The brief
The customer is a US-based avionics integrator building an optical sensor pod for a business-jet retrofit program. Their bracket sits between the pod housing and the airframe longeron. The mating face to the pod has to hold profile tolerance of ±0.01 mm to a datum triplet, and the four M6 mounting holes through the longeron side have true-position ⌀0.02 mm. They had been prototyping with a domestic shop at a unit price they couldn’t justify for production.
RFQ came in on a Monday morning. We ran DFM against our pre-quote checklist, flagged two radii that were tighter than our smallest standard end-mill, and asked if we could open them from R0.3 to R0.5. The customer’s stress analyst came back within the day with the okay. That one DFM round saved us an EDM sinker step that would have added four days to every lot.
Why this job was not routine
Titanium is an easy material to quote for and a difficult one to execute on. The combination of low thermal conductivity, high elastic modulus drop at temperature, and work-hardening means that the same toolpath that works on 6061 aluminum will burn tools, chatter, and move the part after every rib is cut out. On a bracket with a 3 mm minimum wall and four tight-tolerance features on three planes, you cannot simply machine it in one pass.
We planned three phases. Rough mill to within 0.5 mm of final on all features, with continuous-engagement trochoidal toolpaths to avoid dwell. Stress-relief anneal at 704 °C for one hour, furnace cool. Then semi-finish at 0.15 mm stock, and finish at 0.04 mm with a new flute set. The part that goes onto the CMM is the fifth pass it has seen.
Fixturing — the part of the plan that usually kills titanium jobs
We built a dedicated vacuum plate with a sacrificial aluminum base and four locating pins matching holes we drilled in the raw plate before rough milling. The vacuum holds the part flat without clamp distortion on the mating-face side. Two top clamps hold the outer rib during roughing and come off for finishing — the vacuum alone is sufficient once the ribs are at semi-finish stock because the cutting forces drop below the vacuum hold.
This fixture costs us about 14 hours to design and build per new job, and on a run of 85 it more than pays for itself. On one-off prototypes we use a soft jaw setup instead, which takes under 2 hours but costs us roughly 10 % of the tolerance band.
Results from the four lots
Lot 1 (April 2026) shipped 85 parts, all inside profile tolerance. Worst measured deviation on the critical profile was 0.007 mm — inside the band with 30 % margin. Four parts came back from the customer with a cosmetic reject on the as-machined surface finish (they wanted Ra 1.2 μm, we shipped at Ra 1.6 μm). We added a dedicated finish pass with a 2-flute 6 mm ball mill at 8,000 RPM for lot 2 and held Ra 0.9 μm from lot 2 onward.
Lots 2 and 3 shipped a combined 170 parts with no rejects. Lot 4 is on the floor now — we’ve released the first 20 parts through CMM and the worst profile reading so far is 0.005 mm.
What the customer asked us to publish
The customer approved publication of the process description, material grade, and tolerance numbers on the condition that the final application and their company name stay out. Everything above is shipped-to-spec data from the first four production lots. If you are working on an aerospace bracket with a similar profile, the fastest way to get comparable numbers on your part is to send us the STEP file plus the critical-dimension callouts — we’ll tell you within a business day what tolerance we can realistically hold and at what price.