Paper preference: Oral or poster. Session: Radiation Effects on Devices, Circuits and S. Authors (both are with the Jet Propulsion Laboratory):

Transcription

1 Total Dose Degradation of Voltage Regulators A. H. Johnston and B. G. Rax Jet Propulsion Laboratory, California Institute of Technology Pasadena, CA USA Abstract Total dose degradation is studied for low-dropout and conventional voltage regulators. Key parameters are identified, and circuit analyses are used to explain differences in degradation behavior. Internal bandgap reference performance is compared with conventional references. Authors (both are with the Jet Propulsion Laboratory): Allan Johnston (principle contact) Tel FAX allan.h.johnston@jpl.nasa.gov Bernard Rax Tel FAX bernard.g.rax@jpl.nasa.gov Mailing address for all authors: Jet Propulsion Laboratory 4800 Oak Grove Drive Mail Stop Pasadena, CA USA Paper preference: Oral or poster Session: Radiation Effects on Devices, Circuits and S 1

2 Total Dose Degradation of Voltage Regulators A. H. Johnston and B. G. Rax Jet Propulsion Laboratory California Institute of Technology Pasadena, California USA INTRODUCTION Total dose effects in bipolar linear integrated circuits remain an important issue, particularly for space applications. The low-cost process used for the majority of commercial designs incorporates substrate and lateral pnp transistors that are particularly sensitive to total dose damage [1]. Linear ICs are often sensitive to enhanced damage at low dose rate (ELDRS), which impacts them in two ways: more damage occurs at the low dose rate encountered in space; and different mechanisms are frequently observed at low dose rate because the ELDRS phenomenon affects pnp and npn transistors differently. Although several studies have been published on voltage regulators [2-5], they remain one of the most difficult part categories to deal with, partly because total dose damage in those components is strongly affected by specific application conditions. This paper discusses mechanisms and circuit design factors that are important for total dose damage in voltage regulators, contrasting them with more basic linear functions, such as operational amplifiers. TECHNICAL APPROACH Several types of voltage regulators were selected for this study. Most were used in various space systems, with radiation requirements between 10 and 20 krad(si), levels which are low enough to allow commercial devices to be qualified by lot-specific radiation sample tests. Sample sizes of 6-10 parts were typically used for radiation testing, using cobalt-60 gamma rays at dose rates between and 0.01 rad(si)/s. Table 1 lists the part types and manufacturers. The research in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration (NASA). Device Type LM117 *LM137 LM2941 LM2953 Table 1. Part Types Used in the Study Function Pos. general purpose Neg. general purpose Pos. low dropout Pos. low dropout Current Rating Manufacturr 300 ma National 300 ma; 1 A National 1 A National 250 ma National *Two different versions of this regulator were included. The devices included two types of point-topoint regulators, as well as two low-dropout regulators. The negative regulator and the low dropout regulators use lateral pnp pass transistors, while the positive regulator uses an npn pass transistor. Electrical tests were made on an Eagle 300 test system, carefully limiting the duty cycle to avoid overheating the device during testing. This is an important consideration for tests of these types of components. For the low dropout regulators, small incremental radiation levels were used to provide better information about specific failure modes than is possible when more widely spaced radiation levels are used. The electrical tests included standard parameters in the device specification sheets, along with measurements of output voltage and dropout voltage for 11 different load conditions. All of the regulators use an internal bandgap reference voltage; the output voltage is derived from that reference by means of an external resistor network. The stability of the bandgap reference is a basic limitation for their performance, regardless 2

3 of the load. The circuit parameters that provide the most insight into their performance are the bandgap reference, output current, and dropout voltage. However, unlike op-amps and comparators, where the power supply voltage usually has only a slight effect on radiation damage, power supply voltages have a large effect on the degradation of voltage regulators. This is a major stumbling block in establishing electrical parameters that can be used to characterize their circuit performance, unless each specific circuit application is included in the measurement set. TEST RESULTS A. Tests at Low Current These regulators require a minimum load current in order to operate properly, typically 5 to 10 ma. A measurement at low current was used to measure the internal bandgap reference voltage, which is nominally 1.25 V. Figure 1 shows degradation of the bandap reference of the LP2941 (normalized). Significantly less degradation takes place when the device is tested at high dose rate compared to the results at low dose rate. Furthermore, the degradation at high dose rate is nonlinear with dose, saturating at about 2.5%. Fig. 1. Percent change in output voltage for the LP2941 voltage regulator with a load current of 10 ma. Typical results are compared for devices tested at different dose rates, irradiated with all pins at ground. Degradation of the bandgap reference of the LM117 is shown in Fig. 2. Note that far larger changes occur for that device, compared to the changes observed for the low dropout regulator in Fig. 1. Fig. 2. Percent change in output voltage for the LM117 voltage regulator with a load current of 5 ma. The device was irradiated at rad(si)/s with all pins at ground. For both types of regulators, the reference voltage under lightly loaded conditions has only a slight dependence on power supply voltage. B. Maximum Load Current The maximum load current is a complex parameter, particularly for the low dropout regulators. Internal current limiting and/or shutdown circuits are present in all of the regulator designs, and those circuits will interfere with measurements of the load current if the output is shorted to ground. It is possible to determine the maximum output current by applying a pulsed load, limiting the voltage compliance to a value that is approximately 0.5 V below the programmed output voltage, but the conditions where such measurements are valid may change as the device degrades. Manual measurements using an oscilloscope provide a way to ensure that the current limiting circuitry has not been activated, but this is difficult to implement for voltage regulators with high output current. A more thorough discussion of load current degradation and measurement methods will be included in the full paper. C. Dropout Voltage It is more straightforward to measure dropout voltage compared to short-circuit current. Figure 3 shows how dropout voltage changes after irradiation for several different load conditions for the LP2941 regulator, irradiated without bias. Several points are evident. First, even with relatively light loading the dropout voltage 3

4 increases to more than 1 V at higher radiation levels. Although that value is well above the specification limit, it is still much lower than the dropout voltage of the LM117 voltage regulator, making it possible to use the LP2941 under restricted load conditions. Second, much larger changes occur in dropout voltage with higher load conditions. Although some applications may be able to tolerate such high values of dropout voltage, this can be considered the bounding region for most low-dropout regulators. mechanism involved in that region does not depend directly on pnp transistor gain. Fig. 3. Dropout voltage vs. total dose for the LP2941 with all pins grounded. For the three highest load conditions, a third region is evident, where degradation of the pass transistor affects the output. The closely spaced radiation levels allow these various regions to be distinguished, which would not be the case for more widely spaced levels. A similar set of data for a part that was biased during irradiation is shown in Fig. 4 (most parts that are sensitive to ELDRS exhibit more damage when they are irradiated without bias compared to the biased case [3]). For load currents below 500 ma, smaller changes occur compared to the device irradiated without bias (Fig. 3). However, there is very little difference between these two cases at higher currents, when larger changes start to occur in the dropout voltage. Comparing these two figures, the difference in damage expected for the two different bias conditions is clearly evident in regions 1 and 2, but does not appear to affect the response of the devices in region 2. This suggests that the Fig. 4. Dropout voltage vs. total dose for the LP2941 biased actively with a nominal output voltage of 5 V. In contrast to the results for the LP2941, degradation of dropout voltage for the LM117 was only weakly dependent on load conditions, as shown in Fig. 5. The minimum input/output voltage difference for this part is 3 V. The weak load dependence makes it easier to accommodate the degradation by parameter derating, compared to the LP2941. Fig. 5. Dropout voltage vs. total dose for the LM117 with all pins grounded. DISCUSSION Although it is possible to make some general observations, the specific circuit design used in voltage regulators has a large effect on the way that radiation damage affects circuit characteristics, just as for operational amplifiers. 4

5 For example, the reference voltage and start-up circuit of the LM117 are shown in Fig. 6. Lateral pnp transistor Q18 is directly involved in the bandgap circuit. As discussed in [6], the basic bandgap reference circuit relies on high transistor gain, and typically degrades by less than 1% when npn transistors are used, but the results in the previous section show that far more degradation takes place in the bandgap reference of these devices compared to stand-alone references. A SPICE analysis of this circuit, using a pnp model that degrades to 20, shows that the extreme degradation of the reference voltage in the LM117 is due to a combination of degradation of the lateral pnp within the bandgap reference along with mismatch in the current sources, due to the decrease in gain of the multiple collector current source transistors. Further details will be included in the full paper, along with analyses of the low dropout regulator circuits. Fig. 6. Circuit schematic of the internal reference and current sources of the LM117 voltage regulator. Even though the fundamental effects on internal transistors are the same, there are important differences in the way that voltage regulators and conventional linear circuits are affected by radiation damage. For op-amps and comparators, the normal input parameters offset voltage, bias current, and offset current usually dominate the radiation response. They are easily measured, and do not have the strong dependence on power supply voltage and load conditions that are an important practical difficulty for voltage regulators. Voltage regulators are far more involved. High currents are always required, and the interdependence of the input/output voltage difference and load current make it far more difficult to measure radiation degradation with sufficient generality to apply to different circuit applications. The full paper will include results for the other two circuit types, and further develop a more general approach to measuring and analyzing their radiation behavior, and establishing derating factors for circuit applications. Unit-to-unit variability, which is particularly important for high-current applications, will also be discussed. REFERENCES [4] S. S. McClure, et al., Dose Rate and Bias Sensitivity of Low Dropout Regulators, 2000 IEEE Radiation Effects Data Workshop. pp [1] R. L. Pease, Total Ionizing Dose Effects in Bipolar Devices and Circuits, IEEE Trans. Nucl. Sci., 50(3), pp (2003). [2] J. T. Beaucour, et al., Total Ionizing Dose Effects on Negative Voltage Regulator, IEEE Trans. Nucl. Sci., 41(6), pp (1994). [3] R. L. Pease, et al., Enhanced Low-Dose-Rate Sensitivity of a Low-Dropout Voltage Regulator, IEEE Trans. Nucl. Sci., 45(6), pp (1998). [5] L. Bonora and J. P. David, An Attempt to Define Conservative Test Conditions for Total Dose Testing of Linear Bipolar Circuits, Nucl. Sci., 47(6), pp (2000). [6] B. G. Rax, et al., Degradation of Precision Reference Devices in Space Environments, Nucl. Sci., 44(6), pp (1997). \ 5