Sequence Lab

Sequence thinking

Every appliance cycle is a state machine: enter a stage, energise specific loads, wait for a condition or a timer, then advance. A fault is almost always a machine stuck at a stage boundary, waiting for a condition that will never arrive, or advancing past a stage that did not really complete.

That reframes diagnosis. Instead of asking 'what part is broken', you ask 'which stage did it reach, and what condition does it need to leave that stage'. The answer names a small set of components immediately.

• Stuck at a stage = a condition is never satisfied (input, load, or sensing)
• Skipped a stage = the machine believed a condition was already met
• Terminated early = a fail-safe timer or a fault threshold intervened

Duration is a measurement

Stage durations carry as much information as voltages. A drain stage that runs its full timeout instead of ending on a dry pressure reading tells you the sump is not emptying. An ice machine freeze cycle that runs long tells you heat is not leaving the plate. A dryer that heats for the full cycle without reaching the termination temperature tells you airflow, not heat.

Time the stages against the manufacturer's chart. When you cannot get a chart, time them against a known-good identical machine or against the machine's own history if the controller logs it.

Reading a deviation

When a stage misbehaves, three questions place the fault: was the load actually energised, did the physical result occur, and did the sensing report it. Energised but no result is a mechanical or hydraulic fault. Result but no report is a sensing fault. Not energised is a control or interlock fault.

Those three questions apply to every stage of every machine, which is what makes sequence reading transferable across equipment you have never seen before.

Write the expected sequence down before you open the machine. Diagnosing against a written sequence is dramatically faster than diagnosing against memory.

Why sequence knowledge decides the call

Most calls are lost in the first ten minutes, not at the meter. A technician who does not know what the machine should be doing at a given moment has nothing to compare the observed behaviour against, so every reading is interpreted optimistically. A technician who knows the expected sequence sees the departure immediately, and the departure is the diagnosis in almost every case.

Sequence knowledge is also the part of the craft that transfers. Components change with every platform generation; the structure of a cycle does not. Enter a stage, energise loads, wait for a condition, advance. As controls become more integrated and more software-driven, the physical parts get harder to probe and the sequence becomes the most reliable window into what the machine believes about itself.

There is a commercial argument as well. Comparing expected against actual gives you a defensible statement — 'the machine reached stage four and never satisfied the condition to leave it' — that survives scrutiny from a customer, a supervisor, or a warranty administrator far better than 'it seemed like the pump'.

• The deviation point names a short list of components; the symptom alone does not
• Sequence reasoning transfers to equipment you have never seen
• A recorded deviation is defensible; an impression is not

How controls decide to advance

Stages advance on one of three rules, and knowing which rule applies changes what you test. Time-based stages advance when a timer expires regardless of result — a fixed tumble period, a fixed pre-chill. Condition-based stages advance only when a feedback device reports the required state — level reached, temperature reached, harvest switch made. Hybrid stages advance on condition with a timeout backstop, which is by far the most common arrangement on modern equipment.

The failure signatures follow directly. A time-based stage cannot stall; it can only produce a poor result. A condition-based stage can stall forever. A hybrid stage ends at its timeout and typically posts a fault, which is why so many stored codes are literally timeouts wearing a component's name.

Before you test anything, decide which rule governs the stage that failed. If it is condition-based, your first question is what feedback device the control is waiting on and whether the physical condition actually occurred. If it is time-based, your first question is whether the work got done in the allotted time.

• Time-based: cannot stall, can under-deliver
• Condition-based: stalls when feedback never arrives
• Hybrid: ends on timeout and logs a fault that names a condition, not a part

A timeout code tells you the condition was not met in time. It does not tell you whether the physical result failed or the reporting failed. Those are different repairs.

Observing and documenting actual behaviour

Disciplined observation beats fast guessing, and it is a written activity. Before starting a cycle, write the expected stages down the left of a page with expected duration and the condition that ends each stage. During the cycle, fill in the right side: time entered, what you heard or measured energised, elapsed duration, and whether the condition that should end the stage actually occurred.

Run the machine in its normal cycle first where the fault allows it, because service and test modes often bypass the very permissive that is failing. Use service mode second, to command individual outputs once the normal run has told you where to look.

Three questions apply at every stage and they are the whole method: was the load actually energised, did the physical result occur, and did the control receive the report. Energised with no result is mechanical or hydraulic. Result with no report is sensing or wiring. Not energised is control, permissive, or supply.

• Expected sequence written down before the cycle starts
• Normal cycle first, service mode second
• Energised / result / reported — asked at every stage, recorded every time

The five deviations and what each means

Never advances past a stage means a condition-based gate is unsatisfied and there is no timeout, or the timeout is long. Stalls mid-cycle with a fault means the hybrid timeout expired — look at the condition named by the code, physically. Repeats a stage means the control re-entered it because a downstream check failed or a retry counter is running; the interesting question is what the control saw that made it retry. Skips a stage means the control believed the condition was already satisfied, which usually points to a sensor reading a plausible but wrong value, or a jumpered or shorted input. Advances too early is the same failure with a different result — the report arrived before the work was done.

Two of these are commonly misread. A repeated stage is frequently reported by customers as 'it runs forever', and a skipped stage is reported as 'it does not do anything at all' when in fact the machine ran the whole cycle in six minutes. Ask for the duration of the cycle, not just the outcome.

• Stall or timeout: the condition is unmet — test the condition, not the code's part
• Repeat: something failed a downstream check; find what the retry is reacting to
• Skip or early advance: a report arrived that was not true

Machine sequences

Failure modes and what confirms them

Test procedures

Manufacturer differences

Safety and professional boundaries

• Never jumper or defeat an interlock — door lock, airflow proving, high-pressure or thermal limit — to make a sequence advance for observation.
• Observing live cycles means live terminals. Where a panel must be removed, meter first, keep one hand clear, and never reach across an energised board.
• On commercial refrigeration, do not reset a safety-limit trip before you have read the trip history; the reset erases the evidence of which stage kept failing.
• Rooftop and commercial units carry external interlocks and remote start capability. Lock out and tag out before working inside a cabinet that a building control can energise.
• If a sequence cannot be observed safely at the required stage, document what you could observe and escalate rather than improvising access.

Expected readings and what they mean

Field scenarios

Knowledge check