Deceleration on Impacts

How does unity handle the deceleratikon of rigidbodies on impacts? I did some testing.
So far all ive done is crank the fixed timestep to 0.01, and i fired a gravityless-ball v ery slowly at a kinematic wall at a velocity of Vector3.forward (so the magnitude of velocity is 1.0)

, and tracked the magnitude of its velocity during the impact:

in these debug messages, the leftmost number is the magnitude of velocity, and the right one is the current frame number (number of fixedupdate calls since start of running). note also i added a conditional to not display message when the velocity was >=1 or = 0

As can be seen here, it takes about 7 frames to lose most of its speed, but then gets stuck at 8.642675E-06 until frame 393 (15 frames)

I need some help making sense of this. How is unity doing this gradual deceleration, is there a deterministic pattern to it ? And why does it get stuck at some very low value for 15 frames (0.15 seconds) before finally coming to rest?

i’ve been doing some testing, i made this simple test harness script

using UnityEngine;
using System.Collections;
using System.Diagnostics;

public class VelocityMonitor : MonoBehaviour
{
    int i = 0;
    public float j;
    float mag =10f;
    Stopwatch timer = new Stopwatch();
   
    void Start()
    {
        rigidbody.velocity = Vector3.forward*mag;
    }
    void FixedUpdate()
    {
        j = rigidbody.velocity.magnitude;
        if (j < mag && j > 0.0001f)
        {
            timer.Start();
            print(j + "  num " + i);
        }
        else if (timer.IsRunning)
        {
            timer.Stop();
            print("Total Time Taken to stop " + timer.Elapsed.TotalSeconds);
        }
        i++;
    }
}

I also cranked up the fixedTimestep to 0.005 (200 ticks per second!)
with this value, i was able to get readouts for speeds up to about 8 M/s. The faster the ball is going, the less time it takes to decelerate to a stop, it seems. I maybe be wrong, but that doesn’t seem to quite gel with my intuition about physics,

at about 10 m/s it just goes from fullspeed to zero in a single frame, and i have to crank up the timestep some more to see anything in there