Last spring, our small Tier 2 aerospace stamping shop landed a dream contract: 50,000 structural brackets for a satellite constellation, machined from 0.8mm thick Inconel 718, with critical requirements of ±0.01mm on mounting hole positions, ±0.05mm on bend angles, and zero micro-cracks across the entire run. We'd done dozens of automotive and industrial stamping jobs, so we thought we could reuse our standard die design process. We were wrong.
Our first trial run had a 78% reject rate. The holes were oval, the bend angles were 7 degrees off spec, and micro-cracks were spreading from the punch edges on 1 in 5 parts. The customer gave us 2 weeks to fix the issue or lose the contract entirely. We scrapped our initial die design, spent 3 weeks reworking the process, and ended up hitting 99.7% first-pass yield for the full run---but the mistake cost us $72,000 in scrap and rework, and almost destroyed our reputation with a high-value aerospace customer.
Aerospace metal stamping tolerances are in a league of their own. Where automotive stamping usually works with ±0.1mm tolerances for most features, aerospace critical features often fall between ±0.01 and ±0.05mm, with zero tolerance for micro-cracks or surface flaws that could cause part failure in flight. Off-the-shelf die sets don't cut it here, and generic die design rules from other industries will leave you scrambling to hit spec.
Over the last 2 years, we've built 17 custom low-tolerance aerospace die sets for Inconel, titanium, and high-strength aluminum parts, and learned the hard way what works (and what doesn't) for this niche. These 7 rules will help you avoid the mistakes we made, without blowing your budget on overengineered designs.
Lead with material testing, not part prints
The biggest mistake we made on that first satellite bracket job was jumping straight from the customer's part print to CAD modeling, no material testing required. We knew Inconel 718 was a high-strength nickel-based superalloy, but we didn't test the exact mill batch we were using to measure its springback, work-hardening rate, or wear behavior on our chosen die steel.
The result? We designed the die with standard 8% punch-to-die clearance for piercing, but Inconel work-hardens so aggressively that it required near-zero clearance to avoid burrs and hole ovality. We also set the bend die to the final part angle, ignoring that Inconel 718 has 12-15% springback after forming---so every part came out 8 degrees too narrow.
The fix is simple: before you open your CAD software, run 3 core tests on the exact material batch you'll be stamping:
- Tensile and springback tests to calculate exact bend compensation and forming forces
- Wear tests to see how fast your chosen die steel will blunt or wear when stamping your material
- Micro-crack tests to confirm the edge radius you need on punches to avoid work-hardening cracks
For that satellite bracket run, we tested 3 different die steel grades before settling on powdered high-speed steel (CPM 10V) for critical punch and die inserts, which held its edge 3x longer than standard D2 tool steel we'd used for automotive jobs. The $1,200 we spent on material testing saved us $72,000 in scrap on the first run alone.
Design for near-zero clearance on critical features, not standard industry specs
Standard sheet metal stamping uses 5-10% clearance between punch and die for non-critical parts, but that rule goes out the window for low-tolerance aerospace features. For our Inconel brackets, standard 8% clearance on the 0.8mm thick material would have been 0.064mm---enough to cause 0.03mm of hole ovality, already 3x our ±0.01mm hole tolerance.
For all critical features (mounting holes, mating surfaces, bend lines), we used 0.01mm or less clearance, matched to the exact springback and wear behavior we measured in our material tests. We added 0.005mm radius to all punch edges to eliminate micro-cracks from work-hardening, and used PVD TiN coating on critical inserts to reduce wear and friction, so the clearance stayed within spec for the full 50,000-part run.
For bend features, we didn't use standard bend tables from our CAD software. We built a custom bend die with adjustable stops calibrated to our springback test results, so we could fine-tune the bend angle in 0.01mm increments after the first trial run, no full die rebuild required.
Control tolerance stacking from the first CAD model
Aerospace parts almost always require multiple stamping operations: piercing, bending, trimming, sometimes forming. If you design each operation with ±0.005mm tolerance, the total stack across 3 operations is ±0.015mm---already 50% over our critical hole tolerance, before you account for die wear, press vibration, or material variation.
To fix this, we run full tolerance stack analysis on every die design before we cut any steel, and limit total cumulative tolerance across all operations to ≤30% of the final part's critical tolerance. For our satellite brackets, that meant each operation had ≤0.003mm tolerance, so we used precision-ground die blocks with ±0.001mm flatness, and located every feature off a single master datum on the die base, not multiple loose datums that could shift during operation.
We also added 0.001mm resolution adjusting screws on all critical die positions, so we could tweak hole position or bend angle after the first trial run without scrapping the entire die set. That small change let us adjust the hole position by 0.008mm after our first test run, bringing us from 92% first-pass yield to 99.7% in one afternoon.
Build in-process measurement directly into the die set
Most shops design dies first, then figure out how to inspect parts after stamping. For low-tolerance aerospace parts, that's too late: by the time you catch a drift in tolerance, you've already scrapped hundreds of parts. We learned this the hard way on an early titanium bracket job, where die wear caused hole positions to drift 0.02mm after 8,000 parts, and we didn't catch it until we got a customer complaint 2 weeks later.
For all our aerospace die sets now, we integrate in-process measurement directly into the die design:
- Laser micrometers mounted to the die base, triggered on every part, to measure critical hole position, bend angle, and edge radius in real time
- If any feature drifts more than 50% of the tolerance limit, the system triggers an automatic press stop, so we can adjust the die before 10 parts are produced
- A built-in CMM fixture at the unload station, so operators can run a full first article inspection (FAI) in 5 minutes right on the press, no need to send parts to an external lab
For the satellite bracket run, the in-process measurement caught a 0.006mm drift in hole position after 12,000 parts, caused by a loose die mounting bolt. We tightened the bolt in 2 minutes, and avoided scrapping an estimated 8,000 out-of-spec parts.
Design for repeatability, not just one-off accuracy
A lot of custom die shops design a die to make one perfect part, then wonder why it falls out of tolerance after 5,000 parts. For aerospace, most contracts require 50,000 to 200,000 part runs with consistent tolerances, so you need to design for wear and repeatability from day one.
For our die sets, we add 2mm thick wear plates to all critical punch and die surfaces, so when they wear down 0.5mm over 50,000 parts, we can grind them flat and add shims to bring clearance back to spec, instead of scrapping the entire die. We also use hardened, ground guide pins and bushings with 0.002mm total clearance, to eliminate die shift from press vibration---we found that standard 0.01mm guide clearance caused 0.03mm of die shift per stroke on our 200-ton press, which was already out of our critical tolerance.
We also design all our aerospace die sets to be modular: if a customer updates a part print with a small change, we can swap out a single insert instead of building an entirely new die set. For a follow-on contract we did last year for revised satellite brackets, the modular design saved us $18,000 in tooling costs and 3 weeks of lead time.
Integrate traceability into the die design, not as an afterthought
Aerospace parts are governed by AS9100 quality standards, which require full traceability for every component, test, and design decision for 10+ years. We learned this the hard way when a customer asked for documentation on why we used 0.01mm clearance on the critical mounting holes, and we only had a CAD file with no supporting test data. They delayed our $120,000 payment for 3 weeks while we pulled old test records and recreated the documentation.
Now, every custom aerospace die set we design comes with a full traceability package built in:
- Unique serial number etched into the die base, linked to a cloud log of every part run, maintenance event, and wear replacement
- Full material test reports for all die components and stamped material, plus mill certs from the material supplier
- GD&T callouts on every die feature, plus full tolerance stack analysis reports
- FAI results from the first trial run, plus a maintenance guide with clear wear limits for each component
We also add a small QR code to the die base, so operators can scan it to pull up the full die history, maintenance schedule, and part specs in 2 seconds, no paper records needed.
Validate with a fast FAI process before full production
Aerospace customers require formal first article inspection (FAI) before approving full production runs, but most shops send parts to an external lab, which takes 2-3 weeks and delays the contract. We design all our die sets with built-in FAI capabilities, so we can run a full, customer-compliant FAI in 15 minutes right on the press.
For the satellite bracket job, we added 3 precision gauge pins that fit directly into the critical mounting holes, and a digital angle gauge mounted to the bend station, so the operator could confirm all critical features were within spec in 5 minutes. We submitted the FAI report to the customer the same day as the trial run, and they approved full production 48 hours later---2 weeks ahead of our original timeline.
The bottom line
Custom low-tolerance aerospace die sets aren't cheap: our satellite bracket die cost $42,000, 30% higher than our initial estimate for a standard automotive die. But the cost of scrap, rework, and lost aerospace contracts is far higher: that first failed run would have cost us the $1.2M contract, plus $300,000 in penalties for late delivery.
The biggest mistake we see shops make is treating aerospace stamping the same as high-volume consumer or automotive stamping. The rules are different: material comes first, tolerance stacking is non-negotiable, and traceability is built into the design, not added at the end.
If you're working on a low-tolerance aerospace stamping project, drop a comment with your biggest pain point---whether it's springback on titanium parts, die wear over long runs, or FAI delays, we've been there, and we're happy to share what we've learned.