Computer-Aided Predictive Reliability Risk Analysis for High-Temperature Operation of Subsurface Electronics Development

Current design and development of reliable printed circuit board assembly (PCBA) to assure long service life in a hot well environments presents numerous challenges [1]. The necessity of stress analysis by using computer-aided design (CAD) tools for the downhole electronics has been addressed in current literature [2]–[3]. The CAD tools predict MTBF of a PCBA based on the electrical, thermal, and dynamic stress analysis and identifies the stress-exceeded components and lower MTBF parts [4]. The CAD tools also point out critical design errors and weaknesses (a root cause of potential failure) from a reliability standpoint [5]–[6]. The early design change opportunity in the production phase of PCBA product development enables time and cost savings by minimizing the product design cycle iterations. The other added value of this DfR method is the virtual pre-qualification test and the reliability acceptance test that confirms the PCBA reliability target. The swim lane diagrams to plot the interconnections between processes and product teams are shown in Fig. 1.

BQR fiXtress^™ is the software tool that integrates the circuit network topology and functional design error detection with stress analysis, thermal and service life prediction as early as during the schematic and layout definition, before the PCB layout is complete [7]. PCBA design information including BOM, Netlist, component package pin information, physical component placement, mass spatial distribution, and other required component data are imported from the PCBA electronic design software tool as shown in Fig. 2.

When the PCBA design input data and initial quality control (QC) pre-processing is complete, the passive and active component libraries must be filled with information such as technical characteristics and engineering tolerances from technical datasheets. The circuit component package and corresponding pin libraries also must be filled with connectivity information such as external signals, default operating states, functional description, power consumption, operating frequencies and input/output (I/O) DC values. For precise simulation, a schematic review is required to verify if the design is accurate and represents actual operational conditions.

The many difficulties associated with making high-temperature prediction have been noted in the literature [8]. To identify the critical components operating in excess of de-rated values at high temperature, the reliability prediction method from FIDES 2009 [9] has been applied in fiXtress. The general reliability model of FIDES 2009 is based on the following equation:

where;

λ is the item failure rate,

Σ_{Physical_contributions} represent a mainly additive construction term comprising physical and technological contributing factors to reliability, and

Π_{Process_contributions} represent a multiplication term that represents the impact of the development, production and operation process on reliability.

In practice, this equation becomes:

where;

λ_Physical represents the physical contribution,

Π_PM represents the quality and technical control over manufacturing of the item and,

Π_Process represents the quality and technical control over the development, manufacturing and usage process for the product containing the item.

calcePWA^® is the physics-of-failure software that was used. It offers user-friendly modular and integrated analysis of electronic assemblies. The software was developed by the University of Maryland Computer Aided Life Cycle Engineering (CALCE) [10]. When evaluating PCBA designs, engineers should balance the requirements of multiple physics disciplines in the PCBA reliability design optimization. The finite-element analysis (FEA) tool is required to solve the engineering and mathematical physics problem of the entire system in one operation. ANSYS^® Workbench (Professional License) was used in the reliability analysi process and for calcePWA simulation result validation (Fig. 3).

PCBA data exchange files are archived from the electronic design software project file to extract the PCBA layer stack-up sequence details, components package details, and temperature effect details. When the pre-processing is complete the thermal analysis is carried out using specified environmental condition for the PCBA. The substrate, case and junction temperature of individual component are shown in the thermal analysis report.

In the vibration analysis, appropriate boundary conditions and mesh settings must be analyzed to evaluate the structural components and areas which are sources of the first three most-dominant natural frequencies and flexural displacements in each individual PCBA assembly and structural component.

The virtual test can be performed through these thermal and vibration CAD-driven DfR analyses because the test criteria and requirements utilized in the design test qualification of electronic assemblies can be applied into calcePWA as the PCBA life mission profile. During the analysis, the lifetime of the board is estimated from the failure analysis, and the root cause of potential failure is analyzed.

The most convenient and commonly used method of predicting the thermal and vibration aging and failure mode responses is through the creation of FEA models. FEA is used to simulate individual physical phenomena and the interaction between multiple physics disciplines interaction. The FEA workflow chart is shown in Fig. 4.

Because downhole electronic components are prone to failure from shock or vibration loads at the high-temperature environmental conditions, proactive PCBA reliability analysis in electrical, thermal and mechanical contributes long service life and minimizes risk of potential product failures. To precisely predict when this failure may occur, the CAD tools fiXtress, calcePWA [10] and ANSYS must be used in a complementary technique. Even though calcePWA and fiXtress are comprehensive DfR CAD tools, fiXtress is especially valuable for electrical stress analysis based on the PCBA schematic, and calcePWA performs the PCBA layout analysis (Table 1).

Even though the simple way to measure the fatigue life by component mass, geometrical dimension, material property information is to conduct a calcePWA simulation, an ANSYS simulation is required for validation purposes. ANSYS provides detail fatigue models and temperature distributions of complex components, arbitrary component mounting and placement, substrate cavities, mounting brackets, and wire bonding. The process steps in sequential order using three CAD tools are shown in Fig. 5.

From draft fiXtress simulation result with design sources and loads information, the calculated power values and other electrical parameters are used as input for calcePWA and ANSYS simulation to remodel the thermal.

The calcePWA pre-processing entry and fatigue model will be used in ANSYS FEA.

After completing the ANSYS analysis, the thermal and displacements value are compared with calcePWA for validation. Any missing model defined in the ANSYS analysis will be implemented and updated the calcePWA simulation model.

Identifying and prioritizing the risky components, and providing MTBF calculations from reliability-based algorithms of fiXtress and calcePWA.

The objective of the simulation for this example is intended to find the reliability risk within a 10-year operating life. Because the long-term reliability test was conducted using four MDEC PCBAs for 17,109 hours at 155° C in the operating condition, this test result can be used to validate the simulation result. The reliability parameters to estimate the MTBF based on the test result are shown in Table 2 for reference.

The schematic review output is conducted to identify any schematic issues. Here is one good example. The calculated output current of US2901 output driver in Fig. 5 is 4.4mA. The typical current consumption of the status indicator block on the board is 5mA per the data sheet. Even though the minimum recommended supply voltage is 4.5V per the data sheet, the applied supply voltage is 5V - 0.6V = 4.4V. Therefore, changing the value of resistor R26 in Fig. 6 to the smaller value is highly recommended.

The impact of the stress de-rating analysis based on the electrical characteristics including frequencies, voltage, and current via temperature changes must be analyzed. As one case is considered as shown in Fig. 7, replacing the components with ones that are suitable for the voltage rating is highly recommended.

All informative component power consumption evaluated from fiXtress shown in Fig. 8 was implemented for calcePWA thermal analysis to estimate hot zones as shown in Fig. 9.

fiXtress simulation from the de-rating analysis and updated load information was given as input to calcePWA for remodeling the thermal analysis (Fig. 10).

The PCBA is subjected to random vibration stress during the transportation or the actual donwhole deployment. From the simulation, a root cause of the potential future failure (a location facing major stress) due to the dynamic stress can be observed. The calcePWA simulation showed which component or solder joint faced the most critical stress (Fig. 11 and 12).

Vibration and shock analysis from the simulation is useful to find the displacement information on the board, while the natural frequency results could be in question. To cover this shortcoming, PCBA 3D step was created for ANSYS FEA to update the missing models. This validation study using ANSYS enabled evaluation of the calcePWA approach. Under the load detail following the mission profile, the power dissipation, PCBA layer stackup, and MTBF calculation were analyzed. The life cycle testing results can be used to find the weakest solder connections on the PCBA, which could be instructive for lab testing and design improvement. Because the test results showed that the board passed the test representing the life requirement of 10 years, the MTBF calculation from the simulation was validated (Fig. 13).

This section presents other successful examples from this stress analysis CAD tools that are capable of performing thermal and vibration stress analysis based on the engineering practice, characterization tests, and operational environmental stress.

For one existing downhole PCBA, the previous thermal imaging report identified that the hot spots are on only U1 and U6 as shown in Fig. 14. Now fiXtress indicates other hot spots (high junction temperature) on U12 (DC -DC Converter) and U21 (Schmitt Trigger Inverter Gate) according to the simulation result, especially for U21. The thermocouple sensor and infrared radiation image measurements were used to validate the actual observations' correlation with the thermal simulation predictions.

Just like the first case study in this paper, the PCBA 3D step was created for ANSYS FEA to update the missing models. This validation study using ANSYS enables evaluating the calcePWA approach with the same boundary conditions. Response analysis on the PCBA was performed and root mean square of acceleration (Grms) measured over the relevant vibration frequency range with the given input Power Spectra Density (PSD) data at different locations of assembly. In the next example, calcePWA model was meshed with the Tet and Hex element in ANSYS. The critical regions was meshed with finer divisions and the dummy mass (point) elements used to simulate the hinged (single-point support) behavior boundary conditions in ANSYS, as shown in Fig. 15.

Response analysis on the PCBA was performed and Grms was measured with the given input PSD data at different locations of assembly. From the modal analysis, the first, second, and third natural frequencies from calcePWA were validated. The harmonic response analysis (Grms calculated over relevant frequency range from 0 to 1,000Hz) showed that the maximum response and the maximum acceleration occurred at the locations marked in red and shown in Fig. 16.

This paper illustrated a reference DfR benchmark design example that yielded a highly reliable printed circuit board assembly (PCBA) that delivers performance to a design specification MTBF reliability spec (DfR) withstanding high temperatures and vibration, environmental and operational stresses encountered downhole. A comprehensive design for reliability (DfR) based on computer-aided design (CAD) tools analyzed electrical and thermal stress, service life prediction, and identified potential design errors for reliability fast and low-cost iterative design improvement at the schematic-level early in the design cycle when compared to a less sophisticated and more costly approach flow sequence involving design, test, break, analyze and iteration design cycle.

This stress analysis CAD tools can perform reliability risk assessment based on historical failures and operating environment (i.e., engineering practice, failure characterization tests, operational reliability performance data, repair and maintenance database, operational environmental stress data, condition based maintenance driven by the monitoring of key PCBA measurements, etc...), detecting design weaknesses, providing accurate and reliable MTBF predictions and life predictions for the PCBA design iteration. This CAD-based DfR process flow has been shown to improve PCBA performance and reliability by significantly reducing the product development reliability delivery failures, reducing project development costs and schedule, design qualification tests expenses and improving overall product quality and reliability. The experience has shown that this CAD-based DfR product development process flow has enabled product time-to-market cycle acceleration at lower benchmarked costs while consistently delivering the expected product quality, availability, durability and reliability to meet customer expectations. CAD tool-based DfR has significantly improved our product development benchmark metrics and will be the focus trend for CAD tool capabilities investment further enabling our product lines competitiveness and agile market commercial launches.

The authors would like to thank Duraivel Arumugam and Baranitharan Palanisamy for conducting most of the pre-processing and simulation. We would also like to thank Earl Isidro for his feedback for the simulation as the circuit designer and for executing the long-term life test.