Knife Edge Retention Testing: How to Read CATRA Data

The ranking criteria we actually use

For product managers, the right knife edge retention test is not the test with the fattest lab invoice. It is the one that lets you sign off the spec without guessing. Are you approving a new steel after 58-60 HRC readings on the Rockwell tester? Checking heat treatment drift from Batch A to Batch C? Comparing a 15 degree chef knife with a 20 degree outdoor knife after the grinding line changed from a worn 400# belt to a fresh 600# belt? Different jobs. “Which test is best” is the wrong question to ask.

We rank the options below by five buyer criteria: repeatability across production lots, so Batch A and Batch C do not tell two different stories; real cost per sample, including the courier charge; report speed, such as 12 days vs 18 days from sample cutting; fit with the customer's actual use; and whether the result can be written cleanly into a supplier agreement. A test that gives a neat number but cannot be repeated on a 3000 pcs reorder is not a QC system. A test that feels realistic but depends on one operator's wrist pressure is too weak for a packaging claim. We have seen this go sideways: QC pulled the sample after final buffing and found the left side edge 2 degrees wider than the approved pre-production knife.

CATRA, short for Cutlery and Allied Trades Research Association, uses standardized abrasive paper and a fixed cutting cycle to measure how much material a blade cuts over time. The common output is TCC, total card cut. Some reports also show ICP, initial cutting performance, with cutting curves across repeated strokes. A good CATRA test knife report should name the blade type, edge angle, sharpening grit, HRC, sample quantity, and whether the knife was tested as received or resharpened before testing. We ask for those fields because one buyer flagged a report where “15°” was typed on the PO, but the sample sent from the packing table measured closer to 18° per side on our angle gauge.

At our Yangjiang, China factory, normal export production for kitchen and outdoor knives runs around 180,000-220,000 units per month depending on mix. That volume teaches a hard lesson: one strong lab number does not protect you if the production edge varies by 3 degrees per side or the final buffing overheats the apex. The math doesn't work. We run spot checks with an angle gauge and 20x loupe at the grinding line because a sharp-looking edge can lose retention before the carton leaves the warehouse.

Best overall: CATRA TCC for launch approval

CATRA TCC is our first pick when a buyer needs one number to approve a new SKU, mainly for chef knives and EDC knives where the cutting claim goes on the blister card or color box. One number sells. One number also gets abused. We run TCC after pilot sharpening, with the bevel checked on the angle gauge and the behind-edge thickness recorded in mm, usually at 3 points per blade with a digital caliper. No eyeballing. If the geometry and test setup stay fixed, the result tells the product manager whether the gain came from steel or heat treat, not better catalogue wording.

The catch is simple. TCC is not a universal ranking for every knife on the rack. A thin 180 mm chef knife at 15 degrees per side will bite into the test media faster than a thick hunting knife at 22 degrees per side. That does not make the hunting knife worse; it was built for twisting cuts and edge stability in rough use. We have seen this go sideways: one buyer put both samples on the same spreadsheet and asked why the outdoor knife “failed.” Wrong question. The buyer flagged it in red on the comparison sheet, but the two samples had different behind-edge thickness by 0.35 mm. Without controlled geometry, the CATRA number becomes sales spin.

For OEM development, we recommend testing at least three samples per construction after pilot sharpening. Five samples are better if the order is large or the claim is going onto retail packaging. Record the steel grade and heat treatment band, then record the blade thickness behind the edge, bevel angle, sharpening belt or wheel grit, and final stropping method with the same discipline QC uses on an AQL sheet. The grinding line should write down whether it ran a 400 grit belt or a felt wheel with compound, because that small change can move the result by enough to start an argument in sample approval. We have had a PO typo list “satin edge” when the approved sample used mirror stropping. Painful. For example, a 5Cr15MoV kitchen knife at 55-57 HRC should not be judged against a 10Cr15CoMoV knife at 59-61 HRC unless the price target allows that upgrade.

Use TCC to choose steel, approve the edge angle, and check whether a Damascus pattern knife cuts like its core steel should. Do not use one unusually high TCC number as proof that every container will cut the same. The math does not work. QC pulled the sample from one pilot batch; mass production still needs a control plan for HRC, edge angle, and behind-edge thickness before we ship 3,000 sets. We run that control with hardness readings on the Rockwell tester and angle checks at the sharpening station, not after the cartons are sealed.

Best QC control: CATRA curves and ICP

For edge-retention QC, the CATRA curve tells us more than the headline TCC number. ICP is the first bite. TCC is the total paper cut across the full run. The gap and slope between those two points show how fast the apex loses bite after the first paper stack, which is where weak grinding shows up. We once ran two 8-inch chef knives that finished close on TCC, but they split badly in the first 20-40 cycles; QC pulled the sample, checked the apex under a 50x scope, and the buyer flagged the weak first cut before anyone cared about total life.

If the complaint only says "edge retention failed," this is the wrong question to ask. Was the knife dull out of the box, or did batch B fail to match batch A from the grinding line? A high TCC with poor ICP often means the edge is tough but not keen enough, commonly from a fat final bevel measuring 1-2 degrees above spec on the goniometer. A high ICP with a steep drop points to a thin apex or an over-buffed edge with a fragile burr. Looks good first. Then it dies.

A practical CATRA QC agreement should lock the control points before we run mass production. We set sample count and HRC band from the golden sample, then write the measured edge angle on the QC sheet before packing starts. Below is the sourcing table we use as a starting point, then adjust by knife type and target price; the math does not work if the buyer asks for premium CATRA curves on a 1,000 pcs MOQ promo knife with no lab time in the schedule.

If CATRA belongs in your supplier agreement, write minimum TCC plus a curve or ICP requirement. We ship against the PO, and we have seen this go sideways when the PO says "good sharpness" but the buyer expects a clean first-cut curve from cycle 1. One typo matters. If purchasing writes "sharpness ok" while engineering expects ICP data, you can get a passable total number with a poor first-cut experience.

Best factory screen: controlled cardboard cutting

A controlled cardboard test is not CATRA, but it is the best low-cost factory screen we run on production lots. Fast and cheap. It matches the complaint retailers send back, such as “won’t bite tomato skin after one week,” and it catches bad sharpening before packing: folded burrs from the buffing wheel, or heat-treat drift when HRC checks move 1-2 points. For importers buying 3,000-20,000 pieces per SKU, the math does not work if every lot waits for a lab report; we usually need the carton seal date held to 12 days vs 18 days. On the grinding line, QC can pull 50 knives after the #800 belt and find trouble before the blades reach the polybag table.

The key word is controlled. Cutting scrap cartons on the warehouse floor proves almost nothing. A usable cardboard protocol should lock the corrugated grade and thickness in mm, then spell out strip width, cutting length, blade angle, stroke count, and the pass/fail sample photo taped beside the workbench. For example, a 200 mm chef knife might cut 10 mm wide corrugated cardboard strips for 60 strokes, then shave paper or slice 80 gsm printer paper cleanly. A pocket knife needs its own setup because the shorter blade and steeper bevel change the feel; using the chef-knife standard on it is the wrong question to ask.

We often use this as an in-process check after final sharpening and before cleaning. If 5 out of 50 sampled knives show folded burrs or fast loss of bite, the batch should stop for edge inspection. Stop it early. QC pulled the sample, marked the heel and tip with a black Sharpie, then checked the edge under a 10x loupe because most failures show up in the last 30 mm near the tip. The fix is usually a belt change or lower buffing pressure, not a new steel story for the buyer. We have seen this go sideways when a line leader keeps polishing to make the edge look bright.

The weak point is operator variation. Hand pressure changes the result. That is why cardboard cutting should support CATRA data, not replace it. It is also why a buyer should ask for a short test video with the sampling record when dealing with a new knife OEM in China. Ask to see the angle jig, the cardboard label, and the lot number on the inspection sheet; one buyer once flagged a PO typo where “60 strokes” became “600 strokes,” and the whole comparison became useless. Video will not make the test scientific, but it can expose sloppy technique.

Best field check: food and rope trials

Food prep trials and rope cutting work best when the buyer is asking about hand feel, not lab wear alone. A chef knife can post a clean CATRA result and still get complaints if the blade wedges in 25 mm carrot sticks, sticks in potatoes, or feels nose-heavy after 80 repeated cuts on a PP board. It happens. Last month the buyer flagged a 210 mm sample that cut paper fine, but the grinding line left too much steel behind the edge: 0.65 mm measured 1 mm above the apex with a digital caliper. A hunting or tactical knife can show decent TCC, then lose the buyer when QC pulled the sample and found micro-chips after 30 cuts on 10 mm rope, 12 zip ties, and a short light-wood shaving check.

For kitchen knives, we run a basic trial set: tomato skin for bite; onion halves and 25 mm carrot sticks for steering and wedging; raw chicken for cleanup; 3 mm cardboard after washing for quick edge loss. Tomato is simple. It tells us if the apex is still keen without hiding behind a fancy spec sheet. Onion and carrot show blade geometry, especially wedging behind the edge at 0.3 mm to 0.5 mm. Chicken plus washing exposes rust spots near the bolster and handle hygiene problems, which one EU buyer flagged on a satin-finish sample after our inspector wiped it with a white cloth and saw grey residue. For outdoor knives, 10 mm sisal rope and nylon webbing fit the job better than a tomato test; cardboard and light wood feathering show whether the edge chips or just dulls. These are not ASTM-level universal tests. They match how returns land on our desk.

The buyer mistake is treating field trials like a clean ranking without controls. Wrong question. If one tester cuts on a PE board and another cuts on a stainless table, the data is junk by cut 20. If one knife is factory fresh and another has already been through a photo shoot, the result is biased before the first slice. Write down the cut count, material source with batch if possible, cutting surface, cleaning method, and final sharpness check. We use a one-page pass/fail sheet with edge notes in mm, plus the sample code from the PO, because we once had two “black G10 handle” versions mixed after a packing table relabel. The PO typo was small: G10-BK-2 became G10-BK-Ⅱ. Memory turns soft after 6 samples.

This option ranks below CATRA because repeatability is weaker. It ranks above hardness-only checks because it catches geometry, handle comfort, and finish problems that HRC numbers miss. For a private label program, we run one field trial at prototype stage, then CATRA on the final approved sample. The math doesn't work if you pay for lab testing on every handle tweak; we have seen a buyer spend 12 days changing scale texture, then ask why the lab date moved to day 18. That mix gives you customer relevance and a lab reference the buyer can defend in a sourcing meeting.

Best predictor: hardness and edge geometry audits

Hardness and geometry audits are not edge retention tests. They are the best lot-by-lot predictors we can run without delaying a vessel booking from Yantian by 7 days. If the HRC band shifts, the edge angle opens up, or the thickness behind the edge grows by 0.20 mm, the CATRA result will move too. Buyers ask first about steel grade. Wrong question. We have seen a clean 8Cr13MoV blade at 58-59 HRC, sharpened 17 degree per side on the grinding line and checked under a 20x loupe, beat a higher-grade steel knife with a fat shoulder and leftover burr during prep tests on onions and tomato skin.

For hardness, use Rockwell C testing on coupons or hidden blade positions where possible. Finished blades are tricky because thickness, curvature, and surface finish can push the reading off by 1-2 HRC, so the factory needs a written test point and load method, not a guess from the sales desk. We run spot checks with a Rockwell tester before final packing. QC pulls the sample when the first 50 pcs come off heat treatment, then marks the blade map so the inspector is not testing the same easy flat every time. For kitchen knives, common bands include 54-56 HRC for low-cost stainless, 56-58 HRC for 5Cr15MoV and similar mass retail knives, 58-60 HRC for better molybdenum-vanadium stainless, and 60-62 HRC for premium powder or high-carbon cores where chipping risk is controlled.

Geometry needs magnification, angle gauges, or an optical system if the line has one. At minimum, record blade thickness 1 mm above the edge, primary bevel angle, and burr condition: wire burr, rolled burr, or clean apex. Small numbers matter. A 0.25 mm thickness behind edge cuts nothing like 0.55 mm, even if both knives carry the same HRC and steel name on the PO. We use a digital caliper with a thin-jaw tip and recheck doubtful samples under the loupe. For a German-style chef knife, thicker geometry can pass. For a Japanese-style santoku, the buyer flagged it last year when the shoulder measured 0.48 mm and the sample felt dead on tomatoes.

At TANGFORGE, MOQ for custom kitchen knife projects usually starts around 600-1,000 pieces per model, depending on handle, steel, and packaging. At that scale, paying for CATRA on every shipment makes the math break, especially on retail programs with 2-3 color boxes per model and a 12-day packing window vs 18 days if we wait for outside lab results. We have seen this go sideways when the PO said “15 degree” but the approved sample was actually 18 degree per side. Require hardness and edge geometry records on each production lot from Yangjiang or any other China knife base; we ship cleaner when the inspection sheet shows HRC, angle, and thickness before cartons are sealed.

Best claim policy: cautious numbers on packaging

Packaging claims are the easiest place to misuse CATRA data. One prototype with a strong TCC result does not support “stays sharp 5 times longer” on a color box unless the comparison knife, CATRA method, sample count, and confidence level are written on the test request. We had a buyer flag this exact line after QC pulled only 3 pre-production samples from the grinding line; one blade showed a 0.3 mm edge-height difference after regrind. Bad start. In Europe and North America, loose superiority claims cause retailer and marketplace trouble even when the knife is well made. We’ve seen this go sideways.

A safer claim policy is specific and modest. Say the edge was validated by CATRA testing under a named internal specification. State the steel, HRC band, and edge angle only if production can hold them, for example 58-60 HRC with a 15° per side edge checked by angle gauge at the sharpening station. Use claims like “tested for edge retention” or “CATRA-tested cutting performance” when the report is in the file. Percentage claims need discipline: at least 5 samples, a fixed baseline knife, and the same blade geometry, with the grinding line running the approved belt grit. Otherwise the math does not work.

For QC managers, the better place for the number is often the technical file, not the front of the box. Keep the CATRA report with sample photos, the approved sample record with edge-angle data, the steel certificate, the heat treatment log, the inspection report, and REACH, LFGB, or FDA documentation where applicable. We ship retail programs where one PO typo changed “X50CrMoV15” to “5Cr15,” and that small error made the packaging claim harder to defend during artwork approval. For Amazon or retail programs, match the claim to FNSKU-level product records so a later material change does not leave old copy online.

The most defensible approach is boring: define your approved CATRA test knife, lock the sharpening process, set a minimum TCC or curve requirement, and audit production with AQL 2.5 inspection plus functional cut checks on pulled cartons. Short claim. Clean file. On the floor, that means the same belt grit, the same final honing wheel, and a signed first-piece approval before mass sharpening starts, usually before the first 500 pcs move to packing. It will not sound exciting in a sales meeting, but it is how we avoid returns caused by overpromising.

Frequently asked questions