Planning your first heavy-ion SEE test can feel overwhelming. There are facility choices, beam cocktail selections, flux/fluence decisions, test board design constraints, device under test (DUT) preparation and logistics to coordinate. This thread is meant to be a practical walkthrough for engineers who are scoping their first campaign.
Here’s a rough sequence that works well for most programs:
1. Define your test objectives clearly. Are you characterizing SEU, SEL, SEFI, SET, or all of the above? Are you down-selecting technology options with go/no-go requirements? Destructive effects like SEL and SEB require different monitoring setups (current limiting, power cycling logic) than soft errors. Write down exactly what “pass” and “fail” mean for each device before you book beam time.
2. Selecting your facility. The major US heavy-ion facilities include Texas A&M Cyclotron (TAMU), Lawrence Berkeley National Lab (LBNL 88-Inch), Brookhaven National Lab (BNL NSRL and Tandem), and Michigan State University’s Facility for Rare Isotope Beams (MSU/FRIB). Each has different maximum LET capabilities, beam uniformity characteristics, scheduling lead times and procurement mechanisms. TAMU’s cocktail gives you a good spread of LETs up to ~85 MeV-cm²/mg with a single tune; LBNL offers a wider range but you may need multiple tunes; MSU/FRIB provides high-energy beams with long range in silicon, which is valuable for testing devices with thick sensitive volumes or through-package irradiation. European options include UCL in Belgium, GANIL in France, and RADEF at the University of Jyväskylä in Finland; RADEF is widely used for ESA-qualified testing and has a well-established cocktail with good LET coverage.
3. Design your test board for the beam environment. The DUT needs to be accessible to the beam, usually through a vacuum window or in-air. Keep the DUT area clear of tall components. Plan your cable routing to exit the beam chamber. Consider thermal management: some devices need active cooling during irradiation.
4. Determine your LET and fluence plan. For Weibull characterization, you typically want data at 5-8 LET values spanning from onset through saturation. Standard practice is to start at a high LET to confirm the effect exists, then work downward to find threshold. Fluence at each point should be enough to get reasonable statistics — at least 100 events at saturation, and enough particle fluence at low LETs to set meaningful upper-bound cross-sections if you see zero events.
5. Coordinate logistics. Beam time is expensive ($1K-$5K/hour depending on the facility). Have backup DUTs, spare boards, and a clear run plan so you don’t waste shifts troubleshooting setup issues.
What has your experience been? Any facility preferences, lessons learned, or “I wish someone told me this” moments from your first beam campaign? What do you wish you’d done differently in test planning?