Manufacturing Test Analytics

With TofuPilot, monitor your tests first-pass yield (FPY), and capability index (Cpk) at the test phase level.

Overview

When you deploy your test script, it runs many times, and issues arise. Instruments may not connect, limits might reject good parts, or some steps take longer than expected.

The best way to troubleshoot your test scripts is with a database of test logs and analytics, but building that adds more work when you're already managing tests, hardware, and production timelines.

We faced these issues ourselves, building scrappy analytics while scaling drone manufacturing. We saw many other teams doing the same. So, we built TofuPilot a plug-and-play analytics platform for hardware test engineers, on solid web infrastructure, available on cloud or on-premise, and with step-level test analytics.

It’s OpenHTF-compatible and integrates in 5 minutes. And let us know what you think on Discord!

TofuPilot integration

To integrate TofuPilot, install the open-source client:

pip install tofupilot

Add it to your test like this:

import openhtf as htf
from openhtf.util import units
from tofupilot.openhtf import TofuPilot

def phase_one(test):
    return htf.PhaseResult.CONTINUE

@htf.measures(
    htf.Measurement("temperature")
    .in_range(0, 100)
    .with_units(units.DEGREE_CELSIUS)
)
def phase_temperature(test):
    test.measurements.temperature = 25

def main():
    test = htf.Test(phase_one, phase_temperature)
    with TofuPilot(test):                          # just works™️
        test.execute(lambda: "PCB001")

if __name__ == "__main__":
    main()

Each time you run a test, it's automatically uploaded to your workspace.

First-pass yield (FPY)

On the test procedure page, you'll find detailed statistics on execution performance. The most commonly used metric in industrial test monitoring is First-pass yield (FPY). It measures the percentage of parts that pass without rework and helps assess the efficiency of the process.

In TofuPilot, the first-pass yield is calculated in real time for the complete test and for each test phase, so you can quickly see which step impacts the overall yield the most.

A low First-pass yield (FPY) means many parts fail on the first try, often due to poorly calibrated limits, procedural errors, equipment defects, or process variations.

Process capability index (Cpk)

The Process Capability Index (Cpk) measures how well a test stays within its limits. It applies to test steps with numerical measurements and at least one lower or upper limit.

In TofuPilot, Cpk is calculated in real-time for each test step, showing you how capable your test is.

Understanding Cpk

A low Cpk means your test is too variable or not centered, leading to inconsistent results.

• Cpk > 1.33: test process is highly capable.
• Cpk = 1: test process is marginally capable.
• Cpk < 1: test process is not capable; improvements are needed.

A low Cpk often means your process has too much variation: limits are too strict or not centered, or test values are too close to the limits, causing inconsistent or out-of-spec results.

How-to fix low Cpk

1. Identify Test Steps: Use TofuPilot's step selection to find test steps with low Cpk that are dragging down the average.
2. Check Control Charts: Review historical values to see if it's a sudden drop (indicating a quality issue) or if limits were always too strict or poorly centered.
3. Use Filters: Apply filters like revision, batch number, and date in TofuPilot to pinpoint the root cause of low Cpk.

Cpk Calculation

You can compute Cpk by dividing the smaller of the differences between the test mean and the specification limits (low and high) by three times the test standard deviation.

Cpk = min[(USL - μ) / (3σ), (μ - LSL) / (3σ)]

• USL = Upper Specification Limit
• LSL = Lower Specification Limit
• μ = Process Mean
• σ = Standard Deviation