Downloads

Download as Word Document (DOCX)

Case Study: Using Claude for Enhanced Feedback in an Analog Circuit Design Course

Course Context

Course: Analog Electronics Design (EE 3744)
Level: Junior
Enrollment: 65 students
Prior Format: Weekly lab assignments with delayed feedback
Tools: Circuit simulation software, breadboarding equipment
Faculty: Dr. S, Associate Professor of Electrical Engineering

Implementation Challenge

Dr. S identified several challenges in her analog circuit design course that AI assistance could potentially address:

Feedback Bottleneck: Limited time to provide detailed feedback on student circuit designs
Varying Preparation: Students entered with different levels of electronics background
Troubleshooting Delays: Students often waited days for help with circuit problems
Self-Assessment Limitations: Students struggled to diagnose their own design flaws
Documentation Quality: Technical reports often lacked appropriate detail and clarity

Implementation Goals

The integration of Claude aimed to:

Provide rapid preliminary feedback on circuit designs
Develop students' circuit troubleshooting skills
Improve the quality of technical documentation
Support varying levels of student preparation
Free instructor time for higher-level guidance

Implementation Process

Phase 1: Faculty Preparation (Summer before semester)

Tool Selection and Exploration:
- Tested multiple AI tools for technical accuracy in electronics domain
- Selected Claude for its stronger ability to analyze circuit diagrams
- Created Claude Pro account for advanced features
- Developed prompt templates for circuit feedback
- Tested limitations with complex analog circuits
Feedback Framework Development:
- Created rubric for distinguishing AI vs. instructor feedback roles
- Developed circuit-specific prompting guides
- Created sample circuits with common errors for testing
- Established process for validating AI feedback accuracy
Integration Planning:
- Redesigned lab workflow to incorporate AI feedback steps
- Created documentation requirements for AI interactions
- Developed training materials for students
- Updated grading criteria to include evaluation of AI-assisted revisions

Phase 2: Student Onboarding (First two weeks)

Introduction Session:
- 50-minute workshop on using Claude for circuit analysis
- Demonstration of uploading circuit diagrams and schematics
- Example prompts for different types of circuit feedback
- Discussion of Claude's limitations with analog circuits
Guided Practice:
- Provided sample circuit with intentional errors
- Students practiced getting and evaluating Claude feedback
- Instructor-led discussion of response quality and limitations
- Compared Claude's feedback with instructor feedback
Feedback Literacy Development:
- Trained students to critically evaluate AI feedback
- Emphasized verification using simulation and first principles
- Discussed complementary roles of AI and instructor feedback
- Created guidelines for documenting AI assistance

Phase 3: Integration Throughout Semester

Structured Feedback Process:
- Students submitted initial circuit designs for weekly labs
- Used Claude to get preliminary feedback before submission
- Documented Claude's feedback and their response to it
- Instructor provided higher-level feedback focusing on areas beyond AI capabilities
Progressive Implementation:
- Early labs: Basic amplifier circuits with guided prompts
- Mid-semester: Filter circuits with student-developed prompts
- Later labs: Complex systems with integrated troubleshooting
Documentation and Reflection:
- Students maintained logs of Claude interactions including:
  - Circuit design challenges faced
  - Prompts used (including iterations)
  - Evaluation of Claude's feedback
  - Actions taken based on feedback
  - Verification methods used

Implementation Examples

Example 1: Structured Feedback Workflow

Lab Assignment Component:

Before submitting your final amplifier design, complete the following AI-assisted review process:

Upload your circuit schematic to Claude

Use the "Circuit Analysis Prompt" to get initial feedback

Implement any corrections you agree with and document your changes

If you disagree with any AI feedback, explain why using circuit theory

Include both the original feedback and your revision process in your lab report

Student Prompt Template:

I'm an electrical engineering student designing a common-emitter amplifier circuit.
Please analyze this circuit schematic and provide feedback on:

[Student uploads circuit diagram]

1. Biasing stability and operating point analysis
2. Small-signal gain calculation review
3. Potential issues with component values chosen
4. Suggestions for improving performance
5. Any errors you can identify in my circuit topology

For each point, please explain the relevant principles from analog electronics 
and cite specific components/connections in my design.

Example 2: Troubleshooting Assistant

Troubleshooting Workflow:

Students encountering circuit issues documented symptom
Uploaded circuit and measurement results to Claude
Used troubleshooting prompt to get diagnostic suggestions
Tested suggestions methodically and documented results
Reflected on troubleshooting process in lab notebook

Troubleshooting Prompt Example:

I'm troubleshooting a two-stage BJT amplifier circuit that isn't working as expected.

Circuit schematic: [Student uploads circuit diagram]

Expected behavior: Output gain of approximately 65dB, bandwidth from 100Hz to 10kHz

Observed behavior: Very low gain (around 20dB) and significant distortion at input 
signals above 10mV.

Measurements taken:
- DC voltage at collector of Q1: 1.2V (expected 6V)
- DC voltage at base of Q1: 0.7V
- DC voltage at emitter of Q1: 0.1V
- DC voltage at collector of Q2: 8.4V (expected 6V)

Please:
1. Analyze the potential causes of this behavior
2. Suggest 3-5 specific measurements I should take next
3. Identify likely problem areas in the circuit
4. Explain the relationship between my symptoms and possible causes

Example 3: Enhanced Technical Documentation

Documentation Improvement Workflow:

Students drafted technical documentation for labs
Used Claude to analyze clarity and completeness
Received suggestions for improving technical communication
Revised documentation with enhanced descriptions
Reflected on improvements to communication skills

Documentation Enhancement Prompt:

I've written the Analysis section of my lab report for an active bandpass filter 
circuit. Please review my technical writing and provide feedback on:

[Student pastes their technical writing]

1. Clarity of technical explanations
2. Completeness of circuit analysis
3. Appropriate use of technical terminology
4. Logical flow of the analysis
5. Any missing critical information

For each suggestion, explain why it would improve the technical communication.
Focus on helping me communicate circuit behavior more precisely.

Assessment Strategy

Evolving Assessment Approach

Before AI Integration: Traditional Lab Grading

Circuit design correctness (40%)
Circuit performance (25%)
Technical documentation (25%)
Lab procedure completion (10%)

After AI Integration: Process-Focused Assessment

Initial circuit design (20%)
AI feedback evaluation and response (20%)
Revised circuit quality and performance (25%)
Technical documentation with process reflection (25%)
Troubleshooting methodology (10%)

Evaluation Criteria for AI Use

Dr. S developed specific criteria for evaluating appropriate AI use:

Feedback Quality Assessment: Ability to distinguish valid vs. questionable AI feedback
Technical Verification: Use of simulation or calculation to verify AI suggestions
Iterative Improvement: Strategic refinement of designs based on feedback
Documentation Clarity: Quality of explanations about design decisions
Process Reflection: Insights on effective collaboration with AI tools

Potential Outcomes and Considerations

Expected Benefits

More rapid feedback cycles and iterative improvement
Increased student independence in troubleshooting
Improved technical communication skills
More equalized experience for students with varied backgrounds
Instructor time redirected to complex conceptual guidance
Better preparation for industry practice

Potential Challenges

Risk of over-reliance on AI without verification
Possible misdiagnosis of subtle analog circuit issues
Variable student proficiency in prompt engineering
Additional time required to document AI interactions
Occasional misunderstandings of complex analog behavior

Faculty Implementation Considerations

Key Implementation Strategies

Rapid feedback cycles to enable more iteration and learning
Critical evaluation of AI feedback to build stronger circuit intuition
Consistent documentation format for tracking design decisions
Troubleshooting support to reduce student frustration
Reallocation of instructor time to more complex issues

Important Considerations

Technical validation is essential - AI can make confident but incorrect analyses
Variable accuracy with analog circuits - better with digital, less reliable with complex analog
Need for circuit-specific prompt training - general prompts produced generic responses
Value of intentional limitations - restricting AI use in some contexts improved learning
Importance of reflection - metacognitive benefits were as valuable as technical ones

Future Refinement Directions

If implementing such an approach, consider:

Creating a structured prompt library specifically for different circuit topologies
Developing validation protocols for analog circuit feedback
Incorporating industry perspectives on AI-assisted circuit design
Adding comparative exercises contrasting AI and human analysis abilities
Creating advanced troubleshooting scenarios that combine AI and traditional methods

Resources Developed

Circuit-Specific Prompt Library: Templates for different analog circuits
AI Interaction Documentation Template: Standard format for recording feedback
Validation Checklist: Methods to verify AI feedback accuracy
Error Analysis Guide: Common AI misunderstandings in analog circuit analysis
Technical Writing Enhancement Prompts: Templates for documentation improvement

Implementation Advice

For Faculty Considering Similar Integration:

Start with clearly defined circuit types where AI performs more reliably
Teach verification methods alongside AI integration
Use a progressive disclosure approach to gradually reduce prompting guidance
Leverage AI for first-pass feedback but maintain instructor final review
Emphasize process documentation to make learning visible

Technical Considerations:

Claude Pro account provides better image analysis capabilities for circuit diagrams
Circuit diagrams should be clear, high-resolution, and properly labeled
Complex analog circuits require more careful verification
Create a reference library of known circuit types and AI responses
Periodically validate AI responses against instructor expertise

This case study was developed as part of the "Strategies for Integrating Generative AI in Engineering Education" workshop materials in collaboration with Claude-3.7 Sonnet.

Claude in an Electrical Engineering Circuit Analysis Course