November 18, 2026 · Technical Guide
Laser Weld Inspection: Penetration, Porosity and Acceptance Criteria (2026)
A penetration number on a datasheet means nothing until you say where it is measured. Three 1500W handheld welders on the market list 2.5 mm, 4 mm and 5 mm on stainless steel. Same power, same material. Laser weld inspection starts by agreeing on a definition and a section photo, before anyone compares machines.
Last month a buyer sent me three quotes for a 1500W handheld laser welder and asked which supplier was honest.
The penetration numbers were 2.5 mm, 4 mm and 5 mm on stainless steel. Same power, same wavelength band, same material. They cannot all be right.
None of the three is lying. They measured different things and used the same word for it. That is the whole problem, and it is also why laser weld inspection has to start with a drawing, not a machine.
The Three Numbers, Side by Side
| Model | Power | Claimed stainless penetration | Where it is printed |
|---|---|---|---|
| Harsle HW-1500 | 1500 W | 2.5 mm | HW-1500 product table |
| Dato DT-1500HJ | 1500 W | up to 4 mm | DT-HJ parameter table |
| Raymax handheld | 1500 W | 5 mm | 1500W thickness table |
Read the table twice. A 25 percent jump in power buys a 25 percent jump in claimed depth on one sheet, while another supplier claims nearly double the depth at the same 1500W. The third lists 0.3 to 6 mm as a working range and then 5 mm in its own thickness table.
This is not a scandal. It is a missing method. A depth number without a gas, a travel speed, a fit-up gap and a section photo is a marketing sentence.
Penetration Has Three Different Meanings
Before you compare quotes, decide which one you are buying. On the machines we build, the handheld laser welding machine range, the sample report states the definition right next to the number, because a depth without a reference face cannot be checked by anyone.
| Definition | Measured from | What it decides |
|---|---|---|
| Depth from the weld face | Top surface down to the deepest fused point | What most datasheets quote |
| Fused depth at the interface | Length of the joined contact line on a lap joint | Strength of the joint |
| Full penetration | Melt zone reaches the root | Leak-tight and sealed parts |
- Depth from the weld face. How far the melted zone reaches below the top surface. Most handheld datasheets print this one because a microscope can reach it easily.
- Fused depth at the interface. On a lap joint, how much of the contact line between the two sheets actually joined. This is the number that decides whether the part holds.
- Full penetration. The melt zone reaches the root and the seam is continuous through the thickness. Required for sealed parts, risky on thin sheet.
On a 2 mm lap joint, a 1.8 mm depth from the face can still mean 0.6 mm of real interface fusion if the sheets sat 0.5 mm apart. The bead looks deep and the joint is weak. That is the most common failure I see in sample reports, and no photo of the bead will reveal it.
The Standards You Can Cite
Laser welds do not use the same quality standard as MIG or TIG. Get this wrong in a contract and the argument becomes about vocabulary instead of welds.
| Standard | What it covers | Use it for |
|---|---|---|
| ISO 13919-1:2019 | Electron and laser beam welds in steel, nickel and titanium, from 0.5 mm thick | Quality levels B, C and D |
| ISO 13919-2:2021 | Aluminium, magnesium and pure copper | Same three levels, other materials |
| ISO 5817 | Arc welding: MIG, MAG, TIG | Not for laser seams |
| ISO 15614-11 | Procedure qualification for laser beam welding | WPS and PQR paperwork |
| ISO 9606 and ISO 14732 | Welder and welding operator qualification | Operator certificates |
Two sentences in ISO 13919-1 matter more than its tables. The standard does not address the methods used for detecting imperfections, so it gives limits rather than inspection procedures. And the quality level is expected to be prescribed before production starts, preferably at the enquiry or order stage.
In plain words: if your drawing does not name a level, then B, C or D gets decided by whoever complains first. I watched a container of stainless enclosures get argued over for three weeks on exactly this point.
Cut a Macrograph in Twenty Minutes
A macrograph is a polished cross-section photographed through a microscope. It is the cheapest honest test in welding, and almost no buyer asks for one before placing an order.
- Cut perpendicular to the seam. A slanted section shows an exaggerated width and fools everyone, including you.
- Take two sections: first part of the run and last part. One section from the middle says nothing about drift across a shift.
- Grind flat on 120, then 400, then 800 grit. Keep the face flat, or the edges round over and the depth reading shrinks.
- Polish with 3 micron diamond or alumina suspension.
- Etch to expose the fusion boundary: 2 to 5 percent nital for carbon steel, aqua regia or Marble reagent for stainless, Keller reagent for aluminium. Swab for 5 to 20 seconds, rinse, dry.
- Photograph at 10 to 50x with a scale bar in the same frame. A photo without a scale cannot be measured, only admired.
- Measure on the image, not through the eyepiece. Mark the weld face, the deepest fused point and the original interface line before you read any number.
On a production seam, one section per 10 m plus one after every parameter, gas or fixture change is enough for most fabrication work. On a pressure or safety part, follow the inspection plan in your application standard instead of this rule of thumb.
If the problem is the joint rather than the machine, a wobble head widens the usable gap. The wobble versus standard head comparison covers that trade-off in numbers.
Five Numbers to Record
Whatever standard you work to, write these down for every section. A section that is measured but not recorded is a photo, not an inspection.
| Feature | How it is measured | Value on our sample reports |
|---|---|---|
| Penetration depth | Weld face to deepest fused point, as a share of the thinner sheet | 80 percent or more for partial, 100 percent for leak-tight |
| Interface fusion | Joined length along the lap interface | At least 80 percent of the overlap |
| Porosity | Pore area as a share of fused area on the section | 2 percent or less, no pore above 20 percent of thickness |
| Undercut | Depth below the surface at the toe | 0.1 mm or 5 percent of thickness, whichever is smaller |
| Cracks and lack of fusion | Visual plus section | Zero |
| Root condition | Through-thickness seam and burn-through check | Continuous where the drawing asks for it |
Those are our shop values for sample reports, not ISO limits. The real B, C and D limits sit in Table 1 of ISO 13919-1 and are given per imperfection type, so take the figures from the edition named in your contract. The point is not which number you choose. The point is that a number exists before the first part is welded, and both sides signed it.
Counting Porosity Without a Lab
In a macrograph, pores are black shapes inside the bright fused zone. You do not need a metallurgy department to measure them.
- Open the photo in free image software such as ImageJ or Fiji, threshold the black area, read the area fraction. That is your percentage in under a minute.
- Or lay a 5 x 5 mm grid over the section and count the squares containing pores. Rough, but repeatable, which is what a claim needs.
- Record the largest single pore as a share of sheet thickness. A 2 mm pore in a 2 mm sheet is a hole, not porosity.
Zinc-coated steel breaks every rule of thumb, because the zinc boils before the steel melts and the pores line up along the seam even when the settings are right. That case has its own guide: galvanised steel laser welding.
Destructive Tests, and What Each One Proves
Non-destructive checks tell you whether to ship. Destructive tests tell you whether the process is sound. You need both, and they answer different questions.
| Test | What it shows | What it cannot see |
|---|---|---|
| Root bend | Lack of fusion at the root | Porosity in the mid-section |
| Face bend | Surface cracks, undercut, brittle zone | Anything at the root |
| Tensile or shear test | Load per millimetre of seam | Where the weak point sits |
| Microhardness | Hardening in the heat-affected zone | Fusion quality |
A procedure qualification coupon for laser welding typically takes one transverse tensile and two bend specimens, together with the macro sections. That is the shape of the paperwork under ISO 15614-11, and your inspector will ask for it if the part is structural.
Bend tests earn their place for one reason: a seam can show full penetration in one section and be completely unfused 30 mm further along. The section is a window, not a survey.
NDT, and What Each Method Misses
Every non-destructive method has a blind spot. Knowing the blind spot is what makes the report useful.
| Method | Finds | Misses |
|---|---|---|
| Visual, 5 to 10x | Undercut, spatter, craters, colour change | Everything below the surface |
| Dye penetrant | Surface-breaking cracks | Subsurface porosity, lack of fusion |
| Ultrasonic | Lack of fusion, large volumetric defects | Small pores, and difficult below 1.5 mm |
| Radiography | Pores, tungsten inclusions | Tight unfused interfaces |
| Helium leak test | Through-leaks in sealed parts | Anything not through-thickness |
For 1 to 3 mm sheet, ultrasonic is difficult and radiography is slow and expensive. That is exactly why the macrograph still settles most commercial disputes, and why a supplier who cannot send one is a risk on the order, not just on the drawing.
When a Sample Fails, Change in This Order
Most parameter chasing happens in the wrong sequence. Work down this list before touching power.
- Fit-up and gap. Above 0.5 mm gap on a 1 mm sheet, no setting saves the joint. Check the fixture and the cut edge first.
- Focus. On a handheld head, focus set at 0 to minus 2 mm below the surface is where penetration comes from.
- Travel speed. Slower adds depth and heat together. Move in 10 to 20 percent steps and cut a section after each step.
- Power. On thin sheet this mostly adds heat and distortion, not depth. The power selection chart shows where more watts stop helping.
- Shielding gas. Flow and gas type change the pore count far more than the depth. The shielding gas guide has the flow rates.
- Filler wire. Wire fills a gap. It does not repair a bad fit.
If depth is short and the section shows an unfused land at the interface, it is usually item one or two. If depth is fine and the pores line up in a row, it is gas, contamination or coating.
What to Ask For Before You Pay
- A macrograph with a scale bar, on the thickness you actually weld, in your material.
- The penetration definition used: from the face, or at the interface.
- The quality level, B, C or D, written on the drawing.
- One bend test result on a lap joint, not only photos of shiny beads.
- Two sections from different points of the same seam.
A supplier who can send those five items has a process. A supplier who sends only bead photos has a camera. If you are still choosing a machine, start from the handheld laser welding machine range to get the power band right for your thickness, then ask for sections inside that band.
For the cause side of the same subject, the defects and fixes guide covers porosity, cracks and burn-through from where the problem starts.
Frequently Asked Questions
Is a full penetration laser weld always stronger?
No. On thin sheet, full penetration often means burn-through risk and a wider heat-affected zone. A controlled partial weld with a defined depth and no lack of fusion can carry more load than a burned-through seam. Strength comes from the fused area at the interface, not from depth alone.
What porosity is acceptable in a laser weld?
It depends on the quality level you agreed to. ISO 13919-1 puts the dimensional limits per imperfection type for levels B, C and D in its Table 1. For our own sample reports we accept up to 2 percent of the fused area with no single pore above 20 percent of sheet thickness, and zero cracks.
Can I check a laser weld without cutting it?
You can, but each non-destructive method misses something. Visual inspection sees the surface only, dye penetrant sees surface-breaking cracks only, ultrasonic struggles below about 1.5 mm, and radiography is orientation dependent. On 1 to 3 mm sheet a polished section is still the cheapest reliable answer.
How many samples should I cut?
One section per setup change, one from the first and one from the last part of a long run. On a production seam, one section per 10 m of weld is a practical rate. On a pressure or safety part, follow the inspection plan in your application standard instead.
Why does my weld look perfect and fail in the bend test?
Usually the joint was never fused at the interface: a gap wider than 0.5 mm, focus set too high above the surface, or travel speed too fast for the thickness. The cap looks fine because the top surface melted. The interface did not.
Do I need ISO 13919-1 for a simple furniture weld?
Not by law, but the three quality levels give you a shared vocabulary. Pick a level, write it on the drawing, and both sides know what pass means before the first part is welded. Without it, quality is decided by whoever complains first.
How much power do I need for 3 mm stainless?
From the datasheets in this article, 1000W to 1500W is the range, with claimed depth between 2.5 and 5 mm depending on how penetration is defined. Our own test cuts put a tight fit-up 1500W unit at about 2.5 to 3 mm on 304 with argon. Ask for the section, not the number. See the 1500W and 2000W handheld models for the thickness each one is quoted at.
Related Articles
- Laser Welding Defects and How to Fix Them: Porosity, Cracks and Burn-Through
- Laser Welding Parameters Guide 2026: Stainless, Carbon Steel and Aluminium
- Wobble Welding Head vs Standard Laser Welding: Gap, Speed and Cost
- Laser Welding Machine Power: How Much Do You Actually Need?
- Laser Welding Galvanized Steel: Why the Zinc Fights You
- Laser Welding Machine Spare Parts: Price List and Replacement Intervals
Sources: ISO 13919-1:2019, Electron and laser-beam welded joints — Requirements and recommendations on quality levels for imperfections, Part 1: steel, nickel, titanium and their alloys (three levels B/C/D, scope from 0.5 mm thickness, imperfection designations per ISO 6520-1); ISO 13919-2:2021 for aluminium, magnesium and pure copper; ISO 5817 and ISO 15614-11 named for contrast and procedure qualification; claimed penetration figures from Harsle HW-1500 product table, 2026 (2.5 mm stainless at 1500W), Dato DT-HJ parameter table, 2026 (up to 4 mm at 1500W, gap requirement 0.5 mm) and Raymax handheld parameter table, 2025-2026 (5 mm stainless at 1500W, 0-120 mm/s, working temperature 15-35 C); cross-section reading sequence from JMLASER technical note, 13 October 2026; insufficient-penetration definition for thin-plate tee joints, 20 percent or less of plate thickness, from KEYENCE weld penetration measurement application note. Acceptance values in this article are FANY LASER shop values for sample reports, not ISO limits — confirm against the edition named in your contract.
