AC Performance Analyzer & End-of-Line Test Software
An AC performance analyzer for the end-of-line station: a multi-condition sequence chosen for coverage inside takt, capability tracking that warns before the first failing unit, and a signed record per unit.
Schematic
The performance test bench
How capacity is derived from measured air state and measured mass flow.
End-of-line AC testing designed around where failures actually occur
MileSoft AC Performance Test runs a controller-orchestrated test sequence on an instrumented fixture at the end of automotive and appliance air-conditioning lines. The sequence covers refrigerant charge by mass, compressor duty across the operating envelope, evaporator and condenser differential, a sensor sweep and an acoustic signature, engineered to fit the line's actual takt.
Acceptance bands and sequences are held per part number and changed without a station rebuild, and the station validates which configuration it loaded before each cycle. Every unit gets a signed report of all measured values, including its margin to each limit, linked to the unit identity in the traceability record and exportable to customer formats.
Capability is tracked per shift and per line, so a drifting process raises an alarm while units are still passing. A failing unit triggers an andon alert and opens a non-conformance workflow that holds the line. What no end-of-line test can do is catch a part that conforms today and degrades later — the station records the data needed to address that, but the acceptance decision belongs to engineering.
End-of-line AC testing
Three failure classes, and which an end-of-line test can actually reach
You test for the failure modes you anticipated. The warranty data three quarters later tells you which ones you missed.
That is not a criticism of testing; it is a structural property of it. A functional test measures a unit under conditions somebody chose, and a unit that works under those conditions is a unit that works under those conditions. The choice of conditions is the design problem.
| Class | Example | What reaches it |
|---|---|---|
| Present, inside the tested envelope | Undercharge; a dead sensor; a leak | Testing at the right conditions |
| Present, outside the tested envelope | Noise under sustained high load | Broader coverage, at a takt cost |
| Not yet present | A sensor in spec today, out in six months | Nothing in a test — only margin-based acceptance |
The third row is stated first rather than buried in limitations, because a page about end-of-line testing that does not say it up front is selling the wrong instrument.
AC performance analyzer
What an AC performance analyzer should measure, and what each condition costs in takt
Five conditions, each reaching a distinct physical failure mode.
| Condition | Failure it reaches | Takt cost |
|---|---|---|
| Refrigerant charge by mass | Under- or overcharge, and leaks | Low — a settled measurement |
| Compressor duty across the envelope | Behaviour away from the rating point | High — dwell at several conditions |
| Evaporator and condenser differential | Capacity shortfall, restricted flow | Moderate — needs thermal settling |
| Sensor sweep | Dead or out-of-range thermistors and transducers | Low — electrical, fast |
| Acoustic signature | Noise complaints, mechanical interference | Low to moderate — a short capture |
Compressor duty across the envelope is the expensive one and the one a catalogue-point test leaves out: behaviour at the rating point is a single sample of a duty envelope, and warranty failures concentrate away from it. Acoustic capture is unusual at end of line because its failure class is defined by customer perception rather than a drawing, so its band comes from complaint data. Sequences and acceptance bands are held per part number, and the station validates which configuration it loaded before each cycle — otherwise a variant gets tested against another variant's limits.
Coverage under takt
Choosing test conditions from the warranty file, inside a takt budget
Warranty data is normally the scoreboard. Here it is the input.
Every warranty claim is an observation of where in the operating envelope a failure occurred, so a year of claims is a sample of where failures live. Locate last year's claims in the envelope and ask which of your current conditions would have exercised each one — the answer is your coverage figure, and no new instrument is needed.
Worked example: coverage from 240 claims
96 claims cluster at sustained high load, 62 at cold-start cycling, 44 at or near the rating point, 38 scattered. A sequence testing only the rating point covers about 54 of 240 — coverage 0.23. Adding sustained high load takes it to 0.63; adding cold-start cycling to 0.88. The rating point is not useless; it was simply the only condition, and the failures do not live there.
Worked example: what fits inside 74 seconds
Takt is 96 s with 22 s of handling. Ranked by marginal coverage per second, the greedy choice is acoustic capture (12 s), sustained high load (46 s), sensor sweep (6 s) and charge by mass (8 s): 72 s for 0.73 coverage. Cold-start cycling and the evaporator-condenser differential do not fit.
The uncomfortable result: acoustic capture, the condition most likely to be dismissed as a nicety, is by far the best coverage per second, and the differential test that feels like the core of an AC test is the worst value in the set. Check the coverage estimates before acting — the point of the method is that its assumptions are explicit enough to argue about.
Cp/Cpk as early warning
What a capability drift alarm buys, and the trade nobody states
A pass rate says nothing about margin. Capability does — and margin is time.
lead time to first failure = remaining margin beyond 3σ ÷ drift rate
Capability is computed per shift and per line. A drift on one line averaged with three stable ones produces an index that moves too slowly to warn about anything.
Worked example: a worn filling valve
Charge has an upper limit of 640 g; the mean is 592 g, σ is 9 g, and the mean rises 1.4 g a shift. Cpk is 1.78, and the first unit crosses the limit in 15 shifts. An alarm at Cpk 1.33 fires in 8.6 shifts, leaving 6.4 shifts of warning; an alarm at 1.67 fires in 2, leaving 13.
Now tighten the limit to 620 g to improve outgoing quality. The first failure is less than one shift away, and the 1.33 alarm has already been breached. A 3% tighter limit turned a fifteen-shift problem into an immediate one.
Tighter limits improve outgoing quality and shorten the warning; wider limits lengthen the warning and admit worse units. Neither is wrong, but it is a choice a station should expose. The option that dominates both is reducing σ — which widens margin at unchanged limits.
Capability also makes the line-hold policy smarter. A failing unit raises an andon alert and opens a non-conformance workflow that holds the line. A failure while the distribution is centred and stable is probably isolated; a failure after the mean has walked for two shifts is probably the first of many. Without a capability signal, every failure has to be treated as systemic.
Measurement integrity
The test station is a gauge before it is a test
A pass verdict is a number compared against a limit, and it inherits that number's uncertainty.
ISO 5725 separates trueness — how close a mean sits to a reference — from precision, the scatter of repeated measurements, and AIAG measurement systems analysis quantifies both for a specific gauge. For a temperature differential at calibrated airflow, the airflow calibration is itself an uncertainty source governed by ISO 5801 and ANSI/AMCA 210: a differential measured at an airflow known to 5% carries that 5%.
The station manages calibration of its own sensors. That matters because a drifted thermocouple does not produce obviously wrong numbers; it produces plausible ones, and the population of results invalidated is bounded by the calibration interval, not by anything the station notices.
Beyond the test
The failures no test sequence reaches, and what can be done about them
A sensor within specification today and outside it in six months conforms at test time.
accept only if measured value + expected in-service degradation + measurement allowance stays inside the limit
Margin-based acceptance is a different intervention from broader coverage: coverage moves testing toward where failures happen; margin-based acceptance rejects units that pass today and would fail later. It is also more expensive than it looks — rejecting conforming units needs an engineering position that the specification is insufficient for the warranty period, and produces scrap a customer may not accept as chargeable. It is a design and contractual decision the station can enforce but cannot make.
What the station can contribute is the data. Each unit's margin — its distance from the limit — is on its signed report. Join those margins to warranty returns through the traceability record and a plant can compute, for the first time, whether its acceptance limits are the right ones.
Evaluating end-of-line test software
Eight questions to put to any end-of-line AC test station
Each is answerable by demonstration — including by us.
| # | Dimension | Ask the vendor to demonstrate |
|---|---|---|
| D1 | Stated coverage | Which regions of the operating envelope does the sequence exercise, and which does it not? |
| D2 | Sequence provenance | Was the sequence derived from design intent, or from warranty claims? |
| D3 | Marginal value | What would you add if takt allowed, and why that condition? |
| D4 | Stated uncertainty | What is the measurement uncertainty on each key value, and how was it established? |
| D5 | Capability, not yield | Does the station alarm on a distribution shift while all units still pass? |
| D6 | Published lead time | How long, in units or hours, between that alarm and the first expected failure? |
| D7 | Configuration validated | Does the station verify which acceptance band it loaded before each cycle? |
| D8 | Margin recorded | Is each unit's distance from the limit on its report, and joinable to warranty returns? |
D2 predicts whether the station will still catch the right failures in three years. D8 costs nothing and enables the only analysis that can say whether the limits are right.
On evidence: no published outcome is attributed to this station. It appears in a Tier-1 automotive deployment as a downstream cross-check, but that deployment's reported results belong to the kitting and alignment stations and are not claimed here. MileSoft's return model assumes a 75% reduction in field escapes, and that figure is the assumption to test against your own warranty file — a plant whose escapes are mostly in-service degradation should expect substantially less, because coverage cannot reach that class.
Standards referenced: ISO 22514-1 and -2, ISO 5725-1 and -2, AIAG MSA, IATF 16949, ISO 5801, ANSI/AMCA 210, Regulation (EU) 2024/573, IEC 62264-3. For a practitioner's take on end-of-line testing see AC performance analyzer, the end-of-line test station; derivations are in the AC performance test method paper.
Features
Key Features
Built-in capabilities that set our software apart
Multi-condition sequence
refrigerant charge by mass, compressor duty across the envelope, evaporator and condenser differential, sensor sweep, acoustic signature
Engineered for takt
conditions chosen for coverage per second within the line's cycle time
Per-part-number bands
acceptance limits and sequences configured per variant, with the loaded configuration validated each cycle
Capability drift alarms
Cp/Cpk per shift and per line on key measurements, while units still pass
Signed per-unit report
every measured value and its margin to limit, exportable to customer formats
Automatic escalation
andon alert and non-conformance workflow on failure, holding the line
Sensor calibration workflow
calibration of the station's own instruments managed and recorded
Benefits
Key Benefits
What your team gains from adopting our platform
Testimonials
What Our Clients Say
Trusted by industry leaders across manufacturing, HVAC, and logistics.
"Our “Part-Process Matrix” as a concept and Software received 1st runner-up award in MVML Innovation fair 2014. As a customer we appreciate your support and efforts to deliver it on time and most importantly as per customer requirement."
FAQ
AC Performance Analyzer & End-of-Line Test Software — FAQs
Answers to the most common questions about this product. Need more? Book a demo or reach out via Contact.
Why are standard functional tests not sufficient?
What is envelope coverage under a takt budget?
Why report Cp and Cpk rather than pass rates?
Can one station handle multiple variants?
How does it integrate with ERP and traceability?
Is this an AC performance analyzer?
Explore More
Related Products
Complementary tools teams often pair with this solution
Related Reading
Go deeper
Explore the long-form story behind this product.
You Test the Failures You Anticipated: Envelope Coverage Under a Takt Budget, and Capability as an Early Warning
Architecture, Computational Methods, and a Modelled Scenario for MileSoft AC Performance Test
Read the paperRead the full AC Performance Analyzer & End-of-Line Test Software deep-dive
Read articleSee how MileSoft delivers in Pokayoke
Read case studySee your savings in under a minute
Estimate the time and cost this tool could give back — free, instant, no sign-up.
Bring last year's warranty claims and see your coverage
Locate your claims in the operating envelope with us. We will show what your current sequence reaches, what fits inside your takt, and how much warning your capability alarms give.