🔬 Testing 3nm Dualband Filters: High-Res Spectroscopy Reveals the Truth
There is a long-standing debate in the astrophotography community about using ultra-narrow 3nm dualband filters on moderate fast optics (f/5 or f/4). The manufacturers recommend suitability up to f/4 and refer to specialized preshifted high-speed filters for faster optics.
To move past marketing claims and forum speculation, I conducted a rigorous laboratory study. I reverse-engineered seven different 3nm/3.5nm dualband filters (Antlia ALP-T and Altair Astro) alongside an industry-gold-standard Chroma 3nm monoband filter.
🛠️ The Laboratory Setup
Instead of relying on the unlinked "factory test reports" included in the box, I utilized high-resolution spectroscopy on my workbench:
Spectrograph: Star’Ex in HiRes configuration (2400 l/mm grating, 10 µm entrance slit).
Sensor: Cooled ASI585M yielding an unprecedented linear resolution of 0.0075 nm per pixel.
Light Source: A modified 150W slide projector acting as a high-luminance continuum source, perfectly collimated via a 0.4 mm custom brass pinhole and a 50 mm finder lens.
Testing Protocol: Multi-axis transmission profiles measured from 0° up to 15° angle of incidence (AOI) to calculate the effective refractive index (n eff) and analyze structural multi-cavity deformations.
📊 Key Findings from the Lab
Bandwidth Deception (Our Lifesaver): None of the tested "3nm" budget filters were actually 3.0 nm wide. They measured between 3.3 nm and 3.9 nm FWHM. This manufacturing overhead is exactly what saves their performance on f/5 systems.
The "Silicon Lottery" (CWL Variance): At 0° perpendicular incidence, the central wavelengths (CWL) drifted wildly within the exact same product lines—ranging from -0.7 nm (back-shifted) to +1.1 nm (pre-shifted).
The n eff Fingerprint Discovered: By tilting the filters up to 15°, I mathematically isolated the layer physics:
Antlia: Robust multi-cavity design ("camel humps") with a stable n eff of 1.71–1.75. It barely survives f/4 by utilizing flat-top plateaus.
Altair: Drifts erratically in OIII with a very weak n eff of 1.53 (pure Gaussian shape), but steps up to 1.72 in H-Alpha via complex multi-cavity layers.
Chroma: A monoband masterpiece. It boasts a foundational n eff of 1.87 that structurally hardens to 2.13 under steep angles. Its perfect pre-shifted, triple-cavity plateau holds near-maximum transmission all the way down to f/2.8.
📉 Real-World Transmission at f/5 vs. f/4
Integrating the laboratory tilt series across a telescope's true light cone reveals the stark reality of effective transmission (T eff):
At f/5 / 5.7° AOI: Almost all tested filters survive, retaining 73% to 94% effective transmission due to their real-world 3.5nm+ width.
At f/4 / 7.1° AOI: The H-Alpha line lands directly on the 50% FWHM cutoff slope. A "lucky" pre-shifted filter (+1.1 nm) maintains 94% transmission and triumphs at f/4. A poorly centered back-shifted unit (-0.7) drops to 38% transmission, throwing away nearly a third of your gathered photons.
🔗 Read the Full Laboratory Report & See the BASS Diagrams
I have published the complete mathematical breakdown, full Excel control matrices, and overlapping BASS transmission charts mapping the transformation from Gaussian "domes" to multi-cavity "camel humps" on my blog.
👉 Read the full laboratory report with all BASS diagrams on my homepage (german)
📷 OIII-Band
📷 Ha-Band
📷 SII-Band
📷 Blueshift of an Altair Ha-OIII-Dualbandfilter (OIII-Band) at AOI of 0, 3, 5, 7, 10, 15°
📷 Blueshift of an Antila Ha-OIII-Dualbandfilter (Ha-Band) at AOI of 0, 3, 5, 7, 10, 12, 15°