BUYER’S DECISION GUIDE
How to choose a US semiconductor packaging partner.
Choose a US OSAT by matching the partner’s model to your program, not by chasing the lowest per-unit price. Mega-OSATs win on high-volume, locked designs but carry high minimums; boutique wire-bond shops take low volumes but rarely co-design; a co-design partner runs prototype-to-production on one domestic line, iterates alongside your device, and returns die-level yield data. Weigh process coverage, volume fit, workmanship, and traceability before MOQ.
/ US-soil assembly / MIL-STD-883 workmanship / engineer-to-engineer
HOW TO USE THIS GUIDE
Two questions, then seven criteria.
First decide which kind of partner your program needs — the three sourcing models below behave very differently. Then work the seven criteria to compare suppliers inside that model. The criteria are ordered so the hard fits (does the supplier run your process, at your volume, on US soil) come before the differentiators (co-design, data, communication).
Seven criteria, ordered fit-first. Earlier criteria qualify a supplier; later criteria separate the good ones.
THREE SOURCING MODELS
Same word, three very different partners.
“OSAT” covers everything from a high-volume mega-house to a two-bench wire-bond shop. They are not interchangeable. The right model depends on where your device is in its life — not on which is cheapest per unit. This compares the models by category; it names no companies and knocks none.
These are general model patterns. Any individual supplier may sit between them — confirm specifics with each one you evaluate.
CRITERION 01 — PROCESS COVERAGE
Does one partner run every process your device needs?
“We do packaging” spans simple epoxy die-in-package to flip-chip copper pillar with fluxless reflow. A supplier can be excellent at one and have never run the other.
For a device that is still changing, coverage breadth under one roof is worth more than a low headline rate. When wire bond lives at one shop and flip chip at another, every interconnect change becomes a re-qualification on a new line, with a new process history. The relevant families are wire bonding, die attach and die bonding, flip-chip assembly, hermetic packaging, and RF and mmWave packaging. Each links to a detailed capability page — use them to learn what each process family requires and what to ask.
CRITERION 02 — VOLUME FIT
Can the partner take you from first article to low-rate production?
Low volume and prototype-to-production is a different business than high-volume assembly. A re-qualification handoff between a prototype house and a volume supplier re-proves the process on different equipment, under a different quality system.
That handoff costs schedule, NRE, and yield risk, because the production line has no memory of the development history. The alternative is one partner that runs first-article and engineering builds as the normal case and carries the same locked process onto a low-rate line. That continuity requires both the process range to run prototype complexity and the discipline to lock a repeatable recipe — confirm both, and ask the minimum run.
CRITERION 03 — DOMESTIC SUPPLY
Where does the work — and the design data — physically live?
Onshore assembly is a commercial risk decision, not a slogan. Where assembly, process development, and engineering happen determines who can see your design data and how far you can audit the line.
A domestic partner keeps design data, process provenance, and the supply chain under US jurisdiction and open to on-site audit, and it sources primary materials — substrates, die-attach materials, bond wire, sealing materials — from US or allied-nation suppliers with auditable chains. That removes a category of schedule and IP risk that offshore concentration introduces. This is not a claim that offshore is unsafe; it is that, for prototype and low-volume programs, the domestic risk profile usually wins.
CRITERION 04 — WORKMANSHIP
What workmanship standard does the line run to?
A quality claim without a reference standard is a description, not a claim. For high-reliability microelectronics, the workmanship and test-method benchmark is MIL-STD-883.
The question is not whether a supplier mentions MIL-STD-883 — it is whether the line is built to it: documented methods for die attach, wire bond, and hermetic seal; first-article inspection discipline; per-build process-traveler data; and a defined nonconformance path. Traceability is the proof. A supplier who cannot produce a sample traveler from a prior build does not have the system to produce one from yours. Certification schemes such as ISO 9001 signal documented quality management, but they are not a substitute for the workmanship standard that actually governs the bench.
CRITERION 05 — CO-DESIGN & DFM
Will an engineer design the package with you?
Design-for-manufacturability feedback at the start of a program is worth more than any amount of yield data at the end. A mistake caught in design review never becomes a first-article failure.
The question is whether the person who takes your specification is a process engineer who has run your interconnect family — reviewing pad layout, substrate selection, and bump geometry against a known process window — or a sales function that routes to a floor operator you never meet. Co-design goes further: the packaging is developed alongside the device, so the die and its package are optimized together rather than the package being fit to a finished die. Ask how iteration is handled — a failed first article should be a defined step with go/no-go gates, not an undefined cost overrun.
CRITERION 06 — DATA & TRACEABILITY
Do you get die-level data, or a pass/fail count?
For a new or iterating device, the closed data loop is the fastest route to a stable process. Die-level traceability turns a failed build into a specific engineering signal instead of a mystery.
Lot-level records tell you a batch passed or failed. Die-level traceability ties each die to its build history and inspection data, so yield loss is traced to a specific process step and fed straight back into the next iteration. That loop — build, measure, attribute, correct — is what separates a partner that improves your yield from one that just reports it. Ask what data you receive at the end of a build and whether it is die-level or a summary count.
CRITERION 07 — IP & COMMUNICATION
How is your IP handled, and who do you actually talk to?
Your packaging partner handles your most sensitive design data. How that data is controlled — and how quickly you can reach the person who understands it — is a program variable, not a vendor detail.
Ask how design data is stored, who can access it, and how NDA obligations flow to anyone who touches the line. A domestic, single-site partner with a small engineering team is inherently easier to control than a multi-country chain of subcontractors. Communication is the other half: same-timezone, engineer-to-engineer contact turns a week-long relay into a same-day answer, and shortens every iteration. Both are strongest when the same team owns the design conversation, the process, and the data.
WHERE HEISLER FITS
The co-design model, run in practice.
Heisler is the third model above: a domestic OSAT built for low-volume prototype-to-production, co-designing the package alongside the device and closing the data loop for die-level traceability. Here is what that covers concretely, for medical and bio, RF and mmWave, sensor and MEMS, aerospace and defense, and industrial devices.
Gold ball, aluminum wedge, and ribbon bonding — the full interconnect range under one roof. Wire bond →
Eutectic and epoxy die attach, developed to the thermal and mechanical needs of the device. Die bonding →
Flip-chip and copper-pillar assembly for higher-density, higher-frequency interconnect. Flip chip →
Hermetic sealing for devices that need a controlled, sealed cavity environment. Hermetic →
RF and mmWave assembly where interconnect parasitics and substrate choice drive performance. RF / mmWave →
In-house inspection and metrology feed the die-level data loop — yield attributed, not guessed.
FAQ
Answers before you ask.
What is the difference between an OSAT and a contract assembler?
An OSAT (outsourced semiconductor assembly and test) owns the full back-end: die attach, interconnect, encapsulation or hermetic seal, and electrical or visual test, with process engineering and traceability behind it. A contract assembler typically builds to a fixed print and hands the parts back. The practical difference is engineering ownership. An OSAT develops and locks the process and can tell you why a build yielded the way it did; a build-to-print assembler runs the recipe you supply and does not own the process window.
Can a US OSAT handle low-volume prototype builds?
Yes, and this is where a domestic co-design OSAT fits best. Mega-OSATs are built around high-volume, qualified flows and carry high minimums, so a prototype is often out of scope or expensive NRE. A low-volume US OSAT runs first-article and engineering builds as the normal case, iterates the process alongside your device, and carries the same process onto a low-rate production line without a re-qualification handoff. Ask any prototype supplier what their minimum run is and how a first-article iteration is handled if the result misses spec.
Why choose a domestic packaging partner?
Domestic packaging keeps your design data, process provenance, and supply chain inside US jurisdiction and open to on-site audit, and it shortens the engineering feedback loop to same-timezone, engineer-to-engineer conversations. For programs where schedule predictability, IP control, and supply resilience matter as much as unit cost, an onshore partner removes a category of risk that offshore concentration introduces. It is not that offshore is unsafe; it is that the risk profile is different, and for prototype and low-volume work the domestic profile usually wins.
What is MIL-STD-883 workmanship?
MIL-STD-883 is a microelectronics test-method and workmanship standard. It defines how operations such as die attach, wire bond, and hermetic seal are performed and inspected, and the methods used to screen and qualify parts. Used here it is a quality reference, not a market claim: a line built to MIL-STD-883 workmanship methods has documented procedures, first-article inspection discipline, and per-build traceability. The right question for a supplier is what standard governs their line and whether they can produce a process traveler from a comparable prior build.
Which packaging processes should one partner be able to cover?
For a device that is still evolving, the fewer handoffs the better. A single partner that covers wire bonding, die attach, flip chip, hermetic sealing, and RF and mmWave assembly under one roof lets you change interconnect strategy without changing suppliers, and keeps one process history for the whole device. When each process family lives at a different shop, every design change becomes a re-qualification on a new line. Confirm that the specific interconnect your design needs is one the supplier actually runs, not just lists.
What does die-level traceability give a buyer?
Die-level traceability means each die carries its build history and inspection data, so yield loss can be traced to a specific process step rather than guessed at. For a new or iterating device that closed data loop is the fastest route to a stable process: it turns a failed build into a specific engineering signal instead of a mystery. Ask whether traceability is lot-level or die-level, and whether you receive the yield and metrology data or just a pass or fail count.
Process More.
The process starts with a conversation.
Heisler’s scope runs from circuit design and firmware through wafer-level processing, packaging, and supply chain — for programs that need more than an assembly house. Bring us your device: process, geometry, substrate, and volume. We will tell you honestly whether we are the right fit, and what a first engagement would look like.
Request a Capability Brief →/ US-soil assembly / MIL-STD-883 workmanship / engineer-to-engineer access