Intermittent Ultrasound Faults: Why "Works Fine, Then Doesn't" Is Hard to Fix

Safety note: This article is written for qualified biomedical engineering personnel and covers equipment maintenance only. It is not clinical guidance. For anything touching electrical safety or patient safety, the manufacturer's service manual and your local regulatory requirements govern.

Ask any biomedical engineer which fault they dread, and you won't hear "the machine died." A dead machine is a gift. It won't boot, the error code sits on the screen, and the path from complaint to repair is a straight line.

What wears people down is the other kind. The system that fails once on Tuesday, restarts clean, and then runs flawlessly for a week. The department files the same work order for the third time and you arrive to find everything nominal. Nothing is wrong — except that something clearly is.

After enough years on the bench, we stopped calling these hard faults. They're unstable faults, and the difference matters: the failure point isn't static, it moves with temperature, position, vibration, and time. That's why "just restart it" is the worst possible news rather than the best. It means the defect is already there. It simply hasn't settled into a pattern you can catch yet.

1. Why Intermittent Faults Resist Diagnosis

A hard failure gives you a stable observation window. An intermittent one gives you almost nothing — the system is usually behaving by the time you reach the room, and misbehaving again by the time you leave.

In practice, unstable faults fall into three patterns. Identify which pattern you're looking at and the diagnostic field narrows by half before you touch a screwdriver.

Pattern	Trigger condition	Signature	Common root cause
Thermally intermittent	Cold start is fine; fails 20–40 minutes into use	Correlates with runtime; recovers after cooldown	Cold solder joints, aged capacitors, poor heat dissipation
Mechanically intermittent	Fails on movement, height adjustment, rotation, or pressure	Correlates with system posture or cable position	Broken conductors inside intact cable, oxidized connectors
Electrically noisy	Varies by time of day or by neighboring equipment cycling	Tracks with other equipment switching on/off	Poor grounding, degraded shielding, mains coupling

The fastest triage is three questions to the department: When did it start? What was the system doing when it failed? What else was running nearby? The answers usually point straight at one row of that table.

This class of fault — passing self-test, then failing under sustained use — is common enough in ultrasound that it deserves its own breakdown. We covered the layered version in why systems that pass idle self-checks still fail during long scans.

2. Four Root Causes Behind Almost Every Ghost Fault

Symptoms sprawl. Root causes don't. In our experience the overwhelming majority of unstable faults trace back to four categories — and knowing which category you're in prevents the most expensive mistake in the book, which is replacing a board because you misread a cable.

2.1 Hidden cable and connector failures

The cable looks perfect. Inside, conductors are broken, shielding is worn through, or a connector has oxidized loose.

Every movement the system makes — transport, height adjustment, probe rotation, the daily flexing at the strain relief — works a little more damage into it. The result is a connection that is almost made. Pull and reseat the connector and it recovers. A week later, it doesn't.

The signature to watch for: the fault changes when something moves.

2.2 Unstable power and grounding

Voltage sag, elevated neutral-to-earth voltage, a loose receptacle, a degraded earth conductor. None of these will stop a system from booting.

What they will do is produce random lockups, aborted scans, image noise, and error codes that appear without pattern. This is the most commonly missed category precisely because it never produces a stable, reproducible error — and because when you measure at the outlet during a service call, everything reads normal.

2.3 Probe and consumable aging

Element dropout, lens delamination, crystal drift. These are progressive, and they do not trigger a hardware fault code. They show up as degraded image quality, reduced sensitivity, and poor repeatability — which is exactly why they get misdiagnosed as a mainboard problem.

2.4 Software cache and system logic

Long uptimes, accumulated logs, hung processes, corrupted parameters. A surprising share of "complex" faults resolve with a full power-down, a parameter reset, and a cache clear — no disassembly required.

Expert Insight: The four categories differ enormously in cost to rule out. Software takes about five minutes. A board-level hardware fault can take half a day. That asymmetry is the entire argument for working easy-to-hard — not hardest-to-find-first.

For the full taxonomy of ultrasound system faults, see a comprehensive analysis of common technical faults in ultrasound systems.

3. Boot Failures: Tracing the Power Chain

"Sometimes it won't turn on" is one of the most common intermittent faults on an ultrasound system, and one of the most misdiagnosed — because the first instinct is to suspect the mainboard.

3.1 Read the indicator sequence

Power-up on an ultrasound system is staged. The indicator sequence is not decoration; it's diagnostic output.

What you observe	Which stage is failing	Where to look
No indicators, no fans	Primary input	Mains feed, receptacle, main fuse, power switch
Standby light present, no response to power button	Power sequencing incomplete	A rail isn't coming up on time; CPU held in reset
Lights on, then off, cycling	Protection circuit tripped	Output overcurrent/overvoltage; fault on the load side
Boots normally, drops out during scan	Insufficient load capacity	Aged supply module — fine unloaded, collapses under load

The value of this table is that it converts "it won't turn on" — a useless description — into four completely different investigations.

3.2 Power regulation board and control module

The power regulation board establishes and monitors the voltage rails; the data control module sequences the power-up. Both fail intermittently when the underlying cause is a cold solder joint or an aged capacitor, because conduction through those joints changes with temperature.

Which is why the classic field story repeats: first system of the morning boots perfectly; the same system refuses to start after a full day of scanning.

3.3 Battery and 24 V changeover instability

Cart-based systems with an internal battery switch between battery and mains supply. When that changeover circuit is marginal, you get genuinely counterintuitive behavior — a system that won't start while plugged in, but starts fine on battery.

Watch Out: When the symptom contradicts your expectation, suspect the switching logic before you suspect the mainboard. Power faults are the best liars in the machine.

Power problems are especially good at hiding behind symptoms that look nothing like power problems — we went through that failure mode in how power regulation drift hides behind non-power symptoms. For the underlying causes specific to ultrasound supplies, see common causes of power supply failures in ultrasound systems.

4. Image Interference: Working Backward from the Artifact

Interference is the category that most tests judgment, because the same visual symptom can originate in an external source, a cable, a probe, or the main unit — and the cost to resolve those differs by an order of magnitude.

4.1 Rule out the outside first

Start with electrosurgical units. An ESU radiates broadband energy that couples straight into the ultrasound signal chain.

The test takes one minute and saves half a day: have the department power down the ESU and scan again. Interference gone, the source is external. Interference unchanged, it's in the system.

4.2 Testing shield integrity — actually testing it

Transducer cables carry a braided and foil shield around the signal conductors. Repeated flexing breaks the braid internally while the outer jacket stays intact. You will never see this by looking.

The measurement, and the part everyone skips:

1. Set a multimeter to continuity. Put one probe on the connector shell, the other on the shield pin at the system end.
2. Keep both probes in contact and slowly flex the cable along its full length — concentrating on the points that bend in normal use: behind the connector, at the probe neck.
3. A continuity reading that flips as you reach a particular angle is your shield break.

A passing reading on a stationary cable proves nothing. This is the single most-skipped and most-productive step in intermittent ultrasound diagnostics.

4.3 Element dropout vs. cable fault: the test that saves a probe

This is the most practically valuable distinction in the article. The two failures look similar on screen and differ by an order of magnitude in cost.

	Element dropout (probe body)	Cable fault (replaceable assembly)
Artifact position on image	Fixed — always the same location	Varies with cable position
While flexing the cable	No change	Artifact appears/disappears with flex
Pressing the probe lens	No change	No change (resolves after cable replacement)
Moved to a second system	Artifact follows the probe	Artifact follows the probe
Correct action	Probe repair or replacement	Replace cable assembly

The one-line rule: fixed position means probe, moves with the cable means cable. Ten minutes spent on this check comes before any decision to replace a probe.

Expert Insight: Element dropout typically kills one or two elements out of hundreds and reports nothing. Self-test passes, the image is "usable," and one region is permanently darker than its surroundings. These are the faults that get dragged out for months as "the image just doesn't look right and nobody can say why."

For probe-level fault isolation by model, see common faults and diagnostics for the Philips X7-2t TEE probe. To understand how element layout relates to beamforming before you interpret an artifact, see a technical comparison of linear, curvilinear, and phased array probes. Some display anomalies also originate in the supply rather than the screen — covered in why display artifacts can start with power instability.

4.4 Intermittent probe identification failure

The system identifies a probe by reading an identification resistor or memory device inside it. When that path makes intermittent contact, the system misidentifies the probe — wrong presets load, frequency options go missing, image character changes unexpectedly.

To test it: reseat the probe several times and watch whether the recognized model stays stable. If the identification result changes between insertions, you've found the fault.

5. Deep Water I: Turning "Can't Catch It" into "Caught It"

The hardest part of an intermittent fault is never the repair. It's getting the system to fail while you're standing there. Not being able to fix it is manageable. Not being able to reproduce it is what burns the hours.

5.1 Provoke it — don't wait for it

Passive waiting has terrible throughput. Match your method to the pattern from Section 1:

• Thermally intermittent — use a heat gun to locally warm the suspect area, or freeze spray to locally chill it. Temperature change is far faster than elapsed time.
• Mechanically intermittent — systematically flex cables, move joints, press connectors. Change one variable at a time, or you lose the ability to attribute the result.
• Electrically noisy — observe while adjacent equipment cycles; reroute signal and mains cables closer together and further apart.

5.2 Structured waiting, when provocation fails

Some faults genuinely will not reproduce on demand. Your job then is to turn waiting into structured observation:

• Log timestamps of every occurrence — not "it happens sometimes."
• Log the operating context: uptime, what was being scanned, what else was running.
• Track the trend rather than the point. Several days of records will tell you more than one service call ever will.

Failure data collected across years reveals patterns no single visit can. That's the argument behind insights from a four-year predictive maintenance review.

5.3 Substitution bisection: cut the problem in half

When a fault can neither be provoked nor waited out, substitution is the most efficient tool available. The discipline is one variable per step.

Step	Action	Fault gone	Fault persists
1	Swap the probe	Probe-side fault	Main unit or cable
2	Swap the cable	Cable fault	Main unit
3	Swap the power supply / feed path	Supply fault	Inside the main unit
4	Swap the main unit	Main unit fault confirmed	Return to environment and grounding

Swap one, verify, then swap the next. Changing three components at once and finding the fault gone tells you nothing about which one mattered — and it is the most common unforced error in this work.

5.4 Knowing when to stop

Not every intermittent fault justifies unlimited effort. The deciding factor is clinical risk and downtime cost, not technical curiosity.

• Fault affects diagnostic output (degraded image, unreliable measurements) — must be resolved. "Monitor it" is not an option.
• Fault affects only workflow (occasional restart needed, clean afterward) — evaluate cost against benefit.
• Fault touches electrical safety (leakage current, grounding integrity) — remove from service regardless of how intermittent it is. No trade-off exists here.

6. Deep Water II: "Repaired" Is Not "Ready for Service"

This is the section most often skipped, and the one with the highest consequences.

6.1 Why retesting after repair is mandatory

A system that boots and produces a normal image has demonstrated that its function is restored. It has demonstrated nothing about its electrical safety.

A system can suffer a broken earth path or damaged insulation during disassembly and still image perfectly. It is functioning. It may also no longer be safe. Retesting isn't procedural box-ticking — it's the threshold that decides whether the equipment goes back to clinical use.

6.2 Two standards, frequently confused

	Recurrent / after-repair testing	Type testing
Standard framework	IEC 62353	IEC 60601-1
Applies to	In-service equipment and post-repair verification	Design qualification of the equipment type
Performed by	Hospital BME / service organizations	Manufacturers
Characteristics	Performed on site, without dismantling	Requires dedicated laboratory conditions

These are not interchangeable. Citing a 60601-1 type approval from the factory as evidence that a system is safe today is a category error.

6.3 What to measure

Three core measurements:

• Protective earth resistance — verifies the earth path is intact
• Equipment leakage current — current flowing in the protective earth conductor
• Applied part leakage current — current through the portion contacting the patient

Watch Out: Specific limits are deliberately not stated here. Applicable limits vary by standard revision, by applied-part classification (Type B / BF / CF), and by local regulation. They must be taken from the manufacturer's service manual, the current standard text, and your local regulatory requirements. Using the wrong limit is more dangerous than not testing at all, because it produces a false conclusion of safety.

6.4 What a fault log must contain

Intermittent faults have a way of becoming blame events. The only defense is turning the process into a record. A log that can stand on its own carries five fields:

Field	Content	Purpose
Symptom	Objective description, no inference	Separates equipment fault from operator issue
Conditions	Time, operating state, nearby equipment	Supports the intermittency finding
Frequency	Occurrences per total uses	Quantifies severity
Action	What was done, and the result	Demonstrates diagnostic diligence
Reproducibility	Whether and how it can be reproduced	Distinguishes "located" from "not located"

6.5 Talking to the clinical team

The same situation, two ways of describing it:

• "I looked at it and couldn't find anything." — reads as a brush-off.
• "Power, grounding, cabling, and probe are all ruled out. The fault is isolated to the signal processing path in the main unit. I recommend scheduled downtime for a deeper check." — reads as an engineering judgment.

The difference isn't how much you did. It's how much of it you said out loud. Stating what you eliminated matters as much as stating what you found.

The broader framework for managing safety across a system's service life is covered in medical ultrasound life cycle management.

7. The Four-Step Diagnostic Sequence

For an unstable fault with no obvious pattern, work this sequence in order. It consistently shortens time-to-isolation.

Step 1 — Software before hardware. Full power-down for 5–10 minutes. Restart, reset base parameters, clear system logs. Rules out hung processes and corrupted configuration. Lowest cost, so it goes first.

Step 2 — Outside before inside. Check the mains supply, receptacle, earth, and nearby interference sources. Then cables, connectors, and probe consumables. Only then open the chassis for board-level inspection. Reverse this order and you're adding work for nothing.

Step 3 — Find the cause by moving it. For intermittent faults specifically: flex cables, work the joints, press connectors, and reproduce the failure deliberately. This is what isolates a broken conductor, a loose contact, or an oxidized pin. It's the step that separates intermittent faults from hard ones.

Step 4 — Verify by substitution. Cross-swap probes, cables, and power supplies between identical systems to separate a component fault from a main unit fault — and avoid a pointless teardown.

For the general-purpose checklist version of this workflow, see a universal repair checklist before you pay for service. For the repair techniques themselves, see common repair techniques for medical ultrasound equipment.

Parts decision: repair the component or replace the assembly

Once you've isolated the fault, you're making a cost call.

Fault type	Recommended action	Basis
Cable shield broken	Replace cable	Low cost, main unit unaffected
Isolated element dropout	Evaluate probe repair vs. replacement	Number of failed elements, remaining service life
Lens delamination	Replace probe	Acoustic coupling failure; not field-repairable
Intermittent probe ID failure	Check the interface first, then decide	Often a contact problem, not a probe failure
Intermittent board-level fault	Assess against the service manual	Requires dedicated test equipment

The governing principle: confirm the fault location before deciding what to replace. Get that order backwards and you're using parts cost as a diagnostic method. For a structured way to weigh exchange against repair, see service exchange versus component repair.

If you've already narrowed the fault to a probe or cable using the tests above, you can submit your system model, probe model, and observed symptoms to check replaceable part availability and pricing.

8. A Note for the Engineers Who Hold the Line

Medical equipment maintenance has never been about knowing how to open a chassis and swap a part.

What separates an experienced biomedical engineer is a sense of fault patterns, precision about detail, and an uncompromising respect for safety. Every system we keep running steadily is a system that keeps diagnosing patients — and every fault we catch before it fails in use is a fault that never reaches a patient.

Newer engineers often describe intermittent faults as unpredictable and untraceable. In our experience that's not a capability gap. It's a missing method and a missing record.

Work enough of them and the pattern becomes obvious: every random error has a deterministic cause behind it, and every stubborn problem has a rule that explains it.