Common Faults by Symptom, and the First Check for Each

Wednesday, 11:10, at the head office of Northgate Retail. Tom has six hardware tickets in the queue. The subjects, in order, read "dead", "slow", "hot", "battery", "screen" and "keyboard". That is either six different faults or one word each from six people who were in a hurry. Leanne, the store support lead, is standing beside him with the seventh, a laptop that "sometimes just turns off". She has carried it up two floors because she suspects, correctly, that "sometimes" is not a thing the ticket form has a box for. Tom has the diagnostic ladder on the wall and intends to climb it seven times. What he would like, before he starts, is to know which rung each of those words usually lands on. That way he can guess well before he checks properly.

A symptom is what the user sees, hears or feels. A fault is the thing that is actually wrong. The two are related in the way a cough is related to a diagnosis. That is to say loosely and with a lot of history behind each pairing. The first check for a symptom is the cheapest test that tells the two apart. Knowing it saves the ten minutes at the start of each ticket where a technician decides where to begin. This article is part of our Hardware Troubleshooting Workflows series; the ladder itself is in The Diagnostic Order. This article is the field guide that says, for each common symptom, what it usually turns out to be and which rung to look at first.

It covers why symptoms are not faults, the machine that will not power on, and slow. Next come overheating, fan noise and the battery, and the screen, the keyboard and the ports. It covers intermittent faults and how to catch one, and when it is really the dock. It ends with a short war story and the symptom-to-first-check table to copy, in that order.

Symptoms Are Not Faults

The user reports a symptom, the desk fixes a fault, and the gap between them is where diagnosis lives. "The screen is dead" is reported for a machine that is off, or a machine that is on with the display blanked. It is reported for a machine whose backlight has failed, or a machine whose external monitor cable has worked loose. It is also reported for a machine that is perfectly fine but sleeping. Five faults, one symptom, five different first checks. A desk that treats the symptom as the fault orders five screens, and four of them go in a drawer. A desk that treats the symptom as a starting point asks what else is true and narrows the five to one before touching anything.

Each symptom has a usual cause, and "usual" is a statement about the desk's own history rather than a law. On most fleets I have run, "will not power on" was the charger or the cable more often than anything else. On those fleets, "slow" was the disk or a full memory more than it was the processor. "Hot" was dust and a blocked vent more than it was a failing fan. Your fleet's history may differ. The way to know is to read your hardware tickets by symptom and by what fixed them. That is the model-level analysis in Device and Model Hotspots. The first checks below are the ones that hold across most fleets. Replace them with your own as your tickets teach you.

The first check is always a cheap one, and that is not a coincidence. The point of knowing the usual cause is to put the right cheap test first, not to skip the cheap tests in favour of the usual cause. "It is usually the charger" means "try a known-good charger first", not "order a charger". The ladder still applies; the field guide only tells you where on each rung to look. (A guide that let you skip rungs would not be a guide. It would be a hunch with a cover sheet.)

Will Not Power On

A machine that will not power on is usually not getting power. The first check is to prove the power path from the wall to the board. Check for a socket known to be live, a known-good charger of the right rating, and both ends of the cable seated. Check that the charging light on the machine is lit. Then use a long press on the power button to clear whatever state the machine was left in. On a laptop, try it with the charger connected and the battery given twenty minutes to take a charge. A battery drained completely flat over a weekend can leave the machine unresponsive until it has a little charge in it. On a desktop, check the switch on the back of the power supply, the power lead at both ends, and the front-panel light. Most of these tickets end here, which is why this check is first.

If the power light comes on but nothing else happens, the symptom has changed to "powers on but does not start". The first check then becomes the display: is the fan running? Is there any sound of a drive or a chime? Does an external screen show anything? Does the keyboard's own light come on? A machine that powers on with no display is usually a display problem or a sleep state it cannot wake from. The long press followed by a cold start resolves many of them. A machine that powers on, shows the maker's logo and stops is usually a storage or firmware problem, which is a different rung. Write down which of the three you have, because "dead" covers all of them and the next technician needs to know which one.

The check that catches the embarrassing case is the one the user cannot see: the machine that is fine but asleep. The lid sensor or the power setting is putting it into a sleep from which the usual tap does not wake it. Hold the button long enough to force a full power-off, wait, and start cold. If the machine wakes, the fault is a setting or a sensor and the ticket is on rung three. In that case, the user's "it died" was accurate from where they were standing. Never tell a user the machine was just asleep in a tone that implies they should have known. They could not have known. That is why there is a desk.

Slow

"Slow" is the symptom that this series mostly hands to another one, because slowness is more often perception, software or network than hardware. The whole method of turning "it's slow" into a measurement belongs to The Slow Computer. That article decides whether the machine is slow at all, and against what. This article takes over only when the measurement has been made and points at the metal. There are three hardware causes worth knowing by name, because each has a cheap first check.

The first is the storage drive. Its first check is the drive's own health counters, which the operating system's built-in tools can read without opening anything. A drive that is beginning to fail slows the whole machine before it loses any data, because every read is retried. The user experiences it as a machine that takes four minutes to start and freezes for ten seconds at a time. The counters say so plainly, and a drive reporting errors is replaced under rung five without further argument. The signals that let a desk catch this before the user notices are the subject of Failing Hardware Signals. A drive check is the cheapest of them.

The second is memory, and its first check is how much is in use when the machine is slow, which the operating system's own task view shows. A machine with its memory full is swapping to disk. A machine swapping to disk is slow in a specific way: fine for a while after a restart, worse as the day goes on. It is worst with many things open. That is a capacity question rather than a fault. The answer is more memory if the machine takes it, or fewer things open if it does not. If neither applies, the answer is a machine from the next refresh. The third is heat, which slows a machine on purpose to protect itself, and that one has its own section below. None of the three is touched until the slowness article has established that the machine is slower than its peers rather than slower than the user would like.

Overheating, Fan Noise and the Battery

A machine that runs hot, or whose fan is loud, is usually not moving enough air. The first check is the vents: are they blocked by dust, or by the desk surface? Are they blocked by a laptop stand that covers them, or by a bag it has been running inside? Feel where the air comes out. A machine that is hot with a vent that blows nothing is a blocked or failed fan. A machine that is hot with a vent that blows hard is doing its job under a heavy load. The question is what the load is. Dust is the usual cause on machines more than a year old and on any machine that lives in a warehouse, a workshop or a store room. Clearing it is a rung-three job with the right tools or a rung-five job if the fan has to come out. A machine that overheats will also slow itself down to survive, which is why "slow in the afternoon" sometimes turns out to be "hot by the afternoon".

A fan that is suddenly loud with no change in heat is usually a fan on its way out, or a process that has run away. In the latter case, the process is working the processor at full tilt. The first check distinguishes them: look at what is running. A runaway process is rung three and the fix is to stop it and find out why it ran. A fan that is loud at idle with nothing running is rung five. A fan that rattles or grinds is a fan with a failing bearing. It is replaced before it stops, because a fan that stops takes the machine with it on a warm day.

The battery has three symptoms and one first check. "Does not last" is usually a battery that has aged. The operating system's battery health report, which most systems can produce, says in plain words what the battery can hold now against what it held new. A battery that holds well under half of its original charge is replaced. On many laptops that is a rung-five job and on some is a vendor job. "Does not charge" is usually the charger or the cable, so the known-good charger comes first, and only then the battery. And "turns off at a percentage" is usually a battery whose own gauge has drifted. A full discharge and a full charge sometimes corrects that drift and a replacement always does. A swelling battery, where the case or the touchpad is bulging, is not a ticket to troubleshoot. It is a machine to power off, take out of the user's hands and send away today. The desk should say so in its knowledge base in large letters.

Screen, Keyboard and Ports

A screen fault's first check is whether the picture exists at all, and an external monitor answers it in a minute. If the external screen shows a normal picture, the machine is fine and the fault is in the built-in panel, its cable, or its backlight. If the external screen shows the same corruption, the fault is in the graphics hardware or its driver. In that case, rung three comes before anything is replaced. A panel that is dim but readable in strong light is a backlight. A panel with lines or blocks that move when the lid is flexed is the cable or the hinge. A panel with a crack is a crack, and it is a chargeable repair under the rules in Swap, Repair or Retire. The flicker at a lid angle that this series has used as an example is almost always the cable. The cable passes through the hinge and the hinge is where cables go to die.

A keyboard's first check is an external keyboard. A touchpad's is an external mouse, for the same reason: they separate the input device from the machine behind it. If the external one works, the fault is the built-in device, which on a laptop is a rung-five part and on a desktop is a new keyboard from the drawer. If both fail the same way, the fault is a setting, a driver or a stuck key. In that case, the cheap checks start with looking for a key held down by debris. Turn off any accessibility setting that changes how keys behave, and try the machine before the operating system loads. A keyboard that works there proves the hardware and blames the software. A keyboard with a liquid spill is a keyboard with a short life ahead of it, however well it is dried. The ticket should say so.

A port's first check is a different device in the same port and the same device in a different port. Between them, these tell you whether it is the port, the device or the cable. A single port that works with nothing is usually the port and is a rung-five or vendor job depending on the machine. A port that works with everything except one device is the device or its cable. Every port failing at once is a driver or power setting, and that is rung three. The charging port deserves a special mention, because it is the one that takes the most mechanical abuse. It is also the one whose failure looks like "will not power on" or "does not charge". A charging port that only works with the cable held at an angle is a port on its way out. It goes to the vendor before it fails completely rather than after.

Intermittent Faults and How to Catch One

An intermittent fault is one that does not happen when you are looking at it, and it is the hardest kind. The ladder depends on reproducing the fault at each rung and an intermittent one refuses. The desk cannot swap parts for a fault it has not seen, and the user cannot describe a pattern they have not noticed. The first check is therefore not a test but a record: the user keeps a short diary of every occurrence for a week. The diary turns "sometimes" into a pattern that points at a rung.

The diary asks five things at each occurrence, and the template below is the whole of it. Date and time; what the machine was doing; what was plugged in; where the machine was, docked or loose, on mains or battery, warm or cold; and what the user did to recover. Five lines, written on paper or in a note, in the minute after it happens. After a week, the pattern is usually visible. If every occurrence is docked, it is the dock. If every occurrence is after an hour of video calls, it is heat. If every occurrence is on battery under a certain level, it is the battery. If every occurrence is on a Tuesday morning after the weekly update window, it is software. A diary with no pattern after two weeks is a candidate for a longer bench test or a swap, and the honest ticket says which.

INTERMITTENT FAULT DIARY — ONE LINE PER OCCURRENCE, FOR ONE WEEK
Ticket: ________   Device tag: ________   User: ______________

Date/time | What it did | What I was doing | Plugged in | Docked? Mains? Warm? | How I recovered
_________ | ___________ | ________________ | __________ | ____________________ | _______________
_________ | ___________ | ________________ | __________ | ____________________ | _______________
_________ | ___________ | ________________ | __________ | ____________________ | _______________
_________ | ___________ | ________________ | __________ | ____________________ | _______________

An intermittent fault that has been "fixed" will come back, sometimes, which is the nature of the thing, and the ticket should be ready for it. Resolve the ticket with the pattern and the fix written in the resolution note. Tell the user plainly that if it happens again they should reply to that ticket rather than open a new one. Keep the diary attached. Where a requester's reply reopens the resolved ticket, as it does in QuerySolv and in many tools, the second round starts with the whole first round in front of the technician. For an intermittent fault, that is worth more than any part. A fresh ticket for the same fault starts the diary again from nothing, and the user will not thank you for that twice.

When It Is Really the Dock

A surprising share of laptop faults are the dock, the dock's power supply or the cable between them. The first check for any symptom that happens only at the user's desk is to try the machine bare. Docks fail in ways that look like the laptop failing. The external screens go black, which looks like a graphics fault. The keyboard and mouse stop, which looks like the machine freezing. The network drops, which looks like a wireless fault. The laptop does not charge, which looks like a battery fault. In each case the laptop is fine, and the proof is two minutes long: unplug the dock, plug the charger straight in, and see whether the symptom survives. The dock and its cables have an article of their own, Docks, Cables and the Things That Break. The desk should have a known-good dock on the bench for the same reason it has a known-good charger.

The dock's power supply is the part that fails most and is noticed least. The dock still lights up on a failing supply and still passes a keyboard. Meanwhile, it quietly fails to charge the laptop or to drive the second screen. A laptop that is "slow to charge at my desk but fine at home" is a dock supply; a second monitor that "flickers when I plug the headset in" is a dock short of power. Try the known-good supply before the known-good dock, because it is cheaper and it is more often the cause. And when the dock is the fault, the ticket is a dock ticket, not a laptop ticket. The laptop's history should not carry that dock ticket. A laptop with three dock tickets against its name will be retired as a lemon for something that was never its fault.

A Short War Story: The Four Screens at Bluewater Bank

At Bluewater Bank, a technician called Rhys took four tickets in a month from the mortgage team, each reading "screen keeps going black". Rhys replaced four laptop screens over four weeks. Each time, the user was on a loaner for the four days the panel took to arrive. Each time, there was a confident resolution note. The fifth ticket came from the same bank of desks. Rhys, now suspicious, walked over with a known-good laptop and plugged it into the desk, where its screen went black within the hour. The four desks shared a run of docks on one power strip. The strip's supply was failing under load, and every dock on it had been dropping its video whenever the afternoon's calls began. Four panels, sixteen loaner days, roughly twenty technician hours and a month of the mortgage team's patience had gone on a power strip. That strip cost a quarter of an hour to replace. Rhys added "try it bare, at the user's desk" to the top of the symptom table that afternoon. The four replaced screens went into the spares drawer, where they were at least useful later. A fault that lives at the desk is not in the laptop, and the first check is to leave the laptop out of it.

The Symptom-to-First-Check Table

The table below is the field guide in one place. It shows the symptom as the user says it, and what it usually turns out to be on most fleets. It shows the first check that separates the likely causes, and the rung of the ladder where the fault usually lives once the first check has spoken. Pin it beside the diagnostic order card and replace the "usually" column with your own fleet's history as your tickets accumulate. The first-check column changes less, because it is the cheap test, and cheap tests do not go out of date. Nothing in the table replaces the ladder. It tells you where on the ladder to look first.

Tom's six tickets went the way the table suggested, more or less. "Dead" was a charger. "Slow" went to the slowness measurement and came back as a drive with errors. "Hot" was a vent full of store-room dust. "Battery" was an aged battery and a vendor job. "Screen" was a cable at the hinge. "Keyboard" was a biscuit crumb under the shift key, which the user confessed to with good grace. Leanne's "sometimes just turns off" got the diary. After six days the diary said "docked, afternoon, on a call" five times out of five. The dock's power supply went in the bin. Seven tickets, one laptop part, no screens ordered. Tom pinned the table beside the ladder, and the next technician started where he had finished.

Frequently Asked Questions