13.2 Transducers, Couplant, and Straight-Beam Techniques

Official UT topic 1, Review of Ultrasonic Techniques, spends as many items on hardware and 0° methods as it does on v = fλ. The wave in 13.1 is useless until a transducer launches it, couplant (or a water path) gets it into the part, and a straight-beam setup turns time into thickness or a mid-wall echo into a lamination call. Angle-beam weld work in 13.3 is the same machine with a wedge. Learn the 0° package first.

ASTM E114 is the straight-beam contact practice. ASTM E797 is the manual thickness practice. ASME Section V, Article 5 is the Code UT article that still leans on 0° work for material and bonding. None of them publishes a secret ASNT crystal diameter. If the stem gives a frequency, a delay, or a velocity, use those numbers.

Piezoelectric transducers

A piezoelectric element changes thickness when a voltage is applied and produces a voltage when a returning stress wave strains it. That two-way conversion is the entire probe:

1. The instrument dumps a short voltage spike (the initial pulse, main bang) across the crystal.
2. The crystal rings at its thickness-mode frequency and launches a mechanical pulse into the wear face, delay line, or water.
3. Returning echoes strain the same crystal (pulse-echo) or a second crystal (dual / through-transmission).
4. The instrument displays amplitude versus time — the A-scan.

Common element materials you should be able to recognize, not romanticize:

The nominal frequency is set mostly by crystal thickness: thinner crystal, higher frequency. A worn crystal, a cracked wrap, or a dead matching layer is a process-control problem (later evaluation chapters), not a reason to invent a new frequency on the fly.

The crystal does not choose the wave mode by being "an L-wave brand" or "an S-wave brand" in the marketing sense. A thickness-expander crystal on a 0° wear face launches a longitudinal wave into the couplant and the metal. Shear in the part is produced by a wedge (13.3), by a Y-cut or 1-3 composite built for contact shear, or by mode conversion at an interface. If the stem says "5 MHz contact probe" and "0°," think longitudinal until the stem says otherwise.

Frequency, resolution, and penetration

Frequency is the first trade the Level II actually owns.

Raise frequency (5 MHz → 10 MHz, or 2.25 MHz → 5 MHz):

• λ shrinks (13.1). Small pores, thin ligaments, and closely spaced reflectors become easier to resolve if the pulse is also short.
• Attenuation rises. Scatter from grains and absorption eat the pulse. Thick forgings, coarse austenitic welds, and castings go dark.
• Near field lengthens (N ∝ f). The last maximum sits deeper.
• Beam spread shrinks (γ shrinks as λ/D shrinks).

Lower frequency does the opposite: more penetration, more spread, longer λ, usually a worse small-defect bet.

Resolution on the A-scan is not only λ. It is also pulse length. Two reflectors separated by less than about half the spatial pulse length merge into one blob. A 5 MHz probe that rings for eight cycles is a worse resolver than a damped 5 MHz probe that dies in one and a half cycles. Frequency sets the possible λ; damping (next heading) sets how many cycles you actually ship.

Near-surface resolution (how close to the entry surface you can see) is limited by the dead zone: the main bang, ringing, and recovery time of the receiver. Far-surface resolution (separating a backwall from a reflector just above it) is limited by pulse length and by whether the backwall is still a clean reference.

Typical metal frequencies in general industry:

• 1 to 2.25 MHz — thick, attenuative, or coarse-grained product; some castings.
• 2.25 to 5 MHz — the workhorse band for steel plate, pipe, and many welds.
• 10 MHz and up — thin wall, fine-grained wrought product, immersion detail work, some bondlines.

The written procedure names the frequency. "Always use 10 MHz for resolution" is not a procedure.

Damping and bandwidth

The crystal wants to ring. A backing (tungsten-loaded epoxy is the usual picture) damps that ring.

A narrowband probe is a penetration and signal-to-grass tool. A broadband probe is a thickness and near-surface tool. The same nominal 5 MHz stamp can be either. If the stem says "highly damped" or "broadband," it is telling you the pulse is short. If it says "lightly damped" or "high sensitivity," it is telling you the probe rings.

Damping is also why a thickness-gage dual (below) can read a 1 mm remaining wall that a ringing single-element 0° probe cannot: the dead zone shrank.

Single-element, dual-element, and delay-line probes

Single-element pulse-echo uses one crystal to send and receive. It is the default 0° contact probe and the crystal inside most angle-beam wedges. Limits:

• The main bang and crystal ringing occupy the first microseconds. Thin wall hides in that dead zone.
• The near field sits in the part if there is no delay.
• Good for thicker plate, backwall-referenced lamination scans, and any job where the first useful echo is well after the initial pulse.

Dual-element (twin crystal, TR, T/R) puts a transmitter and a receiver in one housing, separated by an acoustic barrier, each on a slight roof angle so their beams cross a short distance below the wear face.

• The receiver is isolated from the main bang, so the dead zone collapses.
• The crossover zone is a weak focus — excellent for remaining-wall and corrosion pits, and for laminations just under the surface.
• Below the crossover the two beams diverge and sensitivity falls; duals are not automatically better on 100 mm plate.
• Duals are the default on many digital thickness gages.

Delay-line probes put a plastic stand-off (polystyrene, acrylic, or a replaceable button) between the crystal and the part.

• The interface echo (end of the delay) becomes the time-zero you actually use. The main bang is off to the left of the useful screen.
• Thin sheet, small-diameter bar, and precision thickness work become possible in pulse-echo.
• The delay also protects the crystal from heat and abrasion. Replace the shoe, not the probe, when the face wears.
• The near field can be designed to live in the delay, so the steel is in a cleaner field.

Do not confuse a delay line with a wedge. A 0° delay is still straight-beam. A wedge is cut to produce an incident angle (13.3).

Contact versus immersion

Contact testing presses the wear face against the part with a film of couplant. It is portable, fast on large surfaces, and the default for field welds and manual thickness. It is also slave to surface condition: roughness, scale, paint, curvature, and temperature all change the transmitted amplitude. That is why a calibration on a smooth IIW block does not automatically transfer to a pitted tank floor without a procedure that says it does.

Immersion testing puts the part (or a squirter column) in water. The water path is a long, uniform delay:

• The interface echo (water → metal) is the front-surface mark.
• The crystal never touches the part. Curved and rough parts become inspectable with a manipulator.
• Focusing lenses and shaped water paths are practical.
• Through-transmission with two squirters is a standard plate and composite setup.
• The water is the couplant. Bubbles, temperature, and wetting agents are process-control items.

Immersion does not magically create shear in the water. The tank path is longitudinal. Shear in the metal, if required, is refracted at the water–metal surface the same way a wedge refracts — Snell's law with v₁ = water (13.3).

Bubblers and squirters are immersion physics without the tank: a flowing water column is the delay and the couplant. Treat them as immersion for exam purposes.

Couplant

Couplant has one job: displace the air film so the pulse is not staring at a steel–air mirror (13.1). Everything else is preference and procedure.

Couplant does not magnetize the crystal, set the frequency, or replace a calibration block. A thicker couplant layer is a short delay and a possible attenuator; on a precision thickness job it is an error source. On a coarse lamination scan a generous film is a gift.

Surface prep is part of coupling. Loose scale, heavy paint, and weld spatter are air pockets. A procedure may allow inspection through a stated coating thickness after a transfer correction. A procedure that says "bare metal" means bare metal. The Level II does not invent a gel that "soaks through" mill scale.

Straight-beam pulse-echo thickness and the backwall

Straight-beam means the beam axis is normal to the entry surface — 0° in the usual shop language. The workhorse display is pulse-echo:

T = v Δt / 2

T is thickness (or depth of a mid-wall reflector). v is the longitudinal velocity of that material unless the stem says you are running contact shear. Δt is the round-trip time from the entry surface (or the interface echo, or the calibrated zero) to the echo you are gating. The 2 is the trip down and back.

Worked example: thickness

A calibrated 0° probe measures a round-trip time Δt = 8.6 µs to the backwall. The procedure velocity for this carbon steel is the typical 0.233 in/µs.

T = 0.233 × 8.6 / 2 = 2.004 / 2 = 1.00 in.

Same arithmetic in SI. v = 5900 m/s = 5.90 mm/µs, Δt = 8.48 µs:

T = 5.90 × 8.48 / 2 = 50.0 / 2 = 25.0 mm.

If you forget the 2, you report double the wall. If you use a shear velocity on a longitudinal backwall, you report a wall that is about 55% of truth (0.128/0.233). Both errors are classic general-exam traps.

Zero the clock the way the procedure says:

• Single-element contact: the wear-face / entry-surface zero after you have accounted for the wear plate. A calibration block of known thickness sets v and the zero together.
• Delay-line or immersion: time from the interface echo to the backwall, not from the main bang.
• Dual / thickness gage: the instrument is usually internally zeroed for the dual path; you still verify on a step block.

Multiples (first backwall, second backwall, third) should be equally spaced if the plate is parallel and the velocity is constant. A second backwall at twice the time of the first is a useful sanity check. If the multiples are not even, you have the wrong zero, the wrong mode, a taper, or a mid-wall reflector you mislabeled.

Lamination scanning and the backwall as reference

A lamination in plate or a stringer in bar is a planar discontinuity parallel to the rolled surface — exactly the orientation a 0° beam wants. The A-scan signature the general exam expects:

• An echo between the entry surface and the backwall, at a time that matches the depth of the plane.
• Loss, breakup, or collapse of the backwall under that same index position, because energy never reached the far surface (or reached it only around the edges of the plane).
• On a total lamination, the mid-wall echo can look like a new, thinner backwall. Thickness mode will happily report half-thickness if you let it gate the first big echo.

The backwall is the reference in 0° work:

• Time reference — thickness and depth.
• Amplitude reference — a sudden drop in backwall height, even without a clean mid-wall spike, is a reason to stop and map (rough back surface, porosity cluster, large inclusion, coupling loss). Coupling loss usually kills both the grass and the backwall as you lift the probe. A lamination kills the backwall while the probe is still coupled and often adds a mid-wall echo.
• Scan integrity reference — if the backwall is not there on sound plate, you are not inspecting. Fix coupling, angle, or the instrument before you call the plate clean.

A 0° beam is deaf to a tight crack that is parallel to the beam (a vertical seam in plate scanned from the rolling face). That is not a lamination problem; that is why angle-beam exists (13.3) and why method selection in Chapter 3 asked about orientation.

Through-transmission conceptually

Through-transmission places a transmitter on one face and a receiver on the opposite face. The quantity of interest is the received amplitude (and sometimes the transit time), not a pulse-echo A-scan of depth.

• A lamination, a disbond, or a porosity field shadows the receiver. The signal drops.
• You do not get a depth from a single through-transmission pair. The defect is somewhere on the line between the two probes.
• It is the right picture for automated plate testers, some composites, and attenuative product where a backwall echo would be gone anyway.
• Immersion through-transmission with two squirters is the same idea with water delays on both sides.

Pulse-echo and through-transmission answer different questions. Pulse-echo: where is the reflector, and how loud is it versus a backwall or a DAC? Through-transmission: did enough energy survive the path? A candidate who says through-transmission "measures thickness better than pulse-echo because there is no divide-by-two" has missed the point. You can compute thickness from one-way time if you know v, but the method's industrial value is loss of transmission, not a better micrometer.

Realistic exam scenarios

A 0.08 in sheet must be thickness-gaged. A ringing 2.25 MHz single-element contact probe shows only the main bang. The probe that belongs on the procedure is a dual or a delay-line (or immersion) at a higher frequency, not "more gain on the same main bang."

A Level II turns frequency from 2.25 MHz to 10 MHz on a 180 mm coarse forging "for definition." The backwall disappears into scatter. Penetration lost; the forging did not get thinner. Go back down in frequency or change methods.

A contact scan of plate shows a mid-wall echo at half the backwall time and no backwall under the probe. That is the textbook lamination (or a mid-wall unbond). Check coupling on adjacent sound plate — the backwall should return — then map the area.

A dry probe on painted tank floor reads random thickness. Couplant through an allowed coating, or strip the coating, per the procedure. Do not average garbage.

An immersion stem asks which mode travels in the 75 mm water path. Longitudinal. The 45° shear, if any, starts at the water–steel interface.

A through-transmission squirter pair loses 20 dB over a 200 mm patch and the operator reports "a lamination at 6 mm depth." Through-transmission did not give that depth. Switch to pulse-echo or a different view to locate the plane.

What transducer and straight-beam items are really testing

If the stem names frequency, ask resolution versus penetration and write λ = v/f. If it names damping, ask pulse length versus energy. If it names dual, delay, or immersion, ask where the dead zone and the near field went. If it names couplant, say displace air. If it names a 0° A-scan, write T = vΔt/2 and treat the backwall as the reference; a mid-wall echo plus a lost backwall is a lamination until proven otherwise. If it names two probes on opposite faces, you are in through-transmission and you do not have a depth. Those six questions are the hardware half of official UT topic 1.