0.021 dB/°C change in gain at 2 GHz, one would likely select the 4-dB attenuator to cascade with the amplifier. This is because, by multiplying the TCA with a 4 dB nominal attenuation, one finds the attenuator should have an S21 vs. temperature slope of 0.020 dB/°C, which would best cancel out the temperature dependence of the amplifier. Remember that a decrease in attenuation with increasing temperature is the same as an increase in the dB value of S21. Using the accurate, measurement-validated Modelithics models of the TCAF-N05 family, which are fit over this temperature range and include substrate effects, we can confirm if the 4-dB value does, in fact, give the flattest temperature response with a quick simulation.

The left side of Figure 10 reveals the change in attenuation for the 4- and 5-dB TCAF-N05 attenuators over the temperature range of -10 to +60°C. The 4dB attenuator model shows a 1.25 dB increase in attenuation over the temperature range of 70 degrees (an average slope of 0.018 dB/°C), as fit to data measured over temperature by Modelithics for mounting conditions specified in the model datasheet. Swapping in the 5-dB attenuator model reveals a 1.54-dB increase in attenuation (0.022 dB/°C avg. slope), which is closer to the ideal 1.46-dB shift (0.021 dB/°C avg. slope) needed to cancel the average temperature dependence of the amplifier in this design.

2009 Attenuators fg10

Figure 10 • Slopes of 4 and 5 dB attenuators along with the results of placing each on input side of

The right side of Figure 10 illustrates the respective responses when cascading 4- and 5-dB attenuators with the TGA8344 amplifier over the same -10C to +60C temperature range. As can be seen, the design with the 5-dB attenuator exhibits a slope ≈1/3 that of the 4-dB attenuator over the investigated temperature range. The 4-dB value gives an average S21 slope of -0.006 dB/°C, and the 5-dB value produces a slope of -0.002 dB/°C. A summary table of these slopes is presented in Table 1. Attenuators are placed before the amplifier as shown in Figure 11.

2009 Attenuators tb01

Table 1 • Summary of temperature slopes

This example shows how having highly scalable, measurement-validated models such as the Modelithics TCAF-N05 family can prevent costly and time-consuming design revisions. Table 1 gives a comparison of the average slopes of S21 in dB over the -10 to +60 °C range investigated for the various designs discussed in the example.

2009 Attenuators fg11

Figure 11 • Schematic representation of attenuator and amplifier.

Summary

Microwave attenuator models that scale with temperature and other design parameters give designers considerable flexibility in evaluating and optimizing their designs. One such application is temperature compensating attenuators, which are widely used for microwave amplifier temperature compensation. Modelithics’ measurement-validated model for the API Inmet TCAF-N05 series was demonstrated to be a useful tool for such designs. In general, temperature dependent, pad scalable, and substrate scalable attenuator models demonstrated herein are very useful in pre-fabrication simulation and assessment of performance of complex assemblies over a wide range of design parameters.

About the Authors

Hetvi Patel, Eric Valentino, Hugo Morales and Larry Dunleavy are with Modelithics.

Acknowledgments

This article is based on the scalable model developed by Modelithics, Inc. for the API Inmet TCAF-N05 attenuators. Special appreciation to API Inmet for funding the modeling work and providing the attenuator samples to complete the characterization needed for the modeling effort. The authors would like to thank Chris DeMartino and Isabella Bedford of Modelithics for their helpful editing comments. API would also like to thank John Woods, formerly with the company, for his contributions to this article.

References

[1] Jimmy Dholoo, Rabih Sinno, “Understanding Temperature & Power Coefficient In Attenuators”, MCE/Weinschel Corporation.

[2] North Mankato, “ATV_Presentation.pdf”, Thin Film Technology Corporation.

[3] Larry Dunleavy, Isabella Delgado, Harvey Kaylie (Mini-Circuits), “Modeling Grounding and Substrate Effects in BroadBand Miniature Surface Mount Attenuators”, Microwave Product Digest, March 2016.