Testing Insights

How to choose tensile tester capacity? 10kN or 100kN

Algorithm first: max test force ≈ material tensile strength × original cross-sectional area, plus 20%–30% margin. But the real buying mistake isn't "too big" — it's only looking at the upper limit, not the lower. Pull a 5kN thin sheet on a 100kN machine and the data is still untrustworthy.

Step 1: Calculate the maximum force you actually need

Take R_m as the highest among all your materials, and S₀ as your largest common sample cross-section. Multiply and add margin — that's the capacity tier to choose.

Values above are for selection estimation; actual per material cert and GB/T 228.1 sample dimensions.

Step 2: Make sure small samples also fall in range

Force error is assessed over the measuring range (GB/T 16825.1 / ISO 7500-1). The controlled interval of typical machines is usually 20%–100% of full scale; below that, indication error is no longer covered by the calibration certificate.

In one line: the fracture force of your smallest common sample should not be below 20% of the chosen capacity.

• 100 kN machine on 7.5 kN Al plate → only 7.5%, out of range, error may exceed ±1%;
• same machine on 47 kN 45 steel → 47%, normal;
• conclusion: this machine can pull steel but not thin Al plate — not "a big machine does everything".

Step 3: Don't miss these four items

• Test type: tension only? or compression, bending, shear, peel? Bending needs 3/4-point bending fixtures and larger test space.
• Yield & modulus?: to measure R_p0.2, R_eL or E, you must have an extensometer (sized by gauge length). Without it, yield is only back-calculated from crosshead displacement — not accepted.
• Speed control: GB/T 228.1 / ISO 6892-1 recommends strain-rate control (Method A), requiring closed-loop strain control — ask before buying.
• Frame rigidity & space: large-tonnage machines deform the frame during small-force tests, mixing into displacement; also confirm the largest sample (with fixture) fits the test space.

Selection checklist (send us and we'll design)

• material list with each highest tensile strength;
• sample form: round / plate / wire / foil / finished part, and size range;
• which metrics: R_m, R_p0.2, A, Z, E, n;
• standard: GB/T 228.1, ISO 6892-1, ASTM E8 or in-house;
• need thermal chamber, extensometer, special fixtures;
• need LIMS / fixed-format report export.

OES grade ID — the most error-prone sample-prep steps

Experience: of questionable OES results, 70% come from sample prep, not the instrument. The spectrometer only reads the point you give it — whether that point is clean, flat and the base metal itself, the instrument can't judge.

Mistake 1: Exciting with scale / coating / plating still on

Spark only burns tens of microns deep. If the surface is oxide scale, paint, zinc plating or carburized layer, you're not measuring base composition. Fix: grind or mill off 0.5–1 mm+ to expose fresh base metal; coated plates must have coating fully removed first.

Mistake 2: Mixing grinding wheels / paper → cross-contamination

Paper used on steel then on aluminum carries Fe in, inflating Al's Fe content. Fix: color-code abrasives by base metal — Fe, Al, Cu, Ti each get their own paper/wheel; a dedicated grinder is better; idle-grind a few seconds when switching.

Mistake 3: Wrong surface roughness

• Too rough (visible scratches): poor seal, argon leak → diffuse discharge, scattered data, light elements C/S/P especially unstable;
• Too smooth (mirror): soft metals like Al get abrasive pressed in during polish, also contaminating;
• recommended: uniform fine striations (Ra ~0.8–1.6 μm); dedicated grinder or 40–80 grit alumina wheel/belt in one pass is stablest.

Mistake 4: Sample too thin / too small

Thin samples dissipate heat poorly → local overheating or even burn-through; backside scale also discharges. Generally: flat excitation face, area fully covering the excitation hole (common Ø12–16 mm), thickness ≥ 3 mm; thin plate, wire, small parts use dedicated small-sample fixtures or mounting, noted in method.

Mistake 5: Random excitation-point position

• ≥ 3–5 mm from sample edge, else argon seal fails;
• avoid porosity, sand holes, inclusions, cracks, welds and obvious segregation;
• space between points; no re-striking on old spots;
• tubes, round bars (curved) must be machined flat before measuring — curved face can't seal;
• take 2–3 points per sample, drop outliers, average.

Mistake 6: Fingers, oil and moisture

Fingerprints bring C, H, O contamination; moisture makes discharge unstable. No touching the excitation face; if needed clean with anhydrous ethanol or acetone and fully blow dry before measuring.

Mistake 7: Argon & standardization ignored

• argon purity ≥ 99.999%; insufficient → white spot, drifting data;
• after startup, bottle change, big temp/humidity change, before continuous runs → do standardization;
• for wide base/grade spans, use type standardization (control samples) — far more accurate than global-only.

How to read the spark spot: a diagnosis table

Common method standards: steel per GB/T 4336, stainless per GB/T 11170, Al & alloys per GB/T 7999; report per the current valid edition.

Metallography five-step method: cut, mount, polish, etch, observe

Prep quality decides the reading. One line for the whole flow: cold cutting, edge-protected mounting, clean polishing, light etching, right observation mode. 80% of bad metallography photos come from step 3 — polishing.

1. Cutting: representative position, low temperature

• define purpose first: banding → longitudinal; grain/decarburization → transverse; failure must include crack source;
• wheel rule: hard material → soft wheel, soft material → hard wheel, avoid clogging & overheating;
• adequate cooling mandatory — overheating changes local structure (re-quench, temper), making the reading completely wrong;
• after cutting, remove another 1–2 mm inward to leave the cutting HAZ outside.

2. Mounting: when to use cold mount

• Hot mount (phenolic / epoxy, ~150–180°C, pressure, minutes): fast, hard, for high-volume regular samples;
• Cold mount (epoxy / acrylic, room-temp cure): for heat-sensitive, coated, porous, tiny or irregular parts;
• measuring case depth, coating thickness, decarburization → must use edge-protecting filler (glass-fiber/ceramic) or dedicated edge-clamp, else edge rounds and depth reads low.

3. Polishing: four-stage progression, clean between

• each step rotate 90°, polish until previous scratches gone;
• ultrasonic clean between steps — one leftover coarse grit is the primary source of a through-scratch;
• polish pressure & time: smaller is safer — high pressure → deep deformation layer, relief;
• cast iron, Al (soft) prone to comet tail; final light polish, shorter time helps.

4. Etching: match reagent, converge time

• Carbon / low-alloy steel: 2%–4% Nital, seconds to ~15 s;
• Stainless: aqua regia, ferric chloride-HCl, or 10% oxalic electrolytic;
• Aluminum: Keller (HF + HCl + HNO₃), 10–30 s;
• Titanium: Kroll;
• rule: shallow over deep — shallow can be added, deep means back to polish; after etch immediately water → alcohol → blow dry, one delay and you get an oxide tint.

5. Observe: four modes each cover a range

• Bright field: default for general structure, grain size, phase composition;
• Dark field: grain boundaries, fine inclusions, micro-relief on polished face — higher sensitivity;
• Polarized: graphite spheroids, non-metallic inclusions, coatings, anisotropic structures;
• DIC: relief without etching — for coating, case, post-surface-treatment cross-sections.

Common rating standards: grain size GB/T 6394 (≈ ASTM E112, comparison/interception), non-metallic inclusions GB/T 10561 (≈ ASTM E45), carburized hardened layer GB/T 9450, induction/flame hardened layer GB/T 5617, decarburization GB/T 224, microstructure general rules GB/T 13298.

Seven common defects: symptom → cause → fix

What does an ONH analyzer measure? The quality issues behind O, N and H

Your mill cert lists C, Si and Mn — but O, N and H are often missing. Yet flake cracking (shatter cracks), strain-aging embrittlement and early fatigue failure are usually caused by exactly these three "invisible elements". They sit at ppm level, beyond conventional OES — only an oxygen-nitrogen-hydrogen (ONH) analyzer can measure them.

First, know what each element wrecks

• H (hydrogen): the culprit behind flakes (shatter cracks). Large forgings, spring steel, line-pipe steel and gear steel are most hydrogen-sensitive — hydrogen-charged steel cracks late, under combined stress and microstructural transformation, often days or even weeks after quenching;
• N (nitrogen): the driver of strain aging. When low-carbon sheet is stamped, Lüders bands (slip lines) appear on the surface and scrap rates climb. In stainless steel nitrogen is actually good (solid-solution strengthening), but poor control during welding causes porosity;
• O (oxygen): present as oxide inclusions, directly cutting fatigue life and cleanliness ratings. Oxygen content in bearing and spring steels is a must-check item for high-end customers.

The principle, in one paragraph

The sample goes into a graphite crucible and is pulse-heated above 2500 °C in inert gas until it melts: oxygen reacts with crucible carbon to form CO/CO₂, quantified by an infrared detector; nitrogen and hydrogen are released as N₂ and H₂, separated and measured by a thermal-conductivity detector. One machine, one fusion, three elements — all within a few minutes.

Ranges & precision: how to read the table

Using the model we deliver (Nake ONH 3500) as an example:

Analysis time is within about 3 minutes, fully automatic. Actual values are governed by the instrument calibration certificate and current standards.

Sampling & prep: at ppm level, contamination is error

• Surface condition: remove scale, oil and moisture. O and H are extremely surface-sensitive — sweat injects hydrogen directly. Gloves and tweezers from sampling to testing, no exceptions;
• Sample form: typically a block (~0.5–1 g) with clean, pore-free surface; chips and powders need dedicated crucibles and methods;
• Test hydrogen promptly: hydrogen diffuses out. Weld seams, deposited metal and pickled parts should be tested as soon as possible or stored cold — wait two weeks and half the hydrogen may be gone;
• Flux & blank: tungsten-tin flux and crucibles carry background blanks; blank drift eats ppm-level resolution — run blank and CRM calibration first thing every day;
• Control samples: type-calibrate with reference materials of similar composition — far better than instrument-only calibration at suppressing matrix effects.

Which industries really need an ONH

• Producers of bearing, spring and gear steel (full O/N/H control + customer audits);
• Welding consumables and welding procedure qualification (diffusible hydrogen in deposited metal is a mandatory test);
• Titanium alloys, powder metallurgy, cemented carbides (H and O define product grade);
• Third-party labs taking outsourced testing from steel mills.

HRC, HRB or HV? Choosing a hardness scale and the limits of conversion

Hardness is not one physical quantity — it is a family of test methods for resistance to indentation. On the same material, HRC 45 and HV 450 can both be "correct" at once. The two real mistakes are using a scale outside its valid range, and taking a conversion table as a measurement.

How the three families divide the work

Valid ranges at a glance

• HRC valid 20–70: below 20 switch to HRB or HBW; above 70, to HRA or HV;
• HRB valid 20–100: for medium-low hardness materials such as annealed steel, copper and aluminum alloys;
• Minimum specimen thickness: about 10× the indentation depth — visible deformation on the back voids the data. Thin sheet and cases (typically 0.1–1.5 mm) are punctured by a single HRC indent; use superficial Rockwell (HRN/HRT) or low-force Vickers instead;
• Cylindrical & spherical surfaces: curvature affects readings. HRC on a convex cylinder needs the standard curvature correction; spheres should not be tested directly — align round bars with a V-anvil.

Conversion tables: reference only, never a result

Hardness conversion tables for ferrous metals (GB/T 1172, ASTM E140) are statistical correlations, reliable only within the specified material class and hardness interval. Three practical boundaries:

• The drawing calls HV for referee testing but you measured HRC and converted — it does not count in arbitration; the specified scale must be measured directly;
• Cross-family conversion error can reach ±30 HV; near acceptance limits (e.g. ≥ HV 600) never conclude from a converted value;
• Special materials like cast iron and austenitic stainless steel need their own tables — applying the general steel table doubles the error.

The five easiest field mistakes

• Indentation spacing too small: centers closer than 3–4× indentation diameter, or nearer than 2.5–3× to an edge — stress fields overlap and readings distort;
• Anvil and indenter not concentric: dirty anvil or a rocking specimen makes Rockwell readings jump — run an intermediate check on a reference block before each batch;
• Punching through the case: a carburized part with 0.8 mm case vs an HRC indentation 0.1–0.2 mm deep — add the soft core's influence and readings read low. For effective case depth, use a Vickers gradient (GB/T 9450);
• Temperature & vibration: standards require 10–35 °C (precision grade 23 ± 5 °C); a station facing a punch press gives unstable readings — the machine is not broken;
• No verification on direct-reading testers: after every scale change, anvil change, or before/after large batch runs, test a reference block — the single most common audit nonconformity here.

Selection decision: four steps to a scale

• ① Thickness & case: solid stock works down the list; thin sheet/layers go straight to Vickers or superficial Rockwell;
• ② Hardness interval: match against the ranges above — which scale's valid window does it fall into;
• ③ Surface condition: rough cast/forged surfaces suit Brinell; machined faces and finished parts suit Rockwell;
• ④ Customer & standard: whatever the drawing, company spec or audit requires — measure that; conversion is for internal reference only.