Handheld Signal Analyzer

• Analog front end

  Compensated 1:1/1:10 divider into a programmable-gain amplifier, then a 400 kHz anti-alias filter feeding the 12-bit ADC. Clamp diodes and a series PTC survive ±50 V abuse, which is the failure mode every cheap USB scope shares.

• Acquisition

  RP2040 PIO state machines capture the eight digital channels at 100 MS/s into a ring buffer over DMA, so the CPU only handles triggering and rendering. Analog capture runs at 1 MS/s per channel with hardware pre-trigger.

• Enclosure and feel

  Two-piece 6061 shell, chamfered edges, blind-tapped M2 posts so no fastener is visible from the front, and a captive 2.8" IPS display bonded behind a scratch-resistant window. Wall thickness dropped to 1.6 mm after an FEA pass at 1.9 mm showed excess margin.

1. Prototype 1

   Square waves showed 22% overshoot that the bench scope did not.

   The divider compensation capacitor was mis-valued against the actual input capacitance. Swapped in a trimmer, tuned against a 1 kHz reference, then read back the final value and fitted a fixed part.

2. Prototype 1

   Digital capture dropped samples above 40 MS/s.

   DMA was contending with the display driver on the same bus. Moved framebuffer writes to core 1 with a dedicated DMA channel and staggered the priorities.

3. Prototype 2

   Baseline noise doubled once the board was installed in the aluminum shell.

   The chassis was floating and coupling into the analog island. Bonded the shell to signal ground at one screw boss near the input jack; noise floor returned to 1.4 mV RMS.

4. Prototype 2

   Display window creaked and flexed under thumb pressure.

   Bonded the window on a 0.4 mm recessed ledge with VHB rather than an unsupported edge glue joint, which also improved the dust seal.

Debugging away from a bench meant guessing. I wanted an instrument I could actually carry: two analog channels plus eight digital lines, real attenuation and protection on the input, and an enclosure that felt like a product rather than a project box.