Prototyping and small-batch builds
A prototype's job is to answer a question. We design the build around the question rather than around the calendar.
A prototype that does not answer a question is an expensive way to feel busy.
Before anything is built we write down what this build is for: does the thermal path work, does the radio reach, does the assembly physically go together in that order, does the sensor see what we think it sees. The answer determines what gets built, how quickly, and how roughly.
Why speed comes from the loop, not the machine
Prototyping time is rarely spent building. It is spent waiting — for a quote, a shipment, a queue slot, a decision.
| Where the time actually goes | What shortens it |
|---|---|
| Procurement of mechanical parts | Additive manufacturing in-house — no quote, no shipping |
| PCB fab queue | Quick-turn EU fabs with a working relationship and a correct dataset |
| Component lead time | Second-source footprints designed in, so a shortage is not a stop |
| DFM rejection and re-release | Design rules built from the fab’s real capability sheet |
| Bring-up guesswork | A structured procedure with pass/fail gates instead of probing hopefully |
Compress those and iteration becomes days. That changes what a prototype is for — you can afford to build one to answer a single question, rather than loading every open question onto one board and waiting six weeks to learn which assumption was wrong.
In-house additive manufacturing
Functional parts in engineering polymers, on our own machines, same day.
Not visual mockups. Polycarbonate and comparable engineering filaments, with infill pattern, wall count and print orientation chosen against the load path, and shrinkage compensated on the dimensions that have to fit. Parts that survive handling, mounting and moderate mechanical load — enough to answer whether the assembly works before anyone commits to tooling.
This is the piece that changes project rhythm most. A bracket that does not fit stops being a two-week problem and becomes an afternoon.
Boards
Quick-turn fabrication and assembly through EU fabs, coordinated by us with a dataset built to their real process window — so DFM is something the board already passes rather than something it discovers.
Then bring-up on our bench, in order: rails current-limited and measured one at a time, clocks and reset, boot straps at the pin, debug attach, boot media, then interfaces individually against the schematic. You get the instrumented report, including what failed.
Truthful scope
In-house: additive manufacturing in engineering polymers, the bench, bring-up and validation, and design for every process below.
Through qualified manufacturing partners: PCB fabrication and assembly, CNC machining, sheet metal, injection molding. We design and coordinate; they build.
Prototypes, source files and the bring-up data are all yours, in native format and under EU jurisdiction, whether or not the project continues past this stage.
Common questions
Straight answers
How fast is a functional prototype, realistically?
Mechanical parts in engineering polymers: same day to two days on our own machines, because there is no procurement step. A quick-turn PCB fabricated and assembled in the EU: typically one to two weeks depending on the fab's queue and stencil requirements. The honest constraint is rarely our side — it is component lead time on whatever part you chose.
Are 3D-printed parts good enough to test with?
For mechanical fit, brackets, enclosure iterations and anything that has to survive handling and moderate load, yes — printed in engineering polymers such as polycarbonate rather than display-grade filament, with infill and orientation chosen for the load path. They are not injection-molded parts and we do not present them as one: surface finish, isotropy and long-term creep all differ. For what a prototype is for, they answer the question.
What does an instrumented bring-up report contain?
What was measured, at what point, against what expected value, and whether it passed. Rails in sequence with current draw noted, clocks and reset confirmed, boot straps measured at the pin, debug attach, then each interface proven against the schematic. Plus what was modified to make something pass, and what is still unproven. The unflattering entries are the ones that decide the next spin.
Can you do small production batches?
We coordinate them with EU fabrication and assembly partners and manage the DFM iterations, but we are an engineering house rather than a contract manufacturer — we do not mark up your build. Past a few hundred units you want a real EMS relationship, and we will help you set it up and hand over a dataset they can build from.
What happens between prototype and production?
A defined transition rather than a hopeful one: DFM and DFA against your chosen assembler's actual process, panelisation, test strategy, a BOM with second sources on anything with single-source risk, and documentation complete enough that a manufacturer builds it without phoning you.
Evidence
Where this has been applied
Competition robot: designing for repeated high-energy impact
A reliability testbed that punishes every shortcut — and where a destroyed capacitor turned into a transient-survivability design rule.
mmWave presence detection for building automation
Radar instead of PIR, because PIR cannot see a person sitting still.