---
id: opt-02
title: "The Hungry Streamer - Self-Optimization"
section: "Optimization & Simulation"
difficulty: "advanced"
estimated_time: 30
prerequisites: ["opt-01", "fund-06"]
objectives:
  - Understand the physical feedback loop between power and plasma conductivity
  - Trace the thermal-electrical evolution of a spark
  - Recognize when and why plasma self-optimizes to R_opt_power
  - Identify physical constraints that prevent optimization
tags: ["plasma-physics", "self-optimization", "thermal-dynamics", "feedback"]
---

# The Hungry Streamer - Self-Optimization

One of the most remarkable features of spark plasmas is their ability to **self-adjust** their resistance to maximize power extraction from the coil. This phenomenon, often described by Steve Conner's principle of the "hungry streamer," is a consequence of fundamental plasma physics and thermal dynamics.

## The Physical Feedback Loop

Plasma conductivity changes dynamically with the power it receives, creating a feedback mechanism:

### Step 1: More Power → Joule Heating

```
Heating rate: dT/dt ∝ I²R

Higher current → faster heating
```

The plasma channel experiences resistive heating (Joule heating) from the current flowing through it. The heating rate is proportional to I²R, so higher currents lead to faster temperature rise.

### Step 2: Higher Temperature → Ionization

```
Thermal ionization: fraction ∝ exp(-E_ionization / kT)

Hotter plasma → more free electrons
```

As temperature increases, more air molecules have sufficient thermal energy to ionize. The ionization fraction follows a Boltzmann-like distribution, increasing exponentially with temperature once the thermal energy approaches the ionization energy (~13.6 eV for many atmospheric species).

### Step 3: More Electrons → Higher Conductivity

```
σ = n_e × e × μ_e

where:
  n_e = electron density
  μ_e = electron mobility
  e = elementary charge

σ ∝ n_e ∝ exp(-E_ionization / kT)
```

Electrical conductivity is directly proportional to the free electron density. More ionization means more free charge carriers, which means higher conductivity.

### Step 4: Higher Conductivity → Lower R

```
R = ρL/A = L/(σA)

σ increases → R decreases
```

The resistance of the plasma channel is inversely proportional to conductivity. As the plasma heats up and becomes more conductive, its resistance drops.

### Step 5: Changed R → New Circuit Behavior

```
New R changes Y_spark, power transfer changes:

If R < R_opt_power: reducing R further DECREASES power
If R > R_opt_power: reducing R INCREASES power
```

This is the crucial step. The circuit's power transfer characteristics depend on the load resistance. From our previous lesson, we know that power is maximized at R_opt_power.

### Step 6: Stable Equilibrium at R ≈ R_opt_power

```
When R approaches R_opt_power:
- Small decrease → power decreases → cooling → R rises
- Small increase → power increases → heating → R falls
- Negative feedback stabilizes at R_opt_power
```

**This creates a stable operating point!** The system naturally seeks the resistance value that maximizes power transfer through negative feedback.

## Time Scales

Understanding the time scales involved is critical to predicting when self-optimization occurs.

### Thermal Response: ~0.1-1 ms for Thin Channels

**Heat diffusion time:**
```
τ = d²/(4α)

where:
  d = channel diameter
  α = thermal diffusivity ≈ 2×10⁻⁵ m²/s for air

For d = 100 μm (thin streamer): τ ≈ 0.1 ms
For d = 5 mm (thick leader): τ ≈ 300 ms
```

**Implications:**
- Fast enough to track AC envelope (kHz modulation in QCW/burst mode)
- Too slow to track RF oscillation (hundreds of kHz carrier)
- The plasma "sees" the RMS or average power, not instantaneous RF cycles

### Ionization Response: ~μs to ms

**Recombination time varies with:**
- Electron density (higher density → faster recombination)
- Temperature (higher temperature → slower recombination)
- Gas composition (different species have different rates)

**Typical:** ~1-10 ms for atmospheric pressure air plasmas

### Result: 0.1-10 ms Adjustment Time

The plasma can adjust its resistance on timescales of 0.1-10 ms, allowing it to:
- Track power delivery changes in burst mode or QCW operation
- Respond to voltage variations
- Seek optimal operating conditions dynamically

## Physical Constraints

While the feedback mechanism drives the plasma toward R_opt_power, physical limitations can prevent this optimization:

### Lower Bound: R_min

**Physical limit:**
- Maximum conductivity limited by electron-ion collision frequency
- Even fully ionized plasma has finite conductivity
- Typical: R_min ≈ 1-10 kΩ for hot, dense leader channels

**If R_opt_power < R_min:**
- Plasma stuck at R_min (cannot achieve lower resistance)
- Power transfer is suboptimal
- Spark cannot extract as much power as theoretically possible

### Upper Bound: R_max

**Physical limit:**
- Minimum conductivity of partially ionized gas
- Cool plasma or weak ionization
- Typical: R_max ≈ 100 kΩ to 100 MΩ for cool streamers

**If R_opt_power > R_max:**
- Plasma stuck at R_max (cannot achieve higher resistance)
- Usually not the limiting factor in Tesla coils
- More common with very weak discharges

### Source Limitations

**Insufficient voltage:**
- Spark won't form at all if V_top < V_breakdown
- No optimization possible without a spark

**Insufficient current:**
- Cannot heat plasma enough to reach R_opt_power
- Spark remains in cool streamer regime
- High resistance, low power transfer

**Power supply impedance:**
- If Z_source >> Z_spark, source impedance limits available power
- The "hungry streamer" is starved by a weak source

## When Optimization Fails

Several scenarios prevent the plasma from reaching R_opt_power:

### Source Too Weak

**Scenario:** Available power insufficient to heat plasma

**Result:**
- Spark operates at whatever R it can sustain
- Typically remains at high R (cool streamers)
- Low power transfer, short sparks

### Thermal Time Too Long

**Scenario:** Burst mode with pulse width << thermal time constant

**Example:** 50 μs pulses with τ_thermal = 0.5 ms

**Result:**
- Plasma cannot respond fast enough
- Operates in transient regime
- Does not reach steady-state R_opt_power

### Branching

**Scenario:** Multiple discharge paths from topload

**Result:**
- Available power divides among branches
- No single branch gets enough power to optimize
- Multiple weak streamers rather than one strong leader

## Worked Example: Tracing Optimization Process

**Scenario:** Spark initially forms with R = 200 kΩ (cold streamer). Circuit has R_opt_power = 60 kΩ. Let's trace the thermal-electrical evolution:

### Initial State (t = 0)

```
R = 200 kΩ >> R_opt_power
Power delivered: P_initial (suboptimal, low)
Temperature: T_initial (cool, ~1000 K)
Current: I_initial ≈ V_top / Z_total (low)
```

The spark has just formed. It's essentially a weakly ionized streamer with high resistance.

### Early Phase (0 < t < 1 ms)

```
Current flows → Joule heating: dT/dt = I²R/c_p
R is high → voltage division favorable → some heating occurs
Temperature rises → ionization begins → n_e increases
Conductivity σ ∝ n_e increases → R decreases
R drops toward 150 kΩ
```

**What's happening:**
- Even though R is far from optimal, some power flows
- Joule heating warms the plasma channel
- Thermal ionization begins to create more free electrons
- Resistance starts to drop

### Middle Phase (1 ms < t < 5 ms)

```
R approaches 100 kΩ range
Now closer to R_opt_power → power transfer improves
More power → faster heating → faster ionization
Positive feedback: lower R → more power → lower R
R drops rapidly: 100 kΩ → 80 kΩ → 70 kΩ → 65 kΩ
```

**What's happening:**
- As R approaches R_opt_power, power transfer increases
- Positive feedback accelerates the process
- This is the "hungry" phase - the plasma eagerly draws more power
- Temperature may reach 5000-10000 K (transition to leader)

### Approach to Equilibrium (5 ms < t < 10 ms)

```
R approaches R_opt_power = 60 kΩ
Power maximized at this R

If R < 60 kΩ: power would decrease → cooling → R rises
If R > 60 kΩ: power would increase → heating → R falls

Negative feedback stabilizes around R ≈ 60 kΩ
```

**What's happening:**
- Feedback changes from positive to negative near R_opt_power
- System naturally seeks the stable equilibrium point
- Small perturbations are self-correcting

### Steady State (t > 10 ms)

```
R oscillates around 60 kΩ ± 10%
Temperature stable at equilibrium (~8000-15000 K for leaders)
Power maximized and stable
Spark is "optimized"
```

**What's happening:**
- Plasma has reached thermal and electrical equilibrium
- Continuous power input balances radiative/convective losses
- The spark maintains maximum power extraction

## What If Physical Limits Intervene?

**Example with R_min constraint:**

```
If R_opt_power = 30 kΩ but R_min = 50 kΩ (plasma physics limit):
  Plasma can only reach R = 50 kΩ (not optimal)
  Power is less than theoretical maximum
  Spark is "starved" - wants more current than physics allows
```

This can happen with very hot, dense plasmas where even full ionization cannot achieve the low resistance needed for optimization.

## Steve Conner's Principle

**The "Hungry Streamer" Concept:**

A spark will adjust its resistance to extract maximum power from the source, subject to physical constraints. The plasma behaves as if it is "hungry" for energy and actively optimizes its impedance to feed that hunger.

**Why this matters:**
- Explains why measured spark resistance tends to cluster around R_opt_power
- Justifies using R_opt_power as a design target
- Helps predict spark behavior in different operating modes
- Guides optimization of coil parameters

## Key Takeaways

- Plasma resistance is not fixed - it dynamically adjusts based on power
- **Feedback loop:** Power → Heating → Ionization → Conductivity → R changes → Power changes
- **Stable equilibrium at R ≈ R_opt_power** due to negative feedback
- Time scales: 0.1-10 ms for thermal/ionization response
- Physical constraints: R_min (hot plasma limit), R_max (cool plasma limit), source limitations
- Burst mode with short pulses may not reach equilibrium
- The "hungry streamer" actively seeks maximum power extraction

## Practice

{exercise:opt-ex-02}

**Question 1:** Why does the optimization work? Why doesn't the plasma just pick a random R value and stay there?

**Question 2:** In burst mode (short pulses, <100 μs), thermal time constants are longer than pulse duration. Would you expect the plasma to reach R_opt_power? Why or why not?

**Question 3:** A coil produces sparks with measured R ≈ 20 kΩ, but calculations show R_opt_power = 80 kΩ. What might explain this discrepancy? (Hint: Consider multiple possibilities)

**Question 4:** Sketch the time evolution of R, T, and P for a spark that starts at R = 150 kΩ with R_opt_power = 50 kΩ. Label key phases.

**Question 5:** Why might a branched spark (multiple discharge paths) fail to optimize? Explain in terms of power distribution.

---
**Next Lesson:** [Thévenin Equivalent Method - Extraction](03-thevenin-method.md)