Thick Film Resistance Measurements: Reject Open-Lead and Compliance-Limited Readings

Measurement purpose

Verify that the test chain recognizes connection and excitation faults before evaluating a printed resistor. The output is a tested validity rule and fault response, not an instrument accuracy specification.

Specimens and conditions

Circuit boundary

Use a de-energized, identified resistor or passive check artifact with accessible terminals. Exclude attached sources and resolve parallel circuit paths before evaluating lead faults.

Protected diagnostic artifact

Select a sacrificial or retained check piece with known contact geometry and a permitted stimulus envelope. Do not deliberately interrupt an energized production circuit or probe a fragile resistor body.

Equipment and records required

• Source and measurement chain: Identify force and sense terminals, supported lead-check functions, compliance behavior, error codes and actual voltage/current reporting for the specific instrument.
• Fault insertion and acquisition: Use an approved switching or disconnected-state method that changes one path at a time. Capture fresh readings and status together, preserving the channel and trigger identity.

Method sequence

1. Establish a valid baseline

   Verify safe connection and obtain a fresh measurement on the check artifact within the intended source limits.

   Record: Artifact identity, connection map, settings, raw voltage/current and status.

2. Challenge the measurement boundary

   With output disabled, introduce a documented single-path fault, restore the protected setup and execute the approved diagnostic cycle.

   Record: Fault identity, observed code, excitation behavior and validity outcome.

3. Restore and verify recovery

   Remove the introduced fault with output disabled and confirm both electrical recovery and a newly acquired data record.

   Record: Restoration observation and proof that a cached value was not reused.

Decision and uncertainty

Only measurements passing connection, excitation, freshness and status checks proceed to resistor tolerance comparison. A fault must never be converted silently into zero, infinity or a previous valid value.

Validity checks do not establish low uncertainty. A valid connection can still have thermal, contact-location and calibration errors requiring separate evaluation.

The test-method owner approves the challenge matrix and safe procedure; authorized quality personnel decide which historical results require remeasurement after a failed validity check.

Traceable outputs

Place validity ahead of tolerance

Consider a ceramic resistor network with a specified terminal-pair resistance. Product acceptance asks whether that resistance lies within the agreed interval. Validity asks an earlier question: did the instrument measure that terminal pair using the intended method? Combining the two into one red light makes troubleshooting difficult because a failed connection and a genuine out-of-tolerance resistor demand different actions.

Implement at least three outcomes: valid and within limits, valid and outside limits, and invalid measurement. An invalid result can stop a lot-release workflow, but it should not be counted as a measured high-resistance defect. Retain its reason so contact-related interruptions can be investigated without distorting the material or trimming process statistics.

Check the voltage required to deliver test current

A current-source measurement requires enough available voltage for the specimen and the complete force path. In a deliberately simplified example, one milliampere through a 2 kilohm specimen requires 2 volts. If the source limit is 1 volt, the requested current cannot be maintained across that resistance. Instrument behavior then depends on its operating mode; a displayed quotient formed with the requested current can be misleading.

Do not solve this by increasing the limit until the display looks stable. The printed element, connected components and fixture have their own permitted stimulus. First check whether the requested current is appropriate, whether force-contact resistance has increased, and whether the selected range supports the intended measurement. Record the limit event even if the instrument still returns a numeric value.

Determine what an open sense lead actually produces

A high-impedance sense input does not guarantee a recognizable open-circuit reading. A disconnected lead can float, couple to nearby wiring or interact with internal protection. Some instruments offer dedicated open-lead detection; others require a prescribed connection check. Use the documented behavior of the installed instrument, not a universal assumption that every open sense wire displays overload.

Place the sense probes on conductor pads identified by the drawing, not on glass-covered resistor surfaces. A probe touching the green protective layer can look correctly positioned in a wide photograph while making no intended electrical connection. Photographing the contact boundary during setup is useful, but electrical challenge tests are still needed to establish the detection rule.

Challenge one fault at a time

Use the matrix below as a method-development worksheet. The listed responses are questions to verify, not guaranteed instrument codes. Start with an output-disabled fixture, introduce the fault through an approved arrangement and keep stimulus within the diagnostic artifact's limits. Never improvise live lead removal merely to reproduce a software error.

Prevent yesterday's valid number becoming today's result

Electrical fault detection is incomplete if the software can attach a stale reading to a new part. Assign an acquisition identifier before triggering, then require the returned result to belong to that trigger and channel. Clear or explicitly manage instrument buffers according to the supported interface rather than assuming that reading the buffer initiates a new measurement.

A useful software test follows one valid result with a forced timeout. The second part must remain invalid even when the display or application memory still contains the first number. Preserve the timeout event and sample identity. A successful recovery should require a new acquisition, not merely a retry of the database write.

Separate a genuine high value from a broken measurement path

After the station correctly detects deliberately introduced faults, a high resistance on a real sample still requires confirmation at its specified terminals. Recheck contact placement and compare a permitted alternate range or measurement route. A high reading that remains stable after controlled remounting is different evidence from a value that disappears whenever a sense probe moves.

This process is not a license to repeat until a passing result appears. Keep the initial observation and each intervention. If a new fixture produces a different value, explain the changed boundary before choosing which result controls disposition. A physically cracked trace, a protected pad and a disconnected cable may all create similar first symptoms but have different product implications.

Store status without hiding the original response

Keep instrument codes in their original representation alongside the translated validity reason. Numeric overflow sentinels, not-a-number values and empty responses must not pass through normal resistance arithmetic. If an export requires a numeric column, leave the resistance field empty for invalid observations and use a separate status column rather than inserting an arbitrary large resistance.

When computing a sample mean, state which observations were valid and why others were excluded. Report invalid-cycle frequency separately. A fixture producing many invalid cycles can be unsuitable for production even if the few remaining resistor values look precise. This distinction makes the reliability of the test process visible without inventing defects in the printed parts.

Revalidate after a measurement-path change

Repeat the relevant fault challenges after replacing a scanner card, changing firmware, moving sense contacts or revising acquisition software. A connector change can alter both the electrical fault response and which software channel reports it. Preserve the earlier matrix so differences are investigated instead of silently accepted.

For a customer drawing review, provide the four-terminal contact map, allowed electrical stimulus and required invalid-data response. This allows the measurement method to be discussed separately from the resistor tolerance. Successful fault recognition establishes a trustworthy data boundary; material selection, trimming stability and final installed behavior remain separate engineering decisions.