Ultrasonic Radar Sweep Spikes On EZ-B V4

Yes - what you’re describing is very typical of servo movement noise coupling into ultrasonic readings on an EZ-B v4, especially when the sensor is reading at the exact moment the servo starts/stops.

A few important observations from your description:

• Static readings are solid your HC-SR04 wiring and divider look good.
• Spikes happen only during sweep motion this strongly suggests timing/settling, not a bad sensor.
• 0 and 255 readings those are classic "bad sample" values during motion or echo loss.
• Movement Panel twitching the avoidance logic is reacting to those bad samples.

What to try in ARC

1) Increase the sweep settle time indirectly

The Ultrasonic Radar skill doesn’t expose a dedicated "settle delay after each servo step" control in the UI, but you can reduce bad samples by:

• lowering sweep speed
• increasing interval
• increasing step size
• pausing scans while the robot is moving

Since you already tried some of this, the next better approach is to script your own scan timing instead of relying on continuous radar polling.

Best solution: script your own scan loop

Use a script to:

1. move the servo,
2. wait for it to settle,
3. take multiple pings,
4. average or reject outliers,
5. then update the radar/avoidance logic.

This gives you much better control than the built-in continuous sweep.

Example JavaScript approach

If you want, I can help you convert this into a script for your exact ports, but conceptually:

// Example: settle + multiple samples + reject 0/255
function readStablePing(triggerPort, echoPort, samples) {
  var sum = 0;
  var count = 0;

  for (var i = 0; i < samples; i++) {
    var d = Ping.get(triggerPort, echoPort);

    // reject obvious invalid spikes
    if (d > 0 && d < 255) {
      sum += d;
      count++;
    }

    sleep(20);
  }

  if (count == 0) {
    return -1;
  }

  return Math.round(sum / count);
}

Then after each servo move:

Servo.setPosition(d0, 90);
sleep(120); // settle
var distance = readStablePing(d5, d6, 5);
print(distance);

If you want to keep using the Ultrasonic Radar skill for display, you can also pause the skill and update it only when your script calls GetPing()/refreshes based on your configured ports.

2) Invert the servo direction

Yes, the radar can be inverted depending on how the servo horn is mounted.

In Ultrasonic Radar:

1. Open the skill’s kebab menu / menu button in the top-right of the skill.
2. Choose Settings.
3. Look for the Invert servo option in the servo settings section.
4. Enable it and test again.

That should make the sweep direction match the real-world motion.

3) About the spike values

A few notes on the specific numbers:

• 255 usually means the sensor returned no valid echo or the reading timed out.
• 0 usually means an invalid or saturated reading.
• When the servo stops, the horn can cause a small electrical/mechanical transient, and the HC-SR04 can briefly misread.

So the best fix is usually not just "more filtering" but filtering plus better timing.

4) Consider adding a guard band in avoidance logic

Even with filtering, I recommend not triggering avoidance on a single sample.

For example:

• require 2 or 3 consecutive close readings
• ignore any reading of 0 or 255
• only react if the reading is below your threshold for a short confirmation window

That prevents twitch-reverse behavior.

5) Hardware note

You already tried the right hardware mitigations:

• separate 5V supply
• decoupling capacitor
• wire routing
• different servo

That points even more toward sampling during servo transients rather than a bad wiring issue.

My recommendation

If you want the cleanest behavior:

• disable continuous radar scanning
• use a custom JavaScript scan loop
• move servo wait settle take 3-5 valid pings average
• ignore 0/255 outliers
• enable Invert servo in the Ultrasonic Radar settings if needed

That will give you much more stable obstacle detection than the default sweep timing.

If you want, I can write you a drop-in ARC JavaScript scan script for:

• your exact servo port D0
• ultrasonic ports D5/D6
• and a "reject 0/255 + average 5 samples" logic

and I can also show you how to make it update the radar while keeping the Movement Panel avoidance behavior.