A connector that cannot tolerate reflow changes the assembly route for the whole circuit card. Its insertion must be scheduled after the relevant oven passes, with a suitable soldering operation and access for inspection. Assembly planning therefore starts with the components and their process limits, then connects printing, placement, soldering and testing into an approved manufacturing sequence.
For circuit cards containing SMT devices, through-hole connectors and hidden solder joints, the useful deliverable is a released process route with evidence at each decision point. Commercial supply arrangements belong in the PCB assembly service scope; enclosure integration and finished-product work extend into electronics assembly. A populated board may still need programming, calibration and system validation before it is ready for use.
Which assembly route fits the components?
SMT and through-hole assembly describe how components connect to the board. BGA, QFN and CSP describe package forms; they are not separate alternatives to SMT. A mixed board may need several soldering operations, but inserting every through-hole part before reflow is not a universal sequence.
| Board condition | Route to evaluate | What must be resolved before release |
|---|---|---|
| Single-sided SMT | Print paste, inspect deposits, place, reflow, inspect joints | Package orientation, paste delivery, thermal profile and test access |
| Double-sided SMT | Define first side, second side and both reflow passes | Retention of first-side parts, cumulative thermal exposure, support and inspection access |
| SMT with through-hole parts unsuitable for reflow | Complete applicable SMT passes, insert those parts, then selective, wave or controlled manual soldering | Component temperature limits, solder-side access, neighboring parts, flux and cleaning compatibility |
| Through-hole parts explicitly suitable for pin-in-paste | Deposit the required paste and insert the qualified parts before the relevant reflow pass | Connector heat compatibility, hole/lead geometry, paste supply, insertion effects and resulting hole fill |
| Sensitive parts or mechanical items fitted late | Add a defined secondary operation | Handling damage, local heat exposure, fastening, contamination and retest requirements |
Samtec's connector selection guidance specifically distinguishes wave soldering from paste-in-hole and requires a paste-in-hole connector to withstand SMT processing. A connector's through-hole pins alone do not establish that compatibility.
Selective soldering applies solder locally, but preheating and nearby thermal exposure still matter. Wave soldering may need a pallet to protect features on the solder side, and that pallet must allow access to the intended joints. Manual soldering also needs controlled tools, instructions and acceptance criteria. The choice is driven by geometry, heat tolerance, repeatability and production volume, rather than a rule that one method is always superior.

Connector placement and surrounding components are inputs to soldering access and assembly sequence.
What does a fine-pitch package actually require?
HILPCB’s SMT assembly capability options include reviewed 0.3 mm-pitch packages, 01005, high-pin-count BGA and QFN, and mixed, flex or rigid-flex assemblies. Specialized routes include flip-chip, wire bonding, die attach, chip-on-board and conformal coating when the package, tooling and process are compatible. Our engineers review design, fiducials and via-in-pad for the selected route and support procurement/EMS handoff; no smaller-than-01005 capability is implied.
A package size identifies a design constraint, not a qualified production process. Murata's LQP02HQ announcement identifies its 01005 inductors as 0.4 × 0.2 mm in plan dimensions. That does not establish an assembler's yield for a board mixing those parts with large power terminals, shields or tall connectors.
For a 0.3 mm-pitch device or a 01005 passive, submit the actual manufacturer part number, package drawing, land pattern and surrounding layout. Confirm the printing, feeding, placement, inspection and rework approach for that combination. Lead pitch, passive size, stencil aperture and placement accuracy are different quantities and should not be substituted for one another.
Bare-die work is a separate process review. COB means chip-on-board: a die is mounted directly on the board, with an appropriate interconnect and protection scheme. Wire bonding and flip-chip attachment have different materials, tooling and validation needs. A conventional SMT quotation should not be assumed to include either route.
A QFN stencil example with its conditions
NXP's AN1902, revision 9, 28 April 2021 provides QFN/SON assembly guidance. Its Table 3 gives these typical starting points, subject to the other components on the board and the actual printing process:
| QFN/SON lead pitch | Typical stencil thickness in NXP Table 3 | Decision for a mixed board |
|---|---|---|
| 0.50 mm | 150 µm | Check whether nearby smaller apertures release paste adequately |
| 0.40 mm | 100–125 µm | Check whether larger joints receive enough paste; a stepped stencil may be an option |
For the exposed-pad land, NXP section 5.2.3 describes a target paste coverage of about 60–65% of the land area. Consider a hypothetical rectangular land measuring 4 × 4 mm: its area is 16 mm², so that percentage corresponds to 9.6–10.4 mm² of nominal paste-covered area.
This calculation is an area budget, not a finished stencil design. Aperture segmentation, web widths, stencil thickness, paste release, the via pattern and trial results determine the actual deposits. It is also not a 35–40% permitted void limit: printed paste area and the soldered joint's void area are different measurements.
NXP allows adjustment for thermal requirements and warns that excess paste can lift the package and impair terminal connections. It also advises against printing directly over unplugged vias. Record the approved stencil revision and inspect the first trial deposits and resulting joints before applying the approach to the production lot.
How should the thermal process be approved?
An oven setpoint is not the temperature experienced by every component. The assembly supplier needs a measured board profile, with thermocouples at representative locations and at sensitive or thermally demanding packages. NXP's QFN/SON guidance includes measurements on the board and package top; the selected device's additional instructions still apply.
Separate the component's classification limits from the solder paste's working window. A maximum package temperature is not a recommended production peak. The approved process must form the joints while respecting component, laminate, connector and material limits across every planned heat exposure.
Record the assembly side, conveyor and oven settings, thermocouple locations, measured traces, paste/alloy identity and the applicable limits. For two-sided assemblies, assess both passes. A substitution with a different body size or thermal mass may require renewed profiling even when its electrical function is equivalent.
Cleaning and coating must also fit the route. A no-clean flux label does not settle every application's residue requirements, coating adhesion or compatibility with connectors and sensors. Agree the required final condition and verification before choosing a cleaning step or omitting it.
Which defects does each inspection or test address?
A test plan needs both a defect list and an observation method. “AOI plus X-ray” is not a complete statement of electrical coverage, and passing a functional test does not reveal every workmanship defect.
| Method | Useful evidence | Remaining limitation to address |
|---|---|---|
| Solder paste inspection | Deposit position and measured paste geometry before placement | Does not establish the final solder joint or its service life |
| Optical inspection/AOI | Visible placement, polarity and accessible joint features | Hidden interfaces and obscured markings need other evidence |
| X-ray inspection | Views of hidden solder geometry, bridges or voids within the method's capability | Projection overlap, resolution and defect orientation affect detection; an image alone does not prove electrical function |
| ICT or flying probe | Electrical checks on accessible, controllable circuits | Parallel paths, access and protection constraints limit individual component coverage |
| Boundary scan | Structural interconnect tests supported by the devices and test model | Requires suitable devices, accurate BSDL/netlist and controllable/observable nets; a JTAG header alone is insufficient |
| Functional test | Specified behavior under the programmed stimulus, load and conditions | Unexercised modes, environmental behavior and product qualification remain outside that result |
QFN wettable flanks can make some terminal features visible, but they do not expose the center pad. NXP recommends optical and X-ray methods for QFN/SON inspection and discusses cross-sectioning for process optimization. The chosen inspection plan should identify the joint or defect being assessed, its acceptance criterion and the population inspected, rather than declaring every package passed because one image looked acceptable.
For powered tests, agree current limits, safe fixture states, firmware revision and how faults are handled. Store measured values where they matter, rather than only a pass label. A diagnostic failure should lead to a controlled disposition and retest, not repeated runs until the board happens to pass. Detailed method selection is covered in circuit board testing.

Accessible pads and concealed connections require different inspection and test approaches.
What must the first-article record contain?
The first article checks whether the released design and proposed process work together. Its approval should identify the configuration that was actually built. A photograph and a “passed” email leave too much ambiguity when the next lot uses a different component, stencil or program.
| Release item | Evidence to request | Decision if unresolved |
|---|---|---|
| Design and fitted configuration | Board revision, BOM/MPNs, placement data, polarity references and approved deviations | Hold affected positions or the build until discrepancies are reconciled |
| Assembly route | Side order, THT exceptions, soldering methods and secondary operations | Resolve incompatible thermal exposure or inaccessible joints before repeat production |
| Printing and soldering | Stencil revision, paste/alloy identity, first-trial observations and measured profile | Revise the process and inspect a new trial when the evidence is inadequate |
| Inspection and electrical release | Applicable acceptance documents/revisions, sampled population, coverage, results and exclusions | Resolve failures or obtain documented disposition; do not call untested features passed |
| Programmed/delivered state | Firmware, calibration requirements, fitted accessories and packaging instructions | Match shipment state to the order; identify work still required by the product team |
| Change and repair control | Approved alternatives, rework history, retest scope and authorization | Reassess the affected process or function before releasing changed boards |
First-article approval does not by itself establish long-term reliability or ongoing process stability. Define what is checked again during production and which changes trigger renewed approval. A new paste, PCB finish, alternate package, rework method or firmware revision can affect different parts of the evidence package.
What should accompany an assembly process review?
Send the released board data, BOM, placement file and assembly drawings together, including variants and omitted parts. Add component handling instructions, package-specific assembly guidance, critical mechanical constraints, connector requirements and any parts prohibited from particular thermal or cleaning operations.
For the test plan, provide expected functions, limits, stimuli, loads, programming files and access requirements. State the required acceptance documents and revisions, any product-specific criteria, records to retain and who may approve exceptions. Include prototype and repeat-build quantities so that stencil, pallet, fixture and programming effort can be evaluated against the actual route.
Ask HilPCB to confirm the proposed sequence, package feasibility, inspection/test scope and first-article deliverables for the submitted design. Keep that manufacturing release distinct from the product owner's system, environmental and regulatory qualification. The resulting quotation should identify the agreed work and any remaining validation responsibilities.
Frequently asked questions
What is the difference between a circuit card and a bare PCB?
A bare PCB provides the patterned conductors, insulation and connection features. A circuit card commonly refers to the populated assembly. That label does not say whether it has been programmed, calibrated, functionally tested or integrated into a finished product; the order must define its delivered state.
Must through-hole components be inserted before reflow?
No. Parts qualified for pin-in-paste may be inserted before their approved reflow pass. Parts that cannot tolerate that process generally need a later insertion and suitable selective, wave or manual soldering operation. Double-sided boards also need a defined side order and evaluation of repeated heat exposure.
Does 60–65% QFN paste coverage permit 35–40% solder voiding?
No. NXP AN1902's approximately 60–65% recommendation concerns paste coverage of the exposed-pad land before reflow. It is not a post-reflow void acceptance limit. The actual joint needs an agreed inspection method and acceptance criteria appropriate to the package, thermal design and application.
Can one stencil thickness serve every component on a mixed board?
Sometimes, but it must be demonstrated for the actual apertures and joints. A thickness suitable for large deposits can impair paste release from small apertures, while a thinner stencil may undersupply other joints. Package guidance, aperture design and printing trials determine whether a common thickness or a stepped approach is suitable.
Does a JTAG connector make the board fully testable?
No. Boundary-scan coverage depends on the implemented devices, their supported pins and instructions, the BSDL and netlist, and the ability to control and observe the intended connections. Programming or processor debugging through JTAG is not proof that all components or board functions have been tested.
When should an approved assembly process be reviewed again?
Review it when a change can affect soldering, inspection, electrical behavior or delivered configuration. Examples include an alternate package, stencil revision, paste change, PCB finish change, new firmware or a revised repair process. Agree the affected checks and approval owner before building changed production units.

