Coffee Grind Size: The Particle Distribution Physics of Extraction
The Hook
You buy a $300 grinder. Your coffee tastes better. You buy a $3,000 grinder. It tastes different—cleaner, sweeter, more transparent. The difference isn’t “grind size”—it’s particle distribution. Cheap grinders produce bimodal dust+boulders. The EK43 produces a unimodal peak at 600μm. The physics of particle distribution explains why.
The Thesis
Grind size is a lie. Grind distribution is the truth. Every grinder produces a range of particle sizes. The shape of that distribution—bimodal (peaks at fines + boulders) vs. unimodal (single peak)—determines extraction dynamics more than the median size. The EK43’s dominance in competitions comes from its unimodal distribution, not its “grind size.”
The Science
Particle Size Distribution (PSD)
Every grinder produces a distribution, not a single size.
| Grinder Type | Distribution | Peaks | Fines % | Boulders % |
|---|---|---|---|---|
| Blade | Chaos | None | 30%+ | 30%+ |
| Cheap burr (conical) | Bimodal | ~100μm, ~800μm | 15-20% | 15-20% |
| Mid burr (flat) | Broad unimodal | ~400-600μm | 8-12% | 8-12% |
| EK43 (flat, 98mm) | Sharp unimodal | ~600μm | 3-5% | 3-5% |
| EG-1 (flat, 83mm) | Sharp unimodal | ~550μm | 4-6% | 4-6% |
Why Distribution Matters: Extraction Physics
Extraction = f(surface area, time, temperature)
| Particle Size | Surface Area | Extraction Rate | Risk |
|---|---|---|---|
| Fines (<100μm) | Huge | Instant | Over-extraction (bitter, astringent) |
| Target (400-800μm) | Optimal | Controlled | Balanced |
| Boulders (>1000μm) | Low | Slow | Under-extraction (sour, saline) |
Bimodal problem: Fines over-extract in seconds; boulders under-extract in minutes. Result: simultaneous over- and under-extraction = muddy flavor.
Unimodal advantage: Narrow size range = uniform extraction = clean flavor separation.
Fines Migration
Fines migration (Gagne et al., 2017, J. Food Eng.):
- Fines migrate downward during brew (percolation)
- Accumulate at filter → clog → channeling
- Channeling = water bypasses coffee = under-extraction
Unimodal = fewer fines = less migration = cleaner cup.
The EK43 Geometry
Why EK43 wins:
- 98mm flat burrs → large cutting surface → fewer fractures
- Parallel burr faces → consistent gap → uniform fracture
- Low RPM (1400) → less heat, less shattering
- Cast steel burrs → sharp edges stay sharp
Result: PSD peak at 600μm, σ (std dev) ≈ 0.3 (log-normal). Cheap grinders: σ ≈ 0.8-1.2.
The Protocol: Measure Your Distribution
DIY Sieve Analysis ($50)
Equipment:
- 5 sieve stack (200, 400, 600, 800, 1000μm) — $40
- Digital scale (0.01g) — $15
- 10g coffee sample
Procedure:
- Weigh 10g coffee
- Sieve 2 minutes (standard shake)
- Weigh each fraction
- Plot cumulative distribution
Target Distribution (Percolation)
| Sieve (μm) | Target % Retained | Cumulative % Passing |
|---|---|---|
| >1000 | <5% | >95% |
| 800-1000 | 10-15% | 80-85% |
| 600-800 | 25-35% | 50-60% |
| 400-600 | 35-45% | 15-25% |
| 200-400 | 15-20% | 5-10% |
| <200 | <5% | <5% |
Peak at 600μm = 35-45% in 400-600μm bin.
Adjusting Your Grinder
| Symptom | Distribution Issue | Fix |
|---|---|---|
| Bitter + sour | Bimodal (fines + boulders) | Upgrade burrs / align |
| Hollow, thin | Too coarse (peak >800μm) | Finer |
| Bitter, astringent | Too fine (peak <400μm) | Coarser |
| Channeling | Excess fines (<200μm) | Reduce fines (align, upgrade) |
The Counterpoint
- Taste > numbers: Some bimodal grinders taste great (alignment, burr geometry)
- Brew method matters: Espresso needs more fines (pressure); filter needs fewer
- Coffee matters: Light roasts = more fines; dark roasts = fewer
- Cost: EK43 = $3,500; Niche Zero = $650 (unimodal enough)
Practical Takeaway
This week: Borrow sieves. Test your grinder. If bimodal, align burrs (shim, realign). If still bimodal, the grinder is the limit. The distribution is the flavor.
Sources: Gagne et al. (2017) J. Food Eng.; Uman et al. (2017) J. Food Eng.; Perger (2016) Barista Hustle; Mahlkönig EK43 technical specs; Barista Hustle “Grind Size” series (2016-2023); Coffee Technicians Guild “Grinder Alignment” protocols.