Skip to main content

4.4 Calibration and Adjustment

Calibration isand adjustment are mandatory final assembly steps that ensure the product's functional parameters meet the specified tolerance limits. This process corrects accumulated manufacturing variation (tolerance stack-up) in components and assemblies. Failure to perform traceable calibration results in reduced accuracy, performance degradation, and non-compliance with system specifications.

4.4.1 Calibration Flow and Process Mandate

Calibration must be integrated into the final refinementassembly thatflow, transformslinking adirectly functionalto programming and safety testing.

A) Flow Integration

Calibration must occur after the device intois one that delivers precision with confidence. By comparing each unit to a trusted reference under controlled conditions, subtle offsetsprogrammed and driftsfunctional, arebut corrected so performance lands safely within specification. This process ties measurement integrity tobefore the serialfinal number,safety with traceable records of “as-found” and “as-left” states that prove not only compliance but also stability over time. When performed with disciplined environments, reliable references, and straightforward math, calibration becomes less about chasing errors and more about building trust into every product shipped.

4.4.1 What this is (and isn’t)

Calibration compares the unit to a known reference and finds error.

Adjustment changes the unit (coefficients, trims, pot turns) to bring it into spec.

Do both with traceable references, in the right environment, and record as-found/as-left to the serial.gate:



4.4.2 Where it sits in the flow

Program (25.3) →– Functional smokeSmoke (25.1) →– Calibration/Adjustment →– Re-verify →– Safety (25.2) → Closeout.

If cal is part of FCT, the script must clearly mark cal steps vs verification steps.



4.4.3 Environment & measurement basics

  • Warm-up: DUTBoth the Unit Under Test (DUT) and the instruments tomust reach thermal steady state (typ.typically 10–10 – 30 min).minutes
  • ) before measurement begins to prevent thermal drift from contaminating readings. Ambient capture:temperature/RH logmust be captured and logged.

B) Instrumentation and Standards

The accuracy of the calibration process is entirely dependent on the test equipment used.

  • Temp/RHTraceability Mandate:—some specsAll applyprimary correction.instruments (DMMs, sources, chambers) must be under a strict, scheduled calibration cycle and traceable (NIST/ISO 17025).
  • Test Accuracy Ratio (TAR): referenceThe measurement uncertainty ≤of 25%the reference instrument must be significantly smaller than the tolerance of the DUT tolerance (4:1 rule;ratio 10:1or ideal)better).
  • Instruments in cal:Hookups: per 18.24-wire (Kelvin) method; station blocks run if a cert is out-of-date.
  • Stable hookups: Kelvinmandatory for low-ohms;ohm/low-voltage shieldedcritical measurements to eliminate lead resistance error. Shielded leads are required for sensitive mV/RF;RF avoid hand heat on sensors.measurements.


4.4.42 ReferencesThe &Closed-Loop fixturesAdjustment (pick the right yardsticks)Protocol

  • Electrical:

    Calibration precisionis DMM,a DCcontrolled, sources,closed-loop electronicsequence loads,executed voltage/currentvia standards,automated decadetest resistors,software, timebase/OCXO/GPSDO.

  • Sensors: temperature chambers/blocks, precision pressure/vacuum source, gas standards, light source panel.
  • RF/Comms: power meter, calibrated attenuators, spectrum/network analyzer (or golden radio + shield box).
  • Mechanicals: force/weight, linear scalescompensating for encoders,variance torquein testersspecific domains.

    A) Calibration Patterns (The Math)

    The mathematical model for motors.

  • Fixtures:compensation short,must repeatablebe paths;kept matingsimple connectors;and guided nests; no clip leads on production.


stable.

4.4.5 Common calibration patterns (use the lightest that works)

  • Offset-onlyOnly (Zero): (zero): measureMeasure at 0 (V or known null),null, store the offset.
  • Two-point gain/offset:Point: measureMeasure at low &and high points; solve the linear equation (y = m·m \cdot x + b.b) to find the gain (m) and offset (b).
  • Multi-point linearization:Point: 3–9For points;highly fit piecewise linear or poly; ensure monotonic LUT.
  • Ratiometric/bridge: calibrate excitation first; then sensor span.
  • Timing: trim RTC/clock via ppm offset against a reference (GPSDO or calibrated counter).
  • RF power/PA bias: step drive, measure output, build gain table vs channel/temp.

Tip: keep math simple and stable; prefer piecewise non-linear over high-order polynomials on small MCUs.


4.4.6 Domains & quick recipes

Voltage/Current rails

  1. Source known value(s). 2) Read DUT. 3) Compute gain/offset. 4) Write constants → CRC. 5) Re-read to confirm ≤ spec.

ADC/DAC

  • ADC: short/known referencesensors (e.g., 0 V & Vref/2)thermistors), computeuse offset/gain.
  • DAC: set codesthree to low/high,nine measure with DMM; solve m/b; limitpoints to codefit range.
a

Temperature

piecewise
  • Two-pointlinear (ice/ambient blockfunction or chamberlook-up steps): store slope/offset; log ambient.
  • If using thermistors, verify table index matches curve (Beta/R-T table)LUT).

B) Adjustment Discipline

  1. Pressure/FlowMeasure (As-Found):

    • Apply two or threea known points;stimulus. zeroThe atsystem vent; store span; runmust leak check if relevant.

    Timing/RTC

    • Count edges vs reference for ≥10 s; compute ppm trim; write; recheck ≤ spec ppm.

    RF

    • Set known channel/power; measure after attenuator; store cal factor; verify ACL R/EMC sanity in shield box if in scope.

    Displays/LEDs

    • Use light panel or sensor; set white/time-out levels; store per-color factors; eyeball uniformity.

    Motors/encoders

    • Home to hard stop or sensor; store zero offset; verify count per rev.



    4.4.7 Adjustment discipline (don’t chase noise)

    • Averageaverage N readings (e.g., 10–32) atto eachreduce point;noise. useLog the medianAs-Found if spiky.value.
    • GuardCalculate & Adjust: Compute the constant; electronically write it to the dedicated NVM partition.
    • Verify (As-Left): Re-measure to confirm the final As-Left value is within spec and meets the guard band: adjust to hit (center of spec,spec, not the edge.edge).
    • Write-protectLimit the Wrench: calFor physical adjustments (trimpots), the adjustment must be minimal. After completion, the trimpot must be sealed with an approved compound (e.g., varnish) and marked with a paint dot to prevent vibrational drift.

4.4.3 Data Protection and Traceability

The calibration constants must be permanently protected against field erasure and linked to the unit's identity.

A) Data Storage and Security

  • NVM Partitioning: Calibration data must reside in a separate NVM partition from the application firmware. This prevents field updates from accidentally wiping the constants.
  • Protection: The calibration region withmust be write-protected and include a CRC/Hash; recipecalculation. refusesThe system must refuse to run if the calibration region's CRC fails.
  • As-foundRegion vs as-leftFlags: alwaysCalibration captured—evendata if(constants, noregion adjustmentflags) wasmust required.
  • Limitmatch the wrench:physical label kit and safety test selection.

B) Logging and Audit Mandates

The complete set of data must be logged and linked to the unit's identity for the audit trail.

  • Mandatory Record: forThe physicalMES trims/pots,log applymust be bound to the unit torque limitsSN and mark with paint dot after.



4.4.8 Data & traceability (what the record must hold)

Attach toinclude: unit SN:

  • Recipe ID, firmware hash, cal algorithm version.
  • References/instruments withinstrument cert numbers & expiry.
  • Ambient Temp/RH, warm-upambient time.Temp/RH, As-Found error, As-Left value, and the calculated coefficients/LUT snapshot.
  • As-foundRetries: values/errors; as-left values/errors; coefficients/LUT snapshot.
  • CRC/hash of cal region; operator/fixture IDs; timestamp.
  • Any retries or anomalies (e.g., slow chamber rampramp) slow).must be captured in the final record.

Store

Final plots for long tests (e.g., temp sweep), at least for NPI and sample lots.



4.4.9 Throughput without cheating

  • Automate: scripted points, auto-calc, auto-write, auto-verify.
  • Ping-pong fixtures or parallel stations for long soaks/chamber steps.
  • Pre-cal modules (sensors, radios) at subassembly level when allowed; perform a short verify in box build.
  • Cache chamber setpoints and test short profiles (two-point instead of five) when the spec permits.


4.4.10 Acceptance cues (fast eyes)Checklist

AreaMandate

AcceptCriteria

RejectVerification Action

Traceability

ReferencesInstruments in cal; 4:1 TAR (Test Accuracy Ratio) verified.

In-cal,Log certsrecords valid

Expired/unknownthe accuracyinstrument Cert/ID and expiry date.

Process Flow

As-foundCalibration occurs after programming, before safety testing.

Warm-up timerRecorded, withinenforced; driftambient policy

Missing;Temp/RH wildly off with no ticketlogged.

Data Protection

As-leftFinal constants written to a separate, write-protected NVM partition.

WithinVerification specconfirms +the guard band

CRC/HashAt edge or still out of specthe cal region is correct.

Adjustment Discipline

As-FoundData storeand As-Left values are logged; adjustment targets the guard band center.

CoeffsFor written;physical trims, the CRCtrimpot OKis sealed

CRC fail;after defaultsfinal restored silentlyadjustment.

Measurement Integrity

Repeatability4-wire (Kelvin) method used for low-ohm/low-voltage critical measurements.

Re-runMeasurement givesinvolves sameaveraging withinN noise

readingsBig jumpsto betweeneliminate runsnoise.

SN Logging

LabelingAll calibration coefficients and status are linked to the unit's Serial Number (SN).

CalAudit date/revconfirms notedthe (ifdata required)

Norecord traceexists onfor unitcompliance that needs itdefense.


4.4.11 Common traps → smallest reliable fix

Trap

Symptom

Fix

Cal before warm-up

Drift after ship

Enforce warm-up timer; log ambient

One-point “cal” on nonzero slope

Still off at span

Use two-point (gain + offset)

High-order fits on noisy data

Wiggly LUT; fails corners

Use piecewise linear; median-average

Using bench gear “as-is”

Hidden offsets

Run zero/standard check each shift

Manual entry of IDs/constants

Typos in NVM

Script reads back; MES push, no typing

Overwriting customer keys/region

Field returns

Separate cal partition; protect IDs/keys

No as-found record

Can’t prove drift vs damage

Always log as-found before adjustment



4.4.12 Pocket checklists

Before

  • SKU/Variant scan → cal recipe & limits loaded
  • Instruments in cal; references warmed; ambient logged
  • DUT at temp; power stable; fixture latched

Run

  • Measure as-found; save
  • Execute points (two/multi) → compute → write coeffs
  • CRC of cal region OK; lock if required
  • Verify as-left meets spec with guard band

After

  • Results (coeffs, errors, plots) to MES by SN
  • Sticker/metadata updated (if required by customer/spec)
  • If fail: raise ticket (module swap or MRB), don’t over-tweak



With calibration done properly, devices leave the line consistent, defensible, and ready to perform in the field without surprises. The payoff is precision that lasts, backed by records strong enough to stand in audits and real-world use alike.


Back to top