Forum Watch: Three Analog and RF Problems That Still Trip Up Engineers

Window Comparator Latching on Zero Volts

Problem: An engineer using an LM339 in a window comparator configuration couldn't figure out why the output LED stayed on with 0V input. The threshold voltages checked out on paper.

The root cause was twofold. First, the pull-up resistor value was too high for the LED drive circuit, limiting current to a level the LM339's open-collector stage couldn't properly sink. Second, and more critically, the input voltage divider network had a subtle imbalance caused by resistor tolerance stack-up. With nominal 1% resistors, the actual thresholds deviated enough that the input fell outside the intended window.

Solution: The thread walkthrough recalculated threshold boundaries accounting for worst-case resistor tolerances. For a ±5V window centered at 2.5V, using 0.1% tolerance resistors brought the threshold spread from ±150mV down to ±30mV. The pull-up resistor was dropped from 10kΩ to 2.2kΩ to ensure the LED current exceeded the LM339's recommended sink current of 6mA.

Results: The circuit functioned correctly after component swapping. The thread provided a useful rule of thumb: for LM339 window comparators, budget at least 2mA through the LED and verify open-collector sink capability against your specific load.

"I wasted two days on this. The datasheet shows the output stage but doesn't explain how resistor tolerance breaks the math on the threshold side."

Cable Velocity Factor and Pulse Delay

Problem: A signal integrity engineer building a 10-meter USB-C extension cable wanted to confirm whether velocity factor affected digital pulse timing or only RF phase relationships.

The confusion stems from mixing phase velocity concepts from AC theory with group velocity behavior in pulse propagation. In non-dispersive media, they are numerically identical, but engineers often don't realize this distinction exists at all.

Solution: The calculation clarified the physics. Velocity factor (VF) directly multiplies propagation delay:

``` t_d = length / (VF × c) ```

For a cable with VF = 0.66 (common for polyethylene dielectric), the 10-meter run introduces 50.5ns of delay. A cable with VF = 0.85 (PTFE) cuts this to 39.2ns. The difference matters for USB 3.2 Gen 2x2, where 10Gbps signaling has bit periods under 100ps.

Results: The engineer settled on low-loss coaxial with VF = 0.85 after running margin analysis. The thread consensus was that velocity factor matters for any digital link exceeding 5% of the bit period. For short runs under 1 meter at USB 2.0 speeds, VF tolerance falls below measurement noise.

Flame Rectification in Gas Ignition Systems

Problem: A thread on gas water heater ignition circuits revealed widespread confusion about how flame rectification sensors actually work versus how people think they work.

The common misconception treats flame rectification as simple conductivity through hot gas. The actual mechanism is more interesting: the flame behaves as a half-wave rectifier due to the asymmetric electrode geometry and DC bias applied through the ignition transformer secondary.

Solution: The discussion walked through the ionization physics. A gas flame produces positive ions and electrons. When the sensor electrode sits at a DC potential relative to the ground electrode (typically 60-120V AC with a DC offset), ion migration creates a rectified current that flows only during one half-cycle. The safety circuit monitors this rectified current; absence of current within a fixed time window triggers a lockout.

Results: Thread contributors documented practical failure modes. Carbon buildup on the sensor electrode increases resistance, dropping the rectified current below the threshold (typically 1-2µA). Ceramic cracking allows moisture ingress. One contributor measured electrode resistance increasing from 80kΩ to over 2MΩ after 18 months of service in hard water conditions.

"I've seen three house fires started by people bypassing the flame sensor. The rectification circuit isn't optional safety theater."

---

##

M4S TAKE

My take: AI claims need scrutiny. The useful implementations reduce cycle time or defect rates in measurable ways. Vague promises about 'optimization' without specific metrics are usually marketing.

Simon McLoughlin