When should hardware engineers stop using JLCPCB/PCBWay and move to engineering-grade manufacturers? This question comes up in every hardware startup, and the answer is more nuanced than "when you need quality."
The Short Answer
Switch when your design requires:
- Impedance tolerance below ±7%
- RF materials (Rogers, PTFE)
- HDI construction (microvias, via-in-pad)
- IPC Class 3 reliability
- Volume > 100 pcs with complex designs
Why This Matters
Budget prototype fabs optimize for one thing: getting you a board that looks correct as fast as possible. They do not optimize for process control, material consistency, or parametric performance.
At our facility, we process roughly 200 orders per month that are "migrations" from budget prototype fabs. The most common pattern: a 4-layer impedance-controlled board that happened to measure within spec on the prototype, but fails on the second build because the process window shifted.
The Technical Triggers
RF Materials
Budget fabs run FR-4 exclusively. If your design operates above 2-3 GHz and needs controlled dielectric properties, you cannot have it made at a budget fab. Period.
The subtle version: your 2.4 GHz WiFi board "works" on FR-4, but you're seeing 2-3 dB more insertion loss than simulation predicted. That's because budget fab FR-4 has Dk variation of ±10-15% batch-to-batch.
Impedance Control Below ±7%
Budget fabs advertise "impedance control" but the reality differs. They don't measure the actual Dk of incoming laminate lots, don't run test coupons on every panel, and their TDR measurement (if performed) is a spot-check on one coupon per lot.
From our data processing boards that "failed impedance" after budget fab prototyping: actual impedance scatter is typically ±12-15% for 50-ohm traces on their standard stackups.
HDI Construction
True HDI requires laser drilling, sequential lamination, and precise registration between buildup layers. Budget fabs offer mechanical drilling down to 0.2mm — they may advertise "blind vias" but these are controlled-depth mechanical drilling, not laser processing.
The Cost Reality
The math at prototype quantities favors budget fabs ($15 for 5 boards vs $200-500). But this misses three hidden costs:
Respin cost: When budget-fab prototype "works" but production doesn't, you waste 4-8 weeks and $2,000-10,000 debugging a manufacturing issue.
Qualification failure: Boards that pass bench testing but fail environmental qualification because manufacturing wasn't controlled tightly enough.
Field returns: The most expensive scenario — boards that ship and fail in the field.
Volume Crossover
At 50-100 boards for complex designs, engineering-grade fabs start competing on per-unit cost because:
- Better yield (98%+ vs 85-95%)
- Panel optimization (20-30% material savings)
- DFM review (catches cost-saving opportunities)
How to Evaluate an Engineering-Grade Manufacturer
Verify these capabilities:
- Material handling: Stock Rogers, Taconic, PTFE? Material certificates per lot?
- Process control: Cpk data for drilling, plating, etching?
- Testing: TDR impedance, 4-wire Kelvin, ionic contamination, microsection in-house?
- Certification: IATF 16949, ISO 13485, AS9100 as applicable?
Summary
The choice isn't "good vs bad" — it's matching your fab's capabilities to your design's requirements. Use budget fabs for what they excel at. Use engineering-grade manufacturers when precision matters.
The expensive mistake isn't paying more per board — it's discovering at qualification that your "working prototype" was working by accident.
Originally published at AtlasPCB Engineering Blog — we're a China-based engineering-grade PCB manufacturer specializing in RF, HDI, and impedance-controlled fabrication.
Top comments (0)