2026-07-09
V-band flanges seal metal-to-metal without a gasket. That means the surface finish of the sealing cone isn't cosmetic — it's the entire sealing mechanism. Surface roughness is measured in Ra (arithmetic average roughness), expressed in microinches (µin) or micrometers (µm). For V-band exhaust flanges, the industry has converged on 32 µin Ra as the sweet spot, and there's real physics behind that number.
Why not smoother? You'd think a mirror finish (4-8 µin) would seal better. It doesn't. At exhaust temperatures (1200-1800°F), the flange faces expand, contract, and micro-slip against each other. A too-smooth surface has nowhere for oxide scale to key into — the mating faces polish each other and start to gall. Worse, ultra-smooth surfaces can cold-weld under clamp load, then tear when disassembled.
Why not rougher? Above 63 µin, the peak-to-valley height exceeds the elastic deformation the V-band clamp can provide. The clamp squeezes the peaks flat, but valleys remain as leak paths. Boost pressure and exhaust pulses find them immediately.
The Ra sweet spot logic:
Real-world example: Turbosmart and Vibrant Performance both spec their V-band flanges at 32 µin Ra on the 20° cone. Cheap eBay flanges often measure 80-125 µin — they look shiny but the machining marks are deep. Builders who report "V-band leaks that won't seal no matter the torque" almost always have surface finish problems, not clamp problems.
Rule of thumb: Convert Ra to peak-to-valley height with Rz ≈ 4 × Ra. A 32 µin Ra surface has roughly 128 µin (0.0013") peak-to-valley. V-band clamps provide about 0.002-0.003" of elastic deformation at the sealing face — so 32 µin gives you a 2:1 safety margin. At 63 µin Ra (Rz ≈ 250 µin, 0.0025"), you're right at the limit.
How to measure: A profilometer stylus dragged across the cone perpendicular to the machining lay. Fingernail test works as a field check — if your nail catches on machining ridges, it's too rough.