NI Multisim 14.1 Automation with MCP

This project connects NI Multisim 14.1 with an MCP server so an AI agent can work with Multisim directly.

The setup uses multisim-mcp and Antigravity to open circuits, edit components, run simulations, read measurements, perform parameter sweeps, and generate reports.

The main goal is to make working with Multisim easier by allowing many circuit tasks to be done using normal language instead of doing everything manually inside Multisim.

What This Project Can Do

With the MCP setup you can:

• Open and inspect Multisim circuits
• Create and modify circuits
• Change component values
• Read circuit and net information
• Run DC operating point simulations
• Run AC frequency sweeps
• Run transient simulations
• Read voltage and current measurements
• Use virtual instruments
• Run parameter sweeps
• Export simulation data
• Generate experiment reports

The Multisim COM interface is used underneath the MCP server to communicate with Multisim.

System Setup

The setup was tested on the following configuration:

Component	Version / Setup
Operating System	Windows 11 64-bit
NI Multisim	14.1 32-bit
Python	3.14 32-bit
MCP Server	multisim-mcp 1.1.0
AI Client	Antigravity

The important part here is the 32-bit setup. Multisim 14.1 uses 32-bit COM components, so the Python environment used to communicate with Multisim also needs to be 32-bit.

Installation

1. Check the Python Installations

First check which Python versions are installed:

py --list

Then check that the Python version you want to use is actually 32-bit:

py -3.14-32 -c "import sys; print(sys.executable); print('Is 64-bit:', sys.maxsize > 2**32)"

The Python installation used in this setup was:

C:\Program Files (x86)\Python314-32\python.exe

The important thing is that the output confirms that Python is running as 32-bit.

2. Install multisim-mcp

Install version 1.1.0 using the 32-bit Python installation:

& "C:\Program Files (x86)\Python314-32\python.exe" -m pip install "multisim-mcp==1.1.0"

Because Python is installed under Program Files (x86), Windows may not allow normal writing to that directory.

In that case, pip can place the user-level scripts under:

C:\Users\<User>\AppData\Roaming\Python\Python314-32\Scripts

This is where the multisim-mcp.exe executable can be found.

3. Check the Multisim Connection

Before configuring Antigravity, it is a good idea to check whether the MCP setup can actually communicate with Multisim.

Run:

& "$env:APPDATA\Python\Python314-32\Scripts\multisim-mcp.exe" --json doctor --connect

The diagnostic checks the Python architecture, Multisim COM registration, the Multisim application, and the simulation engine.

The verified setup showed:

Python Architecture: 32-bit
COM Registration: Found
Live Connection: Connected
Simulation Engine: Ready

The Multisim COM object detected in the setup was:

MultisimInterface.MultisimApp

with CLSID:

{D9CBB7A1-6AD6-438B-AD1B-42FB2DB00CAE}

The Multisim installation was detected at:

C:\Program Files (x86)\National Instruments\Circuit Design Suite 14.1\Multisim.exe

The diagnostic also reported:

automation_ready: true

So the COM connection and simulation engine were working correctly.

4. Generate the MCP Configuration

You can generate the configuration using:

& "$env:APPDATA\Python\Python314-32\Scripts\multisim-mcp.exe" config --client generic --python "C:\Program Files (x86)\Python314-32\python.exe"

The generated configuration looks like this:

{
  "command": "C:\\Program Files (x86)\\Python314-32\\python.exe",
  "args": [
    "-m",
    "multisim_mcp.server"
  ]
}

This tells the MCP client to start the Multisim MCP server using the 32-bit Python installation.

5. Add It to Antigravity

The MCP configuration can be added to the Antigravity configuration file.

Possible configuration locations are:

~/.gemini/config/mcp_config.json

or:

~/.gemini/antigravity/mcp_config.json

Add the Multisim server under mcpServers:

{
  "mcpServers": {
    "multisim": {
      "command": "C:\\Program Files (x86)\\Python314-32\\python.exe",
      "args": [
        "-m",
        "multisim_mcp.server"
      ]
    }
  }
}

After this is configured, Antigravity can start the MCP server when it needs to use the Multisim tools.

How the Setup Works

You do not have to manually start the MCP server every time.

Once the server is configured in Antigravity:

1. Antigravity starts the MCP server when it needs it
2. The MCP server connects to Multisim
3. The AI agent can call the available Multisim tools
4. Multisim performs the requested operation
5. Results are returned to the AI agent

There is no need to keep a separate terminal window running multisim-mcp serve.

The configuration can also be available across different Antigravity workspaces when it is placed in the global configuration.

Using Multisim with Natural Language

Once everything is configured, you can talk to the AI agent normally.

Open and Inspect a Circuit

For example:

Connect to Multisim and open the circuit file at C:\Circuits\AudioAmp.ms14

You can also ask:

List all components, nets, and input/output pins in the active schematic.

To change component values:

Change resistor R1 to 10k ohms and capacitor C1 to 100nF.

These commands can be translated into the corresponding Multisim operations through MCP.

Run Simulations

You can ask for a DC operating point:

Run a DC operating point analysis and show the nodal voltages.

For AC analysis:

Run an AC frequency sweep from 10 Hz to 100 kHz on net Vout and report the -3dB cutoff frequency.

For transient analysis:

Run a 10ms transient simulation with a 1us time step and report peak voltages.

You can also read virtual instrument measurements:

Read the virtual multimeter value on net 3.

The MCP server provides tools for these simulation operations.

Parameter Sweeps

The setup can also be used for testing different component values automatically.

For example:

Run a parameter sweep for resistor R1 across [1k, 4.7k, 10k] and evaluate rise time.

You can also ask it to create a report:

Export the formal experiment report with waveform CSV data.

This is useful when you want to compare several circuit configurations without changing each value manually.

Verified Test Circuit

To test the complete setup, a simple 1 Hz blinking LED circuit was simulated through Multisim.

The test was useful because it checked the whole process:

AI Request
    ↓
MCP
    ↓
Multisim
    ↓
SPICE Simulation
    ↓
Measurements
    ↓
Results

Circuit Netlist

The test circuit used the following SPICE netlist:

* LED Blinking Circuit (1 Hz)
V1 1 0 PULSE(0 5 0 1m 1m 0.5 1.0)
R1 1 2 330
D1 2 0 DLED
.model DLED D(Is=1e-22 Rs=5 N=1.8 Cjo=20p)
.end

The pulse source switches between 0 V and 5 V.

The resistor limits the LED current.

Simulation Command

The transient simulation was run with:

tran 10m 2.0

The simulation produced the expected LED waveform.

The measured results were:

• Pulse voltage: 0 V to 5 V
• ON time: 500 ms
• OFF time: 500 ms
• LED anode voltage: about 2.18 V
• Peak LED current: about 8.55 mA
• Waveform data: saved to CSV

The complete test confirmed that Multisim was able to run the circuit and return simulation data through the automation setup.

MCP Tools

The multisim-mcp setup provides around 55 tools for working with Multisim.

The tools are grouped by purpose.

Connection

connect
disconnect
runtime_status
circuit_info

These tools are used to connect to Multisim and check its current state.

Circuit Management

new_circuit
open_circuit
save_circuit
get_circuit_image

These are used to create, open, save, and inspect circuits.

Components and Netlists

report_netlist
report_bom
enum_components
get_rlc_value
set_rlc_value

These tools make it possible to inspect components and change values programmatically.

Simulation

run_dc_operating_point
run_ac_sweep
run_ac_single_frequency
run_transient
stop_simulation
run_spice_netlist

These cover the main simulation types used in the project.

Virtual Instruments

read_virtual_multimeter
analyze_bode_response
analyze_logic_signals

These tools are useful for reading measurements and analyzing simulation results.

Experiments and Reports

run_circuit_experiment
run_verified_circuit_experiment
plan_experiment_sweep
run_experiment_sweep
generate_report
export_formal_experiment_report

These tools are useful when running complete experiments and generating results automatically.

Analog Laboratory Experiments

The project also contains a set of analog electronics experiments.

Each experiment has:

• A native .ms14 Multisim schematic
• A Markdown experiment report
• Simulation results
• Component values
• Theoretical calculations
• Verification results

The project currently documents the following experiments.

1. Active First-Order Low-Pass Filter

Files:

experiments/RC_Low_Pass_Filter.ms14
experiments/RC_FILTER_EXPERIMENT_REPORT.md

Main components:

R1 = 1 kΩ
R2 = 1 kΩ
C1 = 100 nF
Op-Amp = OPAMP_5T_VIRTUAL
Supply = ±12 V

The theoretical cutoff frequency is:

fc = 1 / (2πRC)

For the given values:

fc ≈ 1591.55 Hz

So the expected cutoff frequency is approximately:

1.59 kHz

The passband gain is:

Av0 = -R2 / R1

which gives:

Av0 = -1

or approximately:

0 dB

The Multisim circuit and report are included in the project.

2. Variable Regulated DC Power Supply

Files:

experiments/variable_power_supply_bridge_rectifier.ms14
experiments/BRIDGE_RECTIFIER_EXPERIMENT_REPORT.md

Main components include:

10:1 Step-Down Transformer
3N246 Full-Wave Diode Bridge
2200 µF Smoothing Capacitor
LM317K Voltage Regulator
10 kΩ Variable Potentiometer
33 Ω Load Resistor

The expected output range is approximately:

1.75 V to 13 V

The circuit includes the transformer, bridge rectifier, filter capacitor, adjustable regulator, and load.

3. RC Phase Shift Audio Oscillator

Files:

experiments/rc_phase_shift_oscillator.ms14
experiments/RC_PHASE_SHIFT_OSCILLATOR_EXPERIMENT_REPORT.md

The circuit uses:

LM741 Op-Amp
±15 V supply
3-stage RC network
C1 = C2 = C3 = 0.01 µF
R1 = R2 = R3 = 1.5 kΩ
R4 = 15 kΩ
R5 = 1 MΩ potentiometer

The theoretical oscillation frequency is:

f0 = 1 / (2πRC√6)

For the given component values:

f0 ≈ 4331.8 Hz

or approximately:

4.33 kHz

The required loop gain is approximately:

|Av| ≥ 29

The project documentation gives a feedback resistance requirement of approximately:

R5 ≥ 435 kΩ

for the stated condition.

4. Inverting Op-Amp

Files:

experiments/opamp_inverting_amplifier.ms14
experiments/OPAMP_INVERTING_AMPLIFIER_EXPERIMENT_REPORT.md

Main components:

LM741
±12 V supply
R1 = 10 kΩ
R2 = 100 kΩ
Input = 2 Vpk at 1 kHz

The theoretical gain is:

Av = -R2 / R1

Therefore:

Av = -10

This corresponds to approximately:

+20 dB

The output also has a 180° phase reversal because the amplifier is configured as an inverting amplifier.

The negative sign in the gain represents this phase inversion.

5. BJT Common-Emitter Amplifier

The project also includes a BJT common-emitter amplifier experiment.

File:

experiments/bjt_ce_amplifier.ms14

Report:

experiments/BJT_CE_AMPLIFIER_EXPERIMENT_REPORT.md

The documented circuit uses a:

2N2222 transistor

The reported voltage gain is approximately:

Av ≈ -18.2

The experiment also includes DC bias validation.

Experiment Summary

Experiment	Multisim File	Report	Main Result
Active Low-Pass Filter	RC_Low_Pass_Filter.ms14	RC_FILTER_EXPERIMENT_REPORT.md	Cutoff ≈ 1.59 kHz
Variable DC Power Supply	variable_power_supply_bridge_rectifier.ms14	BRIDGE_RECTIFIER_EXPERIMENT_REPORT.md	1.75 V to 13 V
RC Phase Shift Oscillator	rc_phase_shift_oscillator.ms14	RC_PHASE_SHIFT_OSCILLATOR_EXPERIMENT_REPORT.md	Frequency ≈ 4.33 kHz
Inverting Op-Amp	opamp_inverting_amplifier.ms14	OPAMP_INVERTING_AMPLIFIER_EXPERIMENT_REPORT.md	Gain = -10
BJT CE Amplifier	bjt_ce_amplifier.ms14	BJT_CE_AMPLIFIER_EXPERIMENT_REPORT.md	Gain ≈ -18.2

The project keeps the native Multisim .ms14 files and Markdown reports. It does not keep separate .cir files.

Project Structure

ni-multisim-automation/
│
├── README.md
│
├── assets/
│   ├── rc_active_low_pass_filter_schematic.png
│   ├── rc_active_low_pass_filter.png
│   ├── variable_power_supply_bridge_rectifier.png
│   ├── rc_phase_shift_oscillator.png
│   └── opamp_inverting_amplifier.png
│
├── experiments/
│   ├── RC_Low_Pass_Filter.ms14
│   ├── RC_FILTER_EXPERIMENT_REPORT.md
│   │
│   ├── variable_power_supply_bridge_rectifier.ms14
│   ├── BRIDGE_RECTIFIER_EXPERIMENT_REPORT.md
│   │
│   ├── rc_phase_shift_oscillator.ms14
│   ├── RC_PHASE_SHIFT_OSCILLATOR_EXPERIMENT_REPORT.md
│   │
│   ├── opamp_inverting_amplifier.ms14
│   ├── OPAMP_INVERTING_AMPLIFIER_EXPERIMENT_REPORT.md
│   │
│   ├── bjt_ce_amplifier.ms14
│   └── BJT_CE_AMPLIFIER_EXPERIMENT_REPORT.md
│
└── ANALOG_LAB_EXPERIMENTS_MANUAL.md

GitHub Setup

If you want to upload the project to GitHub, first open PowerShell in the project directory:

cd "c:\ni multisim sutomation"

Initialize Git:

git init

Add the README and gitignore:

git add README.md .gitignore

Create the first commit:

git commit -m "Initial commit: NI Multisim 14.1 MCP automation setup and documentation"

Then set the main branch:

git branch -M main

Add your GitHub repository:

git remote add origin https://github.com/<YOUR-USERNAME>/<YOUR-REPO-NAME>.git

Finally push the project:

git push -u origin main

Replace the username and repository name with your actual GitHub details.

Final Notes

This project is mainly about connecting NI Multisim 14.1 with an AI agent through MCP.

The useful part is that circuit work that normally requires a lot of manual interaction can be controlled through commands such as:

Open this circuit
Change R1 to 10k
Run an AC sweep
Measure Vout
Find the cutoff frequency
Run the experiment
Generate the report

The current setup has already been tested with Multisim's COM interface and a working transient LED simulation. It also includes several analog laboratory circuits that can be opened and simulated directly in Multisim.