Abstract-We describe a new RF and microwave power sensor MMIC design. The circuit incorporates a number of advances over existing designs. These include a III-V epitaxial structure optimized for sensitivity, the figure-of-merit applicable to the optimization, a mechanism for inbuilt detection of load ageing and damage to extend calibration intervals, and a novel symmetrical structure to linearize the high-power end of the scale.
Topside metal helps spread the heat dissipated by HBTs and reduces both the thermal resistance at a given temperature, and its temperature dependence. In this paper we describe new methods for including topside metal cooling effects in both an electrical method of extracting HBT temperature and a geometry-based method for calculating HBT temperatures.
I. ELECTRICAL METHODThe electrical method for extracting HBT temperature extends a model [ 11 that uses V h (at fixed Ie) as a thermometer for sensing junction temperature. Previous methods essentially limit temperature and power ranges so that thermal resistance R* can be taken to be nearly constant, then determine this constant (Rth) by curve-fitting to V at various V,, and ambient temperatures (Tamb). In our modification to the electrical method, we take advantage of knowing that a highly temperature dependent thermal resistance can have its variation completely described by just two parameters. We then simply extend the curve-fitting of V to two parameters (RthO and n) instead of one, and intentionally use a wide range of power dissipation and temperature (> 200°C). The key to this method is having the general form for the functional dependence of thermal resistance that can be derived from the Kirchhoff transformation [2]: belle belle 15 (~1.0%) 10.031 (=3.6%) 0.20 8.5 (=0.6%) 0.018 (2.1%) 0.11 Here Rho is the low power thermal resistance at 300°K (= -. .Rb[Tmb=300"K, Pdiss+Ol), and n is associated with the tern-* perature dependence of GaAs thermal conductivity (= T'").One can simulate the effect of putting other thermal resistances in parallel or in series with the resistance in (1). One finds that over a wide range of temperatures and power dissipations that the equivalent thermal resistance also follows the form of (1) but with altered values of RthO and n. Furthermore, this is found to hold whether the added resistances are temperature dependent or not. Hence (1) can be used to describe HBTs with topside metal thermal shunts. Since topside metal has nearly constant thermal conductivity (n=O), the effective value of n in (1) will be reduced to a value less than for n of GaAs (1.135). Using this method, one finds that fitting errors (ms) are only about 0.4 mV (or 0.4"C) compared to -5°C for previous methods. Furthermore, this method allows the thermal resistance to be calculated for any temperature or power dissipation, even outside the range of the experimental data. Figure 1 shows an example of this method at I, = 5 mA. Rth0 results vary by -1% over a wide range of le (Table 1). Variation between devices for Rho is also about I%, while n varies by about 4% between devices ( Table 1)-0 ln 0 0 7 h ZG 7 i = 0 0 cu 0 ln cu 7 2 3 4 5 Vce Fig.1. Vbe and T, vs. v,, and Tamb at 1, = 5 mA, for a typical 2 X 6 pm2 HBT with a pair of 10 x 10 vm2 thermal shunts. Fit has Rth0 = 1530 IUW and n = 0.884. The rms fitting error is 0.32 mV, or about 0.3"C. The chip is AuSn soldered to a large MO shim and probed on a hot chuck.Table 1. Variation of RthO and n between device...
We discuss the many factors affecting the reliability of GaAs HBTs that we have encountered starting from the early days of AlGaAs-emitter HBTs through the present day use of InGaP-emitter HBTs. We discuss both wearout and infancy failure modes and try to distinguish fundamental (i.e., unavoidable) from nonfundamental failure modes. We have found that infant failures are dominated by substrate dislocation density, which can limit long-term-reliable circuit sizes to under ~1000 transistors.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.