How an ultrasound system is built: from probe to picture

Follow an echo from the probe face to the monitor and name the block that each common complaint points to, so that you can separate a probe fault from a system fault before you open a cover.

What the machine actually does

An ultrasound system does one simple thing thousands of times a second. It sends a short burst of sound into the body and then listens. Every boundary between tissues reflects a small part of the sound back. The system turns the time each echo takes to return into a depth, and the strength of the echo into a brightness. Do this along a few hundred lines across the probe face and you have one B-mode frame. Repeat the frame many times a second and you have real-time imaging.

The arithmetic rests on one assumption: sound travels through soft tissue at about 1540 metres per second. The system does not measure this speed; it assumes it and calculates depth from it. Echoes from deeper structures take longer to return and also come back weaker, because tissue absorbs sound, and it absorbs higher frequencies faster. That trade-off — high frequency for fine detail near the surface, low frequency for reaching deep organs — is why one department needs several different probes. Lesson 2 returns to this.

For a service engineer the useful model is a chain of blocks, each passing the signal to the next. When the image is wrong, the question is always the same: at which link did it go wrong? This lesson walks the chain in the order the signal travels, and for each link says what it does and how it usually fails.

The probe: the first and most fragile link

Inside the probe is an array of small piezoelectric elements — commonly a hundred or more on an ordinary one-dimensional array, and thousands on a matrix probe. A voltage pulse makes each element ring at its resonant frequency; a returning echo makes it produce a tiny voltage. In front of the elements sit one or more matching layers, which help the sound pass from hard ceramic into soft tissue, and the acoustic lens — the rubbery face the patient touches — which focuses the beam in the thin direction across the array. Behind the elements a backing block damps the ringing so that pulses stay short. Each element has its own fine conductor running through the cable to the connector.

Every one of these parts can fail in the field. Elements crack or lose their connection, and each lost element is a lost contributor to the image. Lenses are cut, worn by abrasive wipes, swollen or dried by the wrong chemicals, and separated from the layers beneath (delamination). Cables are run over by chair wheels, kinked in drawers and wound tightly around the hand, and the fine conductors break inside an undamaged jacket, usually close to the strain relief at either end. Connectors are dropped, pins are bent and contacts collect dust. The probe is the part the user handles hundreds of times a day, so it is no surprise that a large share of ultrasound service calls end at the probe.

The lens, the housing and the cable jacket are also the insulation between the patient and the electronics. A crack is not only an image problem; it is an electrical safety problem. Lesson 3 deals with that.

Connector, probe recognition and switching

A probe connector carries far more contacts than an ordinary cable connector, because every element needs its own path, plus shields and control lines. A mechanical latch pulls the connector firmly onto the port. Most probes also carry a small identification circuit or memory, so that the system knows which probe is attached and loads the right presets and the right acoustic output limits. A probe that is 'not recognised' may therefore have a fault in the identification path, in the connector pins or in the system port — not necessarily in the array.

Most systems have more probe ports than sets of front-end electronics, and many probes have more elements than the system has channels. Relays or electronic multiplexers sit between the ports and the channels. They select the active port, and on many designs they also select which group of elements is connected to the channels for each scan line. On many systems you can hear the relays click at start-up or when you change probes. A failed relay or multiplexer path gives a fault that follows the port or the channel, not the probe — one of the most useful clues you have.

Transmit and receive: high voltage out, microvolts back

The transmitter, or pulser, produces the drive pulse for each element with precisely controlled timing and amplitude. The pulses come from a dedicated high-voltage supply inside the system. Their amplitude depends on the model, the probe and the mode, and the system limits it to keep acoustic output within the permitted range. Two facts matter here. First, there is high voltage on the transmit circuits and at the probe connector while the system is scanning, so covers come off only with the system switched off, disconnected from the mains and allowed to discharge as the manufacturer describes. Second, a shorted probe or a damaged port can overload the transmit supply. On some systems one faulty probe can trigger a transmit-power error that stops imaging on every probe.

On the way back the echoes are tiny, many orders of magnitude smaller than the transmit pulse. A transmit/receive switch protects the sensitive receiver while the pulse goes out. Each receive channel then has a low-noise amplifier, time-gain compensation (TGC) — extra gain applied to later, deeper echoes to make up for attenuation — and an analogue-to-digital converter. Faults here typically take out a group of channels on one board, and they show on every probe that uses those channels.

Beamformer and image processing

The beamformer is what makes an array more than a row of separate microphones. On transmit it fires the elements with small time delays so that their wavefronts add up into a beam that is focused at a chosen depth and, on some probes, steered to a chosen angle. On receive it delays and sums the signals from many elements so that echoes from the point of interest add up and echoes from elsewhere partly cancel. Current systems do this digitally, in dedicated hardware or largely in software. Because every line is built from many elements, a dead element or a dead channel rarely gives a blank screen. It gives a vertical stripe of lower brightness, or a softer, noisier region, because one contributor to each affected line is missing.

After beamforming, the processor detects the echo envelope for B-mode, calculates frequency or phase shifts for colour, pulsed-wave and continuous-wave Doppler, and converts the lines from the probe's geometry into the rectangular pixels of the monitor (scan conversion). Faults at this stage tend to be mode-specific: B-mode is normal but colour is missing or noisy, or CW Doppler fails while everything else works. Remember that many apparent faults at this stage are presets, filters and user settings, not hardware.

The PC: an ordinary computer in an unusual box

Behind the imaging electronics, almost every current system runs on a PC-type computer: a standard operating system, a hard disk or SSD, memory, a graphics card, a CMOS battery, USB hubs for the control panel and a network interface for DICOM. It fails like any PC. Disks wear out, fans and filters clog and the system overheats, mismatched memory causes random freezes, a flat CMOS battery after months in storage gives start-up errors, and a failed graphics card leaves the monitor showing 'no signal'. The power supplies and the cooling serve both halves of the system, so a cooling problem can appear as an imaging problem and the other way round.

Treat this half with the same discipline as the imaging half. Read the logs before you replace anything, check cooling first because it is free, and never replace a disk or reinstall software without a valid software image and a backup of the presets and options for that exact system. Options and licences are tied to the system; they are restored through the manufacturer's procedure, never worked around.

Reading a complaint against the chain

When the call comes in, place the complaint on the chain before you travel. Three questions do most of the work: does it happen on one probe or on all probes, on one port or on all ports, in one mode or in all modes? The answers point to a block:

• One probe, on any port — dark vertical line, flicker when the cable moves, 'not recognised': the probe, its cable or its connector.
• Every probe on one port, other ports normal: the port, its pins or its switching.
• Every probe on every port, stripe in the same place: receive channels or the beamformer.
• Weak image on every probe in every mode, once settings are excluded: transmit supply or pulser.
• One mode only, for example colour or CW Doppler: processing, presets or that mode's own hardware path.
• Freezes, reboots, no boot, blank monitor: PC, disk, memory, graphics, cooling or power supply.

On-site checklist

• Ask three questions before travelling: one probe or all, one port or all, one mode or all.
• Ask what changed recently: a dropped probe, cleaning, transport, a software update, long storage without power.
• Write down the exact error text and save the system logs before replacing anything.
• Move the suspect probe to another port, and put a known-good probe of the same type on the suspect port.
• After a transmit-power or similar error, remove the suspect probe and restart with known-good probes only.
• Open covers only with the system off, disconnected from the mains and discharged as the manufacturer describes.
• Back up presets, options and configuration before any disk or software work.

Check yourself