The Level 2 method: logs, the subsystem map, test scans and phantoms
Find the first causal error in a CT log, place the failure in the scan sequence, choose a test scan, phantom or swap that separates two suspects, and know what you are and are not allowed to change.
Where Level 1 ends
At Level 1 you kept the scanner safe, warmed, calibrated and clean, and you knew where your scope ended. Level 2 begins when the scanner fails anyway: an error stops the scan, an artefact appears that the air calibration does not cure, the gantry will not rotate, the table will not move. Your job now is to say which subsystem has failed, prove it with a test, and either repair it within your authorisation or hand over a clear, evidence-based request to whoever is authorised.
The expensive mistake at this level is not ignorance. It is replacing the module that reported the error. A CT scanner is a chain: mains, power distribution, the slip ring, the generator, the tube, the collimator, the detector and DAS, the data link, the reconstruction computer, the host and the network. A fault anywhere upstream is often reported by a block downstream that simply noticed something was missing. Change that block and you have spent money, lost a day and kept the fault.
Read the log for the first error, not the loudest
Every modern CT keeps a system log with a time stamp, a severity, the subsystem or board that raised the event and a code or text. Manufacturers grade severity differently, but the idea is common. The NeuViz 16 troubleshooting manual, for example, separates events that bring the system down, errors that put a parameter outside specification while scanning may continue, warnings that do not stop work, and plain information. Learn the grades on the machines you look after, because a screen full of warnings can hide the one error that matters.
Work backwards in time from the moment the scan stopped. The first event in a burst is usually closest to the cause; what follows is consequence. A generator that loses its supply will complain, then the DAS will report missing data, then reconstruction will report an incomplete scan, then the host will report an aborted exam. The last line on the screen is the furthest from the cause.
Before trusting any time stamp, check the system clock. A flat backup battery in a real-time clock or timekeeper can reset the date, scramble the order of events and confuse functions that depend on elapsed time, such as tube cooling calculations. Check, too, how your system numbers its parts. Board slots, detector modules and channels may be counted from zero or one, from different ends, or in a format that must be converted; read the manufacturer's explanation before you order a module by number.
- Export or photograph the log before rebooting; some systems overwrite it.
- Note the exam, protocol, kV, mA, rotation time and gantry angle at the moment of failure.
- Ask the operator what they saw and heard, and whether anything changed: a power cut, a move, a new protocol, a software update, cleaning.
The scan sequence is your subsystem map
A scan is a fixed sequence of stages, and each stage needs a known set of subsystems. Placing the failure in the sequence narrows the suspects before you open a single cover. Several service manuals organise their troubleshooting this way; the NeuViz 16 manual walks from initial state through scan start and hardware preparation to X-ray on, the data path and finally image quality.
In practice: if the system does not complete initialisation, think power distribution, communication between the stationary and rotating parts, controllers and firmware. If it fails during preparation, ask which unit did not report ready: table, gantry rotation, tilt, collimator, generator, detector. If it fails at X-ray on, think generator, tube, rotor start and high-voltage arcs. If exposure completes but data are lost, think data link, slip ring and the receiving board. If everything completes but the image is wrong, you are in calibration, detector, geometry or reconstruction territory.
The same question from Level 1 still helps on images: is the problem before the patient (tube, filter, collimator: usually affects the whole image or a whole projection) or after the patient (detector, DAS, data: usually local, regular and repeatable)?
Reproduce it with the cheapest test
A fault you cannot reproduce cannot be proven fixed. Look for the smallest test that makes it happen. Can the gantry rotate with no exposure? Does a scout complete but a helical scan fail? Does it fail only at high kV, only at high mA, only at fast rotation, only after twenty minutes of work, only at a particular gantry angle? Every yes or no removes a group of suspects.
Manufacturer service tools usually offer rotation tests, generator and tube tests, detector and DAS tests, data-link tests and table tests. Use only the tools available at your level of authorisation. Any test that produces X-rays is a radiation exposure: the room is cleared, the door closed, the warning lights working, and you follow the local radiation rules and Level 1 lesson 3 in full.
For image problems, scan a phantom, not a patient. A water phantom, scanned with the same protocol on which the problem was seen, tells you whether the artefact belongs to the scanner or to the patient and protocol. Compare with the baseline you recorded at acceptance or after the last major repair. If the artefact disappears on the phantom, look at the protocol, the patient and positioning before you look at hardware.
Swap, isolate, measure - one change at a time
Swapping identical parts is a powerful test on a CT scanner, because the machine is full of repeated units: detector modules, DAS boards, power supplies, fans, brushes. Moving a suspect module to another position and seeing whether the error follows it is often faster than any measurement. Do it with discipline.
Change one thing at a time and write down what you changed. Swap only parts with the same part number and a revision the manufacturer allows together. Return each part to its original place if the test proves nothing. Think before swapping a power supply into a circuit that may have a short: the fault that killed the first part can kill the second. And remember that many replacements require a calibration before the system is usable again. The NeuViz 128 calibration manual, for example, lists for each replaced component the calibrations that must follow; other manufacturers use similar tables. A swap test that leaves the scanner uncalibrated is not finished.
What you may change, and what you may not
Your authority on a CT scanner is set by your employer, your contract, the manufacturer's training and the law of the country. In Russia, technical service of medical equipment for other organisations is a licensed activity; in Kazakhstan, work with devices that generate ionising radiation needs a licence from the atomic energy regulator; and in both countries radiation safety rules require trained, admitted personnel. Check the current rules that apply to you before you work.
Within that authority the usual safe territory is: reading logs and running tests your access level offers, cleaning, inspecting and re-seating connectors, user and service calibrations you are trained for, and replacing modules by the manufacturer's procedure followed by the required calibrations and checks.
Outside it, and outside this course: using service passwords or keys you were not given, activating options or licences, changing radiation-related parameters without the required acceptance or constancy tests afterwards, loading firmware or software other than by the manufacturer's procedure, and bypassing any interlock, door switch, emergency stop or protection, even 'just for a test'. If a repair needs a step you are not authorised for, stop and document the evidence for the party who is.
Triage before you travel, and keep a fault diary
Many CT faults can be narrowed down before you reach the site. Ask the operator to read you the exact message, not a summary of it, and to send a photograph of the screen and of the log page if the system allows. Ask when it started, whether it is constant or comes and goes, whether it follows a particular protocol, and what the room temperature is. Ask whether there was a power cut, a storm, building work, a water leak or a cleaning visit in the gantry room.
With that information you can decide which spares and tools to take, whether you need the manufacturer's service software and access level, and whether the scanner can still be used safely for some exams in the meantime. That last decision belongs to the clinic's responsible person, but they need your technical opinion in writing to make it.
Write as you go: the as-found state, the log excerpt, each test with its result, each change, and the state as left. A good diary turns an intermittent fault into a pattern, lets a colleague continue tomorrow without repeating your work, and protects you when the customer asks why a part was or was not replaced. It also becomes the record the customer needs for the scanner's service history.
| Part | What it does | How it fails | |
|---|---|---|---|
| 01 | Mains, PDU and UPS | Feed and distribute power to gantry, table and computers | Sags under load, a missing phase, a UPS that feeds only part of the system |
| 02 | Slip ring and brushes | Carry power and often data onto the rotor | Dirt, wear and arcing cause power drops and data errors |
| 03 | HV generator | Makes stable kV and mA for the tube | kV or mA regulation errors, inverter faults, arcs |
| 04 | X-ray tube | Produces the beam; anode, bearing, filament | Arcing, rotor start errors, filament open, oil and heat problems |
| 05 | Collimator and filters | Shape the beam in z and across the fan | Blade jams, position errors, wrong slice width |
| 06 | Detector and DAS | Convert X-rays to digital readings | Channel or module dropouts, temperature errors, rings |
| 07 | Data link | Sends readings from rotor to stationary side | CRC or framing errors, lost data, aborts at speed |
| 08 | Recon, host and storage | Reconstruct, display and store images | Disk, file system, network and software faults |
| 09 | Gantry drive, tilt and table | Move rotor, gantry and patient precisely | Drive errors, encoder loss, brakes, collision sensors |
The tube that was 'hot' every morning
A Canon/Toshiba Aquilion would not allow scanning after the consoles were switched on: an error appeared, and after a reset the tube heat display showed 100 percent, then slowly 'cooled' until the system let the operators work. Nothing in the morning workload explained a hot tube. The engineer treated the heat reading as a calculation rather than a measurement and asked what the calculation depends on: elapsed time. The timekeeper battery measured far below its rated voltage, so the system lost track of time whenever it was powered down and assumed the worst. The timekeeper was replaced, the date set, and the filament current calibration repeated as the procedure required. The morning lock-out stopped.
Some 'readings' are calculations. When a value is impossible, ask what the system computes it from - here, time - and check the clock before blaming the tube.
Five emergency stops still pressed
A CT gantry switched itself off in the middle of normal work and would not come back. The next morning the engineer did not start with boards or power supplies. He walked round the system and checked every emergency stop button, including the less visible ones, and found five of them latched in. The story came out quickly: a table foot pedal had jammed, staff had pressed emergency stops in a hurry, and not all were released afterwards. The pedal was freed, the buttons released, and the system started normally. The fix took minutes; a parts-first approach would have taken days.
Before any deep diagnosis, check the simple safety chain: emergency stops, door and cover switches, pedals. Then ask what happened just before the fault.
The detector module with the wrong number
On a GE scanner with an older detector, the log blamed a module by number, but the engineer was not sure the logged number matched the physical position: on some older software versions the number in the error had to be converted to find the real module, while newer versions report it directly. Rather than order a module on a guess, he swapped two modules and watched whether the error followed. It did, which identified the failed module with certainty before anything was ordered.
Log numbering conventions differ between systems and software versions. Confirm with the documentation or a swap test before ordering an expensive module.
On-site checklist
- Export or photograph the log before any reboot.
- Check the system clock and log order.
- Find the first error of the burst and the scan stage where it failed.
- Record protocol, kV, mA, rotation time, angle, time and room temperature at failure.
- Reproduce with the cheapest test or a phantom, room cleared for any exposure.
- Change one thing at a time; same part number and allowed revision only.
- Run the calibrations required after every replacement.
- Keep a fault diary: as-found, tests, changes, as-left.