Your H14 factory test certificate reads "99.995% efficiency." But that's the whole-filter average. If a 5 mm² spot has a pinhole, local penetration could be 5 to 250× the overall — and the certificate won't show it.
Why "Overall Efficiency" Is Not Enough
HEPA/ULPA filters are thousands of glass fiber layers cross-laid. Any tiny defect during manufacturing — broken binder, fiber displacement, a foreign object trapped in the seal — can create a local pinhole.
Overall efficiency tests measure the whole filter's average penetration. A pinhole might push its own local penetration 10–100×, but averaged over the full face, the filter still reads 99.99% instead of 99.995%. Certificate still passes.
The problem: air takes the path of least resistance. A single 5 mm² pinhole may leak more particles than the rest of the filter combined. Customer installs a "passed" filter, cleanliness doesn't meet spec — the usual cause is an undetected local leak.
The Fix: PAO Scan Test
PAO (Poly-Alpha Olefin) is a synthetic oil-mist aerosol with a median particle size of 0.3 μm — conveniently near HEPA MPPS. Inject it upstream, probe-scan the entire downstream face, find any point where local penetration spikes.
EN 1822-4 / ISO 29463 / IEST-RP-CC034 all require per-unit scan testing for H14 and above.
Why PAO instead of real dust?
- ▸Uniform, stable particle size (real dust is too variable)
- ▸Refractive index suits optical detection
- ▸Doesn't clog the filter (filter still usable afterward)
- ▸Non-toxic (DOP retired due to carcinogenicity)
The Three-Step Flow
Biểu đồ 1: Ba bước thử PAO Scan
Phát aerosol thượng lưu → đo nền → quét mặt hạ lưu
Phát aerosol PAO
Laskin hoặc máy nhiệt, MMD ~0.3 μm
Đo nền C₁
OPC hoặc photometer
Quét C₂
Đầu dò quét mặt hạ lưu kiểu S
EN 1822-4 / IEST-RP-CC034 yêu cầu scan test cho H14+.
Upstream — aerosol generation
A Laskin nozzle or thermal generator atomizes PAO. Concentration held at 10–100 μg/L; upstream challenge concentration uniform across the filter face.
Baseline — upstream concentration C₁
An optical particle counter (OPC) or photometer reads upstream concentration as the 100% reference.
Downstream — point-by-point scan
Probe held 20–30 mm from the downstream face, moved in an S-pattern across the whole filter at ≤ 50 mm/s. Downstream concentration C₂ logged in sync.
Local penetration at each point:
P (%) = (C₂ / C₁) × 100
Pass / Fail Criteria
Biểu đồ 2: Tiêu chí Pass/Fail (EN 1822 / ISO 29463)
Hiệu suất tổng × xuyên thấu cục bộ — phải đạt cả hai
| Cấp | Hiệu suất tổng ≥ | Xuyên cục bộ ≤ | Xuyên tổng ≤ | Ghi chú |
|---|---|---|---|---|
| H13 | 99.95 % | 250× tổng | 0.05 % | Scan tùy chọn |
| H14 | 99.995 % | 5× tổng | 0.005 % | Scan từng đơn vị |
| U15 | 99.9995 % | 5× tổng | 0.0005 % | Scan từng đơn vị |
| U16 | 99.99995 % | 5× tổng | 0.00005 % | Scan từng đơn vị |
| U17 | 99.999995 % | 5× tổng | 0.000005 % | Test đơn điểm thay thế |
Đạt tổng không có nghĩa không rò cục bộ.
Both thresholds must pass:
- 1Overall efficiency ≥ spec (H14 ≥ 99.995%)
- 2Local penetration ≤ 5× overall (for H14 and above)
Example: an H14 measures overall penetration 0.003% (pass). But one spot reads 0.02% — that's 6.67× the overall, exceeding the 5× threshold. Fail. The filter must be reworked or scrapped.
Three Common Failure Modes
Failure 1: seal-bead pinhole
The most common. Gas bubble in the sealant during cure, or a fiber trapped during application, creates a pinhole. Signature: anomaly points cluster around the frame perimeter, center of the filter reads normal.
Failure 2: internal media displacement
The media shifted during manufacturing, a small crack opened inside a pleat. Signature: points appear in the center area, seemingly random, but all on the same pleat.
Failure 3: incomplete sealant fill
The gel between media and frame was discontinuous, air bypasses through the frame-media gap. Signature: high penetration along an entire edge, not single points.
Practical Notes
1. Scan Test is mandatory, not a premium
Many customers cut the scan test to save cost. The saving is <5% of filter unit price, but the risk is the entire filter failing to deliver spec. Not worth it.
2. Require per-unit scan test report
Not "this batch sampled 10%, passed." Every filter needs its own scan test report with serial number, scan path map, anomaly list.
3. Re-test at the site
Factory tests measure the filter alone. After installation in an FFU/AHU, the seal between frame and housing may still leak. Customers should run an In-situ Leak Test at site — verifying "filter + installation" as a combined system.
"H14 factory passed" and "H14 performance once installed" are two different things. A scan test sits between them. Don't skip it.
