I Wasted $1,200 Learning This: The Real Checklist for CO2 Laser Glass Etching (A 20W Setup)

I run the production floor at a small shop that does custom laser engraving. We've got a thunder-laser CO2 system, and in my first year (back in 2019), I made a classic mistake. A client needed 120 wine glasses engraved with a wedding date. I ran a quick test on a scrap piece of glass, it looked fine, so I loaded the batch and hit 'run'. The result? 120 glasses with a cloudy, brittle frosted look. $1,200 in ruined product, plus a two-week delay for replacement. The problem? I didn't have a proper checklist for the material.

This article is that checklist. It's for anyone wondering 'Can a 20W laser etch glass?' The answer is yes, but only if you follow these specific steps. If you skip one, you'll end up like I did—with a trash bin full of expensive mistakes. Here are the 5 checks I now use for every single glass job on our thunder bolt laser machine.

Step 1: The Material Prep (The Step Most People Skip)

This is where I lost $1,200. I didn't prep the glass surface. It's not just about cleaning it. You need to understand why you're doing it.

What to do:

• Wash and dry: Use a mild soap and water. No alcohol-based cleaners. Alcohol can leave a residue that causes the laser to micro-fracture the surface.
• Apply a wet layer (the secret): Take a paper towel, wet it with water, wring it out so it's damp, and lay it flat on the glass. You don't want it dripping, but it needs to be visibly wet. This helps dissipate heat and reduces thermal shock. Check your results: The surface should be uniformly damp. Any dry spot will show up as an inconsistent etch.

Why this matters for a 20W system: A lower-power laser like a 20W needs to 'soak' on the material longer. That heat builds up. If you don't manage the heat with moisture, you get the cloudy 'sandblast' look you see on cheap souvenirs.

Step 2: Focus Calibration (The 'Paper Test' vs. Reality)

Everyone learns the 'paper test' for focus. You slide a piece of paper under the head, bring the nozzle down, and see where the paper burns to find the focal point. That's fine for flat wood. For curved glass, it's a trap.

What to do:

1. Place your actual glass object on the honeycomb bed.
2. Don't use the paper test. Lower the nozzle manually until it's just barely touching the center of the curved glass surface. Raise it by exactly 0.5mm to 1.0mm. You want the focal point on the surface, not below it.
3. Check your results: Run a 5mm x 5mm square test. If the edges of the square are sharp, your focus is good. If they're blurry or the spot size looks too big, re-focus.

The assumption here is that the exact focal point is best. But for glass and many transparent materials, the 'sweet spot' is slightly above the material. This is one of those causation reversals people get wrong—people think a tighter beam is always better. For glass, a slightly diffused beam creates a smoother, less brittle etch.

Step 3: Power & Speed Settings (The 20W Sweet Spot)

Forget the default settings in your software (LightBurn or RDWorks). They are designed for wood and acrylic. For glass with a thunder-laser CO2 20W machine, you need to radically lower your power density.

What to do:

Here is the starting point I use for a standard wine glass (curved, clear):

• Power: 35% to 45%
• Speed: 100% to 150% (mm/s)
• DPI: 300 (Lower DPI reduces heat soak)
• Passes: 1 (Do not use multiple passes. The second pass will crack the glass.)

Why the high speed? You want the laser to 'kiss' the surface, not burn into it. If you go too slow, you don't just ruin the etch—you increase the chance of the glass cracking from heat. The first test etch I showed you? I used 20% power and 50mm/s. I was so afraid of breaking the glass I went too slow and too cold, which created that cloudy mess.

Step 4: Post-Etch Cleaning (The 'No Harsh Chemicals' Rule)

You will have a white, chalky residue after the etch. Don't scrub it off with a scouring pad. Don't use glass cleaner with ammonia.

What to do:

1. Let the glass cool for 5 minutes. It's physically hot, even if it doesn't look it.
2. Rinse the glass with warm water from the tap. Do not use cold water—thermal shock can crack the glass.
3. Gently wipe the residue off with a soft, damp cloth. The etch should reveal itself as a smooth, 'sand-etched' look—very fine, not frosted or chunky.
4. Check your results: Hold the glass up to a light. You should see a clean, white mark. If you see tiny chips or a rough texture, your power was too high or your focus was off.

Step 5: The 'Reject' Check (A 30-Second Inspection)

This is the last step and the one I created after my $1,200 mistake. You don't need a magnifying glass. You need a simple, quick method to separate good pieces from bad before they go to the client.

What to do:

• Look for hairline cracks: Rotate the glass against a dark background. Any tiny crack around the etched area is a reject. It will get worse with washing.
• Check for 'sticky' residue: Run your fingernail over the etch. If it catches or feels rough, the power was too high and the glass is micro-fractured. Reject.
• Look for overall consistency: Is the etch uniform? If part of the letter is darker than another, it means your focus wasn't consistent for a curved surface. Reject.

One more thing: I don't have hard data on how many etches this checklist saves. What I can say anecdotally is that before I wrote this down, we had a reject rate of about 12% on glass. After training my team on this five-step list? We're under 3%. The difference between a $200 savings on the cheaper machine? That $200 savings turned into a $1,500 problem when I had to replace the 120 glasses. The cost of the machine is the start. The cost of redoing work is the real killer.

Prices as of March 2025: We sourced replacement glasses at $10/each from a local supplier. Verify current rates before ordering.

So, if you're running a CNC cutting machine manufacturer like thunder-laser and trying to figure out glass etching, treat this as your base. Print it out. Hang it next to your machine. You'll still make mistakes—everyone does—but you won't make the expensive ones.