The Level 2 method for X-ray: logs, exposure history, the subsystem map and test exposures
Collect the error list and exposure history, place an X-ray fault in the exposure sequence, choose a test exposure, phantom or meter reading that separates two suspects, and know what you may and may not change.
Where Level 1 ends
In X-ray Service — Level 1 you learned to make a room safe, to measure kV, time and output with a meter, to check collimation and image quality and to keep a record. Lesson 8 of that course gave you the basic method: facts first, subsystem next, cheap and likely causes before expensive ones, one change at a time. Level 2 uses the same method, but on faults that survive the first round of checks: the error that comes back after a board swap, the image that is wrong although every measurement passes, the stand that moves by itself once a week.
The costly mistake at this level is not lack of knowledge. It is believing the module that reported the error. An X-ray system is a chain: building supply and earth, mains contactor, generator with its DC link or energy store, high-voltage tank and cables, tube and housing, collimator and filters, grid, AEC chamber, detector or cassette reader, image processor, workstation, network. A failure anywhere upstream is usually noticed and reported by a block downstream. Replace that block and you have spent the customer's money and kept the fault.
Read the error list and the exposure history
Almost every modern generator and system controller keeps an error list with a time stamp, a code and often the kV, mA and time requested. Many also keep separate histories: the Philips BV Family service software, for example, offers a complete system log file, a tank history, an energy storage history on battery or capacitor models, and separate error lists for individual controllers. Older analogue or locally built systems may show only a code on the console and a few LEDs on a board; then the LEDs and the user's account are your log.
Work backwards from the moment the exposure failed. The first event in a burst is usually nearest the cause; the rest are consequences. A supply dip will show first as an undervoltage or DC link message, then as a kV regulation error, then as an aborted exposure, then perhaps as a lost link to the detector. Look also for repeated warnings in the days before the call: arcs counted but tolerated, rotor start-up retries, a temperature alert every afternoon. These are the history of the fault.
Check the system clock before trusting the order of events. A flat backup battery on a console, a generator board or a PC can reset the date and, on some systems, the stored calibration or settings as well. Field engineers meet this often enough that a clock error or a sudden loss of settings should make you think of the battery first and the board second.
- Photograph or export the error list before rebooting or clearing it.
- Write down the exam, workstation (table, wall stand, fluoroscopy), kV, mA, mAs or time, focus and AEC field at the moment of failure.
- Ask what changed: a power cut, building work, a new device on the same circuit, a moved unit, a software update, earlier service work, cleaning.
The exposure sequence is your subsystem map
An exposure is a fixed sequence, and each step needs a known set of subsystems. If the system does not finish its start-up, think supply, low-voltage power supplies, the communication between console and generator, and the safety chain of limit switches and emergency stops that many systems check at power-up. If it stops at preparation, ask which unit did not report ready: anode rotation, filament pre-heat, detector ready, bucky or grid in position, collimator in position, door contact. If it stops at the moment of exposure, think high voltage: DC link or energy store, inverter, tank, cables, tube arcing. If the exposure completes but no image arrives, you are in the detector link, the reader or the software. If an image arrives but looks wrong, the question is whether the cause is before the patient (tube output, filtration, collimation, AEC) or after the patient (grid, detector, calibration, processing, display).
Rooms with several workstations give you a free comparison. One generator and tube often serve a table bucky, a wall stand and sometimes a second tube or a fluoroscopy table. A fault that appears on every workstation lives in what they share: supply, generator, tube, collimator. A fault on one workstation only lives in what is specific to it: its detector or cassette tray, its grid and AEC chamber, its cable and its selection signals, its position sensors. Ask the radiographers to try the same exam on the other workstation before you open anything; the answer costs one exposure and often halves the list.
That last split is worth a habit. A fault before the patient changes the dose or the beam: a meter reading, a half-value layer, a light field check will show it. A fault after the patient usually leaves the meter readings normal and shows a fixed pattern on a flat-field image.
Reproduce it with the cheapest test
A fault you cannot reproduce you cannot prove fixed. Find the smallest test that makes it happen. Does the anode run up without an exposure? Does a low exposure work and a high one fail? Does it fail only at the wall stand, only on the large focus, only with AEC, only after twenty exposures, only when the table is tilted? Each answer removes a group of suspects.
Make test exposures without a patient, with the room cleared, the door closed and the warning signs working, as in Level 1 lesson 2. Use a non-invasive meter for kV, time and dose, a flat-field or uniformity exposure for the detector, a phantom or a known test object for geometry and image quality. Step one parameter at a time: kV in small steps at fixed mA and time, then mA at fixed kV. When an error appears at one step and not the one below, you have found a threshold, and the threshold tells you a lot. A fault that appears above a certain kV regardless of mA points at insulation: tube, cable or tank. A fault that appears above a certain power points at the supply, the DC link or the inverter.
If the fault is intermittent, set up to catch it: leave a power quality recorder on the supply, note every occurrence with time and technique, ask the users to photograph the screen. A list of ten occurrences with conditions is worth more than an hour of guessing.
Swap, isolate, measure: one change at a time
Swapping identical parts is a strong test when the system has them: two detectors, two identical motor drivers, two stands on one generator, two focal spots. Swap one thing, test, and put it back if nothing changed. Isolation is the other tool: disconnect a pedal, a hand switch or a remote panel (with the system off and by the manufacturer's procedure) and see whether a false command or a single-fault message disappears. Measurement closes the argument: supply voltage under load, the voltage between neutral and protective earth, a low-voltage rail on a board, a potentiometer over its full travel.
Bring the tools that make these tests possible: a calibrated non-invasive X-ray meter with its dose and kV probes, a true-RMS multimeter and a clamp meter, a power quality recorder that logs the minimum voltage during an exposure, an oscilloscope with properly rated isolated or differential probes for low-voltage control signals, a set of attenuation plates or a uniformity phantom, a test object for resolution and contrast, and the baseline report from acceptance or the last major repair. Without the baseline most measurements are just numbers; with it they become evidence of change.
Never 'test' by bridging a safety circuit, defeating a door contact or raising a fuse rating. A safety chain that opens is telling you that something is wrong; your job is to find the open contact, not to remove the question.
What you may and may not change
At Level 2 you will see service software, configuration tables and calibration menus. Seeing them does not mean you may use all of them. The rule is simple: you change only what the manufacturer's procedure for your authorised level allows, and only after you have saved what was there. Many menus carry warnings for good reason; a reset of calibration data, for example, returns a system to default values that do not match its tube, its detector or its AEC, and the room then gives wrong doses until it is recalibrated in full.
You do not use passwords, service keys, dongles or option activations you were not given, and you do not copy calibration or configuration files from another unit: filament and AEC data belong to one tube and one detector. Licence and software work, HV tank work, tube replacement and generator calibration belong to people trained and equipped for them. When you reach that line, stop and hand over a clear package: the error list, the conditions, your measurements, what you excluded and what you suspect. Lesson 8 returns to this for the return to service.
Errors not in the manual, and an earth loop that was gone
A digital radiography room with a CPI Indico IQ generator and an Opera table began to show generator errors in the 903-905 range, which the engineer could not find in his documentation. A colleague's first remote suggestion, a stuck exposure button at power-up, did not explain it. On site the engineer measured the supply: one phase sat near 260 V while the other two were around 200-220 V, and the voltage between neutral and protective earth was 15-30 V. That pointed away from the generator and towards the building. The earth electrode outside had been damaged during snow clearing, so the earth and neutral references had shifted. Once the earth loop was restored, the errors stopped. The quality of the neutral from the utility was raised with the hospital as a separate issue.
When errors make no sense, measure phases and the neutral-to-earth voltage before reading another page of the manual. A broken earth is both a safety hazard and a source of 'impossible' faults.
Hung at boot by a limit switch
A Spektr-AP radiography unit hung at power-up: every console button flashed and the screen stayed on the boot sequence. The engineer started where many would, with the console batteries and the 15-pin cable to the generator, and he watched the transmit and receive LEDs on the link. All of that was in order. The real cause was simpler: the stand was sitting on an emergency limit switch, in this case at the end of the source-to-image distance travel, and the controller would not complete start-up with that circuit open. Once the stand was moved off the switch by the manufacturer's method, the system started normally.
A system that will not finish start-up may be refusing to, because its safety chain is open. Check limit switches and stops before the communication path.
A single-fault error that lived in a pedal
A radiography system sometimes showed a 'single fault' error at power-up, typically after half an hour of normal use and a restart. Earlier, a new foot pedal had made it disappear for a while. The engineer read the circuit: the single-fault line ran in series through the joysticks, the buttons and the pedals, so any of them could open it. With the system off he checked the pedal contacts pin by pin and found one pin broken away in the pedal connector, making contact only some of the time. Repairing the connection cured the fault for good.
When a monitoring chain runs through several devices, trace the whole chain. A part swap that helps 'for a while' usually disturbed the real fault rather than fixed it.
On-site checklist
- Save the error list and histories before rebooting.
- Check the clock and backup batteries when events are out of order or settings are lost.
- Measure phases, phase balance and N-PE voltage at rest and during exposure.
- Place the fault in the sequence: start-up, preparation, exposure, image transfer, image quality.
- Reproduce without a patient; step kV, then mA, and note thresholds.
- Swap or isolate one thing at a time and put it back if nothing changed.
- Save configuration before any change and stay within your authorised level.