What is it

Link to Production Application

https://floating-savannah-01315.herokuapp.com/app/landing.html

Running the Application Locally

1. Install LTspice.

   • Under Tools -> Control Panel -> Netlist Options, check the Convert 'µ' to 'u' option - this forces LTspice to encode its output files as UTF-8. This application is only compatible with UTF-8-encoded files.

2. Install Python 3. This application was tested on Python 3.9.x.

3. Install MongoDB.

4. Clone this project to your local workspace. Then, run pip install -r requirements.txt under the project root.

5. Start your local development MongoDB server using the mongod executable.

6. Run server.py to start your local development server.

7. Go to localhost:5000/app/landing.html for the landing page of the web app. To start using the web app, you must upload the following LTspice files:

   • .cir netlist file
   • (optional) .asc schematic file, if you wish to render the circuit schematic
   • (optional) .log operating point analysis log file - you won't need this if your circuit is already in small-signal form

Architecture

This application is divided into the following modules:

circuit_parser.py

Parses LTspice netlists and operating point analysis log files to create small-signal circuits - these are represented as NetworkX multigraphs.

Currently, the parser supports only a small subset of LTspice component types and syntaxes. The module can be extended by subclassing Component, and overridng parsing methods for new component types.

Note that the application only uses small-signal circuits as an intermediary step towards computing the signal-flow graph (SFG). For debugging and verification purposes, small-signal circuit multigraphs produced by this module can be visualized in one of two ways:

1. Render them on-screen using matplotlib (for more details, consult NetworkX library documentation)
2. Print the small-signal circuit to a netlist file using the Circuit.netlist property

dpi.py

Performs driving-point impedance analysis on a small-signal circuit, and outputs a signal-flow graph (SFG). The SFG class parses the small signal model generated by circuit_parser.py represented in NetworkX multigraph and store the result in another multigraph object called SFG.graph. The dpi analysis follows the procedure: it first render the impedence for each node in the circuit and then generate short circuit current and voltage nodes, which is called the driving pointer impediance analysis.

mason.py

Applies Mason's gain formula to calculate the transfer function and loop gain of an SFG (the SFG is represented as a NetworkX digraph). The module uses NetworkX's implementation of Johnson's algorithm to find all simple cycles within the SFG (one of the main steps in the formula), and is therefore relatively efficient.

/ltspice2svg

A thin wrapper around the ltspice2svg package. It converts .asc schematic files to a web-friendly SVG format that can be displayed in the browser. The wrapper is needed because the original package can only be run as main, and won't work correctly when imported. The wrapper solves this issue by invoking it as a subprocess.

db.py

Performs operations against a mongodb instance including: * creating and storing circuit documents * serializing and de-serializing compiled transfer functions (these are used to compute bode plot point. They are expensive to compile so are cached in the database). * modifying circuit documents

This module should probably be re-factored because it embodies too much un-related logic. For example:

• the create circuit method also invokes the circuit parser, when it's better to pass in an already parsed circuit
• the module compiles symbolic transfer functions into faster numerical ones internally. This logic should probably be separated out.

When deployed to Heroku, if the environment variable DB_URI is specified, the production MongoDB instance URI in the cloud.

server.py

Runs a development server on port 5000. * Front-end web pages are served from the static ./public directory. * An exhaustive list of API routes are available in api_specification.html. Some example functionalities provided are: * Update the component values in a circuit * Create a new circuit * Plot the gain and phase of a circuit's transfer function * Find the transfer function expression of a circuit

Note that when deployed to Heroku, the application uses Procfile to launch a production-quality waitress server instead of flasks's built-in development server.

How to deploy

This application is currently deployed to Heroku. The deployment process itself is quite straightforward, you simply have to reference Heroku's Getting Started with Python guide. Note that there is an additional step of setting the DB_URI environment variable in the console to point to your production mongodb instance.

The database instance is hosted on mongodb atlas cloud. To view its contents, obtain the connection string for your cluster from https://account.mongodb.com/account/login. Then, install MongoDB compass and connect to the cluster to view its contents.