A sensor board can pass every dry bench test and still drift in humid air. Learn how no-clean flux residue causes PCB leakage current and electrochemical migration, and how to set leakage budgets, run humidity tests, and validate cleaning, guard rings, and conformal coating.
A precision sensor board can look clean, calibrate correctly, and run for days on a dry engineering bench, then develop offset, slow drift, or intermittent saturation after installation in a humid enclosure. The missing circuit element is often not in the schematic: it is a moisture-dependent leakage path across the assembled PCB. When that path also supports electrochemical reactions, an initially reversible measurement error can become lasting corrosion or a metallic short.
For high-impedance electronics, “no-clean” is not an unconditional reliability guarantee. It describes a residue intended to remain after a qualified soldering process, not proof that every assembled board will meet every leakage requirement in every environment. The relevant question is whether the finished assembly preserves the required insulation under its actual temperature, humidity, contamination, and electrical-bias conditions. [2][5]
High-impedance interfaces deliberately handle very small currents. Examples include photodiode transimpedance amplifiers, electrochemical sensor inputs, capacitive sensing front ends, and buffered high-resistance voltage sources. A parasitic path that is irrelevant to a low-impedance digital output can dominate one of these measurements.
Under dry conditions, surface contamination may produce little measurable current. As moisture accumulates, ionic material associated with flux residues or other contamination can support conduction between conductors. Whether this becomes a practical problem depends on the residue chemistry and quantity, conductor spacing, voltage, humidity, temperature, and exposure history. [2][5]
This creates several diagnostic traps:
Think of cleanliness as an electrical design variable. A circuit requiring teraohm-scale isolation cannot be qualified by appearance alone.
Flux removes oxides from solderable surfaces and limits reoxidation while the joint is hot. The remaining material depends on the flux formulation and the thermal process it experienced. A no-clean formulation is intended to leave acceptable residue under qualified conditions; material that did not receive the intended thermal exposure can behave differently. [2]
That distinction matters around selectively soldered features, shielded areas, and hand-reworked components. A board assembled with a qualified solder paste can still acquire a different residue population when an operator adds liquid flux during repair. Qualification must therefore cover the assembly process, not just the paste designation. [2][5]
It is equally important not to label every visible no-clean residue as conductive or dangerous. Residue presence does not automatically establish a reliability defect. Suitability depends on the application and the completed process. [5]
Where moisture interacts with soluble ionic contamination, the resulting surface film can carry current. An engineer can initially model that behavior as an effective resistance between conductors, although the actual path may vary with time, humidity, voltage, and local chemistry.
A sensitive input does not need to be adjacent to a particularly high voltage to suffer. It only needs a voltage difference large enough, relative to the available insulation resistance, to inject more current than its error budget allows.
Several paths can exist simultaneously:
Flux is one possible contributor, not the only one. Fingerprints, salts, oils, and cleaning-agent residues also matter to assembly surface preparation. [6]
Surface leakage and electrochemical migration are related but not interchangeable. Leakage describes unwanted current. Electrochemical migration involves electrochemical reactions and transport of metal species through an electrolyte under bias, potentially producing conductive metallic growth between conductors.
A useful conceptual sequence is:
The supplied research identifies the combination of contamination, moisture, and electrical bias as a driver of electrochemical migration and corrosion. It does not establish a universal humidity, spacing, voltage, or exposure-time threshold at which every board will fail. [2][5]
Corrosion can also damage conductors or terminations without creating a short. Thus, the eventual failure may be excessive leakage, intermittent bridging, a persistent short, or increased resistance and an open connection.
Drying can remove a conductive moisture film; it cannot restore metal that corrosion has consumed. Conversely, a dendritic bridge can be fragile or intermittent, so the absence of a stable short after drying does not prove that migration never occurred.
“Very high impedance” is not a usable manufacturing requirement. The design needs a maximum acceptable parasitic current and a test method capable of resolving it.
The following calculations are illustrative design examples, not measured results from the research.
For a sensor represented by a Thevenin source resistance R_SOURCE, small unwanted current produces approximately:
V_ERROR ≈ I_LEAK × R_SOURCE
Assume:
The entire allowable unwanted current would be:
I_MAX = 100 µV / 10 MΩ = 10 pA
That is the total budget, not automatically the PCB allocation. Input bias current, protection-device leakage, connectors, cables, and other contributors must share it.
Suppose only 20 µV is allocated to PCB surface leakage:
I_PCB_MAX = 20 µV / 10 MΩ = 2 pA
For a neighboring conductor approximately 3.3 V away from the input potential:
R_SURFACE_MIN ≈ 3.3 V / 2 pA = 1.65 TΩ
A board can have no visible contamination and no ordinary continuity-test fault while missing this requirement badly. The circuit does not require a recognizable short to become inaccurate.
For a transimpedance amplifier, unwanted current entering the summing node produces approximately:
V_OUT_ERROR = I_LEAK × R_FEEDBACK
With an illustrative 100 MΩ feedback resistor, 10 pA of unwanted input current produces 1 mV of output error.
The model needs attention to topology. Leakage from a supply rail into the summing node appears as an input-current error. Leakage across the feedback resistor can instead alter the effective feedback impedance and therefore the gain. Different physical contamination locations can produce different symptoms in the same amplifier.
Build a table of sensitive nodes, neighboring voltages, allowed leakage, and operating conditions. Include startup, standby, and unpowered states when relevant: a neighboring signal that sits near the input during normal measurement may move to a rail during another mode.
Do not rely on nominal cancellation between leakage currents. Their magnitude and direction can change independently with moisture and bias.
A practical design requirement looks like this:
The assembled sensor-input region shall contribute no more than its allocated input-current error under the defined humidity, temperature, bias, and operating-mode sequence.
That statement connects circuit accuracy to manufacturing qualification far more effectively than “use no-clean flux” or “PCB shall be clean.”
The first diagnostic objective is to classify the electrical behavior without destroying evidence.
| Observation | Possible interpretation | Useful next step | What it does not prove |
|---|---|---|---|
| Offset rises with humidity and substantially recovers when dry | Moisture-dependent surface leakage | Repeat controlled humidity exposure while logging bias and error | That no corrosion or migration occurred |
| Error returns at similar conditions over repeated cycles | Persistent contamination or another moisture-sensitive path | Compare controlled cleaning and guarded test configurations | That flux is necessarily the source |
| Sudden current spikes or brief rail saturation during humid bias | Intermittent bridging, unstable surface conduction, or another intermittent fault | Capture time-resolved current and inspect the sensitive region | That a permanent short must be present |
| Low resistance remains after drying | Persistent bridge, damaged component, or another permanent path | Localize electrically and inspect microscopically | That drying alone can recover the assembly |
| Resistance increases or a connection becomes open | Possible corrosion or another interconnect failure | Inspect conductors and terminations; consider cross-sectioning | That all humidity failures are shorts |
| Cleaning improves performance | Evidence supporting a surface-contamination contribution | Repeat environmental testing and analyze removed contamination where practical | That the improvised cleaning method is production-ready |
These are troubleshooting interpretations, not standalone acceptance criteria. Multiple mechanisms can coexist.
Before cleaning or baking a field return:
Cleaning every returned board immediately can remove the evidence needed to distinguish flux contamination, environmental deposits, and physical damage.
A useful test is not simply “put the board in a humid chamber.” It should separate environmental sensitivity from assembly-process variation.
Consider a small comparison matrix:
Use multiple samples where practical, and avoid changing flux, layout, cleaning, and coating simultaneously. Otherwise, an improvement may be real but its cause remains ambiguous.
An illustrative exploratory sequence might use a fixed chamber temperature of 25 °C, with plateaus at 40%, 60%, and 85% relative humidity, followed by a controlled dry recovery. These are example engineering setpoints, not a prescribed IPC qualification profile or universal failure thresholds.
Choose the actual profile from the product’s operating environment and the applicable qualification requirements. Define dwell time using the test objective and observed stabilization, rather than assuming that chamber-air equilibrium means the assembly has equilibrated.
Bias the relevant conductors during exposure. Electrochemical migration depends on electrical bias as well as moisture and contamination, so an unpowered humidity soak does not reproduce every powered field condition. [5]
Record:
Separate noncondensing humidity testing from condensation testing. Dew-point crossings can create a different moisture condition from steady humid air. If the product encounters condensation, test it deliberately rather than allowing uncontrolled condensation to confound another experiment.
At picoampere-scale budgets, the test setup becomes part of the design problem. Cable insulation, feedthroughs, connectors, switching hardware, and the fixture itself can add humidity-sensitive leakage.
Use instrumentation and cabling appropriate to the expected current range, with guarding where supported. Run fixture blanks or equivalent controls so that chamber-induced leakage in the measurement system is not attributed to the PCB.
Avoid attaching a conventional voltage probe directly to a high-impedance node without evaluating its loading. Often, monitoring a buffered output and using a separate leakage test structure provides a cleaner experiment.
After humid operation, retain the electrical logging while returning the assembly to controlled dry conditions. Then repeat exposure if the sample remains suitable for testing.
The sequence reveals information that a single endpoint misses:
Recovery is evidence about the mechanism, not a blanket reliability pass.
No single inspection or cleanliness number establishes every aspect of reliability.
| Method | Main question it addresses | Important limitation |
|---|---|---|
| Magnified visual inspection | Are accessible residues, particles, or damage visible? | Cannot establish ionic cleanliness or electrical reliability |
| ROSE testing | Is extractable ionic contamination changing relative to the controlled process? | Bulk measurement can obscure local problems; not a reliability prediction |
| Ion chromatography | Which extractable ionic species are present? | Chemistry alone does not reproduce the circuit’s operating behavior |
| Surface insulation resistance testing | How does insulation behave under the defined humidity and bias conditions? | Relevance depends on specimen, geometry, process, and test conditions |
| Water contact angle | Is surface wetting consistent with a qualified reference before coating? | Does not measure ionic cleanliness, SIR, coating thickness, or adhesion |
| Product-level humid functional testing | Does the actual assembly meet its accuracy and leakage requirements? | Requires controlled conditions and a trustworthy measurement setup |
The research identifies IPC-TM-650 2.3.25D for resistivity of solvent extract, or ROSE, and IPC-TM-650 2.6.3.7 for surface insulation resistance. It also explicitly cautions that ROSE is a process-control tool, not a predictor of reliability. [1]
This distinction matters for high-impedance boards. A bulk extraction result may track whether a wash process has shifted while still failing to describe contamination concentrated between two critical input pins.
Water contact angle can add useful surface-preparation information when compared with a known-good, validated reference. However, the supplied source states that no IPC standard establishes a universal contact-angle limit for coating readiness. Treat it as a supplementary process indicator, not an electrical acceptance test. [1]
A guard reduces leakage into a sensitive conductor by surrounding it with a low-impedance conductor held close to the same voltage. It is not simply a ground ring.
For a voltage input, the guard may follow the buffered input potential. Around a transimpedance summing node, the appropriate guard potential may be the amplifier’s reference potential, depending on the topology and input common-mode voltage.
Consider another illustrative calculation. A 10 GΩ surface path with 3.3 V across it carries:
I = 3.3 V / 10 GΩ = 330 pA
If a correctly implemented guard reduces the voltage across the final path into the sensitive node to 10 mV, the current through the same assumed resistance becomes:
I = 10 mV / 10 GΩ = 1 pA
This demonstrates the electrical principle, not a guarantee for a real contaminated board. Guard discontinuities, connector geometry, package surfaces, guard-driver errors, and contamination that bypasses the guard still matter.
Check guard-driver stability and capacitive loading. Do not automatically connect a guard to a noisy or incorrectly referenced signal, and do not assume guarding eliminates the need to control contamination.
Practical layout measures include:
Low-standoff components deserve particular attention because cleaning requires access for wash chemistry, removal of contaminated liquid, rinsing where required, and drying. Accessible-surface brushing does not establish that trapped residues have been removed. [6]
Specify the actual soldering and rework processes, not merely “no-clean.”
A useful process-control package identifies:
No-clean residue suitability depends on qualified processing. Underheated residue, including residue associated with hand rework or selective soldering, may not behave like the material produced by the qualified reflow process. [2]
The production acceptance plan should therefore cover representative rework as well as pristine first-pass assemblies. Otherwise, the process most likely to introduce local excess flux may be absent from qualification.
Cleaning is not synonymous with applying isopropyl alcohol and waiting for it to evaporate. The process must remove the relevant contamination from the assembly, including difficult geometries, without creating material-compatibility problems or leaving unacceptable residues.
Some no-clean residues resist moisture and do not dissolve in plain water. Cleaning chemistry must match the flux; deionized water may serve as a rinse without being an effective primary cleaner for a non-water-soluble residue. [2][6]
A validated sequence addresses:
The supplied manufacturing guidance emphasizes evaluating chemistry, equipment parameters, rinsing, and drying as a controlled system. It also distinguishes visual cleanliness from ionic cleanliness and long-term reliability. [5][6]
Do not turn one successful solvent-cleaning experiment into a production instruction without further validation. A troubleshooting clean can establish that surface condition matters while remaining unsuitable for repeatable manufacturing.
Coating should be an additional environmental defense, not a method of burying an unresolved contamination problem. Unsuitable residues can remain trapped beneath a coating and affect adhesion, uniformity, and long-term protection. [5]
The research also reports that flux residues under coatings can contribute to blisters, leakage, and dendrites during humid exposure. [1]
A practical coating-readiness workflow is:
Coating-process guidance in the research specifically highlights cleaning, controlled storage after drying, masking, cure parameters, and first-article verification. It also warns that a coating can pass initial UV inspection yet later delaminate if surface preparation was inadequate. [4]
Keep standards and acceptance requirements in their proper roles. IPC-CC-830 addresses conformal-coating material qualification and performance; it should not be treated as a universal substitute for assembly-specific coverage, thickness, workmanship, and adhesion requirements. The research discusses coating preparation alongside J-STD-001, IPC-A-610, and IPC-TM-650 cleanliness and SIR methods, underscoring the need for multiple forms of evidence. [1][4]
The decisive result is not simply that the coating looks continuous. It is that the prepared and coated assembly meets the product’s electrical and environmental requirements.
Before releasing a high-impedance sensor board, require a short evidence package that connects design intent to production:
This closes the gap between “the flux is approved” and “this sensor assembly remains accurate in its deployment environment.”
Not always. No-clean describes a residue intended to remain after a qualified soldering process, not a guarantee that every assembly meets every leakage requirement. Residue presence alone does not prove a defect, but high-impedance sensor boards, boards with hand rework or selective soldering, and boards that will be conformal coated deserve evidence that the finished assembly preserves the required insulation under its real humidity and bias conditions. [2][5]
Moisture can combine with ionic contamination from flux residues, fingerprints, salts, or cleaning agents to form a surface conduction path. Under dry conditions that path may carry almost no current, but as humidity rises it can inject enough leakage into a high-impedance input to cause offset, slow drift, or saturation. [2][5][6]
Leakage describes unwanted current. Electrochemical migration involves electrochemical reactions and transport of metal species through an electrolyte under bias, which can grow conductive metallic dendrites between conductors. Contamination, moisture, and electrical bias together drive migration and corrosion, so a reversible leakage problem can become permanent damage. [2][5]
No. ROSE (IPC-TM-650 2.3.25D) is a process-control tool, not a predictor of reliability, and a bulk extraction result can hide contamination concentrated between two critical input pins. Surface insulation resistance testing (IPC-TM-650 2.6.3.7), ion chromatography, and product-level humid functional testing answer different questions. [1]
No. Coating is an additional environmental defense, not a way to bury contamination. Residues trapped under a coating can affect adhesion and have been reported to contribute to blisters, leakage, and dendrites during humid exposure, so clean, verify, and dry the assembly first, then retest the coated board under humidity and bias. [1][4][5]
A guard surrounds the sensitive conductor with a low-impedance conductor held close to the same voltage, so little voltage appears across the final leakage path. In the illustrative example, a 10 GΩ path with 3.3 V across it carries 330 pA, while the same path with only 10 mV across it carries about 1 pA. It does not remove the need to control contamination.
AI-driven electronics workflows, including platforms such as IntelCAD.ai that translate natural-language requirements into schematics and PCBs, can help make leakage constraints explicit early. Engineers can specify high-impedance nodes, allowable input-current error, guard requirements, adjacent-voltage restrictions, and assembly-access constraints before layout decisions become expensive to change.
The important boundary is verification: generated design intent is not evidence of ionic cleanliness, successful cleaning, or humidity reliability. The strongest workflow connects the schematic and layout to leakage calculations, manufacturing requirements, and measured environmental results, then uses those results to improve the next revision.
Design your next board by describing it.
IntelCAD.ai turns plain-language requirements into schematics and PCBs.
Start Your Design