Every shop has a version of the same afternoon. A new material shows up, someone starts changing settings and marking test blocks, and by the end of the day there is a scrap plate covered in marks nobody can tell apart and a notebook page that reads like a weather log. The problem is the method. Testing one setting at a time on a fiber laser hides the way those settings interact, so it takes far more marks to learn far less.
Quick Answer
A structured grid does the same work in a couple of hours: pick the two settings that carry most of the effect, hold everything else fixed, mark every combination on one coupon, and read the whole map side by side under one light.
Changing one laser setting at a time hides the interactions between them. Speed, power, and pulse frequency do not act independently on a fiber laser, so a result you get by dropping speed alone tells you almost nothing about what happens when you drop power at the same time. A grid that varies two settings together shows the whole surface at once.
The direction of each setting is easy enough to state. Lower speed, higher power, and lower pulse frequency give you a stronger, deeper mark. Higher speed, lower power, and higher pulse frequency give you a weaker, shallower one. Knowing the direction does not tell you where the usable window sits for the alloy and finish sitting on your bench, and the window is what you are hunting for. Laser Marking 101 walks through how the three interact in more detail.
Sequential testing also breaks the comparison. Mark one block, look at it, change a setting, mark another, look at that one. You are comparing a mark you are holding against a mark you remember, often under different lighting and half an hour apart. Human eyes are terrible at that. Put twenty blocks on one plate and the differences that matter jump out immediately.
At Jimani we develop parameters this way in the job shop because customers rarely send one part type and never send unlimited time. The grid is how we get to a defensible answer before the sample parts run out.
Test speed and power on the first grid, because on most fiber laser jobs those two carry the majority of the effect. Hold pulse frequency, fill spacing, pass count, and focus height fixed for that run. Fix the lens for the entire study. Once speed and power have narrowed the range, run a second, smaller grid on whichever remaining setting the application depends on.
Pulse frequency deserves its own run rather than a spot in the first grid, and physics limits how far you can push it. Every fiber laser carries a pulse energy ceiling around 1 millijoule to protect its internal optics, and on a fixed pulse width laser that ceiling arrives at roughly the laser's wattage expressed in kilohertz. A 20 watt laser reaches it near 20 KHz and is restricted from operating below that. Below the ceiling there is nothing left to find, so the useful test range runs upward from there. Pulse width is a fourth setting on a MOPA laser, since shorter pulses raise peak power while shortening the time energy is applied to the part. Reaching it takes both the hardware and the software. A standard Jimani Hybrid ships with a JPT MOPA laser and Leopardmark, and Leopardmark does not support variable pulse width, so that setting only becomes available to you with a Prolase upgrade.
The technique you have chosen decides which settings matter most. For ablation and engraving, speed and power lead. For stain marking on stainless steel and titanium, focused spot size dominates, because the whole effect depends on heating the surface enough to form an oxide layer without vaporizing anything. Spot size is not a number you type in. Changing it means changing the focusing lens or adjusting a beam expander, which makes it a hardware decision you settle before the grid rather than a variable inside it. Set the optics for the technique first, then test the settings the software controls. If you are not certain which technique the job calls for, settle that using our breakdown of ablation, engraving, and stain marking.
Use a coupon of the same alloy and the same finish as the production part. Mark a grid of small cells, with speed varying across the columns and power down the rows, and mark each cell's settings next to it as part of the job file. Five speeds by four power levels gives twenty combinations on one plate in a few minutes of run time.
Bracket wide on the first plate. Jimani's published working range for the damage pass on Type II anodized aluminum is 12 to 15 watts at 25 to 40 inches per second, and while those numbers were developed for that specific material and finish, they make a reasonable starting point for most ablation work. A first grid running roughly 8 to 20 watts against 20 to 45 inches per second puts that known answer near the middle of the plate and shows you what failure looks like on both sides. Seeing the failures matters as much as finding the winner, because that is how you learn how much margin you have.
Make the cells representative. A half inch square of solid fill behaves differently than the .1 inch tall characters the customer needs, since fill density and line count change how much heat lands on a given area. Mark a sample of the real content in each cell, or at minimum the smallest feature on the part. Kerf width is material dependent and is not the same number as the calculated spot size, so a fill spacing that gives clean coverage on stainless can leave stripes on brass. Leave enough space between cells that slag or discoloration from one does not contaminate the next, and keep the coupon at a single Z height so focus is not quietly a third variable.
Prolase 10 builds the whole array from a single object. Open the object's properties and the Object Step and Repeat panel takes an X Array and Y Array count, a spacing value for each direction, and two laser parameters to increment. Set Parm 1 to Speed and Parm 2 to Freq, give each one an increment, and one text or shape object becomes a complete test matrix. Two parameters is the limit, and every other setting comes from the material file.
Here is a real configuration. X Array 5 and Y Array 5, X Spacing .1 inch and Y Spacing -.1 inch, Parm 1 set to Speed with an increment of .5 inches per second, Parm 2 set to Freq with an increment of 2 KHz. One dialog, twenty five marks, no objects placed by hand and no settings typed in twenty five times. Parm 1 increments along the X axis, stepping across each horizontal line of marks, and Parm 2 increments along the Y axis as the array works down the plate. The Pass selector next to each parameter ties the increment to a specific laser pass, and Local Coordinate System keeps the array positioned relative to the base object rather than to the marking field.
Two is a hard limit, and understanding where the other values come from matters more than the limit itself. The material file supplies the starting speed, frequency, and power for the pass you selected. The array then increments the two parameters you chose and leaves everything else exactly where the material file put it. Choose speed and frequency and every mark in the matrix burns at the same power, the one sitting in that material file pass.
Power can be one of your two variables if that is the study you need. It cannot be a third. To run the same speed and frequency sweep at a different power, change the power in the material file and mark the array again, which gives you a second plate. Plan for that before you cut the coupon rather than after, because it changes how much material you need and how you label what comes off the bench.
There is a labeling feature worth knowing about as well. Rather than writing down which cell got which settings, Prolase can mark the changing parameter value into the object itself, so the plate documents its own settings as it marks. That matters more than it sounds like it should, because an unlabeled test plate is worthless the moment it leaves the bench. Using it requires a serial number type variable text object rather than plain text, and it needs room. The marked values take up more space than a short label, so open up the array spacing enough that one mark does not run into the next.
One parameter will never appear in that array, and the reason is worth understanding. Focused spot size is a function of the lens, so changing it means changing hardware rather than a software setting. Spot size gets decided when you choose the lens and set the beam expander. If a study needs to cover more than one spot size, that is a second plate and a second setup rather than another column.
Building a twenty five cell matrix by hand is the part of this method that discourages people from using it. Each object needs its own position and its own laser settings, and a transcription error in one cell quietly corrupts the result. A note for anyone running Leopardmark, which is built on the Prolase 7 code base: this array capability came with Prolase 10 and does not appear there, so the matrix gets built object by object. Combined with the variable pulse width limitation above, that gives a shop developing parameters for new materials two reasons to look at the upgrade, and a shop marking the same three parts every week almost no reason at all.
Score the grid against the acceptance criterion the customer will use rather than against appearance. Contrast gets judged under their inspection lighting, depth gets measured with a gauge, and a Data Matrix code gets read on their scanner. The right pick is the middle of the region that passes, not the single best looking cell.
What you want out of the plate is a process window, not a point. If exactly one cell passes and its neighbors fail, the process is fragile and the next lot of parts will find that out for you. Anodize thickness varies between finishers, alloy lots vary, and a setting with no room around it will drift out of tolerance without anyone changing anything. Take the center of the passing region and you have absorbed that variation in advance.
Cycle time enters after the window is established. Among the settings that pass, faster is better, but a first article is the wrong place to buy seconds by giving up margin. On deep engraving, remember that multiple passes at higher speed with different fill angles will give you cleaner troughs and edges than one slow pass, which tends to leave a slaggy bottom and slag piled at the edges. If the whole grid fails at maximum power, the answer is a bigger laser rather than a longer afternoon, and we lay out those wattage thresholds separately.
Run the chosen settings on actual production parts as a confirmation batch, five to ten pieces from the same lot, before committing to a full run. The grid proved the settings on a flat coupon. The confirmation batch proves them on the part, with its real geometry, its fixturing, and its surface variation.
Geometry is where coupon results most often fall apart. A curved surface moves the marked area away from the focal point, and as a rule of thumb you can mark roughly 60 degrees around the circumference of a 1 inch diameter part before you have to rotate it. A stepped part can put two features at heights the lens cannot hold in focus at once. Fixture repeatability matters too, since a fixture that lets the part sit a few thousandths high on every fifth load will produce a mark that fails on every fifth load.
Write down what you proved while the parts are still on the bench: material, finish, technique, the settings, fill spacing, pass count, lens, and the acceptance criterion you scored against. That record is the beginning of a parameter library, which is where this series goes next.
If you have a material you would rather not spend the afternoon on, send us a sample part. We run this method in our own job shop most weeks, and we will tell you the technique, the window we found, and what it took to get there.