Close

Log #2: The Software Pivot - Moving to "Automotive Scale"

A project log for Turning Cheap Audio Chips into Industrial TDR

Turn a $3 USB soundcard into a cm-accurate industrial TDR. We use DSP and formal logic to find automotive wire faults for under $5.

lszl-szkeLászló SZŐKE 10 hours ago0 Comments

The Hardware Wall (and How We Broke Through It)

Now that the V1 Interface is soldered and we're safely using the 13 kΩ resistor (Automotive Bridge) to protect the soundcard, we hit a massive software wall.

When you push an audio signal through a 13 kΩ resistor and look for the reflection from a wiring harness, the returning signal amplitude (Magnitude) drops into the $10^{-5}$ range. That’s deep below the noise floor of a standard Realtek soundcard. The previous software version treated this return signal as a hardware error. We had to completely pivot our DSP architecture.

MetaSpace v8.0 Architecture: The "Statistical Shield"

To solve the noise problem, we restructured the MetaSpace engine into six independent, professional-grade modules. The most critical upgrade was creating what we call the "Statistical Shield":

  1. Massive Integration Windows: We ditched standard audio processing. The DSP engine now uses a vector-averaged Lock-in amplifier with a massive 16,384 sample window to integrate the signal out of the noise floor.
  2. Atomic-Sync: To combat the inherent jitter of Windows/Realtek drivers, the measurement now starts with a synchronous chirp. This "Atomic-Sync" cancels out jitter with fractional-sample precision, allowing the software to know exactly when the physical response begins.
  3. Deterministic Stability: Even on consumer soundcards, we achieved a measurement stability of 0.0001 radians (~0.005 degrees).

Real-World Validation: It Works!

We empirically calibrated the V8 engine using a 60 cm precision cable, and the results are incredibly promising:

What's Next?

The core http://localhost:8080 engine is stable. The next step is macroscopic field testing: taking the hardware out to the garage and testing it on actual 3 to 6-meter automotive wire harnesses to measure larger phase jumps (0.3 - 1.5 rad).

Discussions