Making a CNC Glass Cutter | Avid CNC

[@CCCfeinman55]

Well, you pegged the problem earliest and best, when you described the propagation of the crack through the vertical plane of the glass being imperfect and uneven. Folks I know that need dimensionally correct pieces will use a 2 x 72“ grinder with a water drip and silicon carbide or ceramic belts with the piece mounted in a precisely positioned puck clamp and rotated against the belt to take off a few microns here and there until they get the exact diameter they want. The only other thing I know of to get better results, is to use a machine type press to split the glass just outside the wheel line, with consistent pressure all around. Or, getting closer but still shy of machine tolerances, use a professional grade glass cutting ($130-$2,000) machine. Or pay for someone to try a waterjet. Sadly, short of some of the above, you’re stuck with the kind of deviation that you’ve been seeing. Great video! Some days are learning days, aren’t they?

[@Clough42]

Yeah, this is what I'm concluding as well.

[@Idontwantagoogleyoutube]

@@Clough42 i should think the easiest method for simple circles is to cut a known dimension out of a material you can control and then score off that with a fixed offset. if you want 10.091, why can't you make a 9.591 circle blank, and score referenced off that with a centerpoint mounted in a 0.5 holder? basically an inlay technique but instead an outlay

[@Clough42]

@@Idontwantagoogleyoutube the problem isn't the score. The problem is that the crack does not propagate straight down through the glass.

[@ZXLMaster]

​@@Clough42 Has anyone suggested using the drag diamond engraver to score your glass? If not, let me be the first. But as you said, the score is not the issue. Food for thought.

[@Clough42]

@@ZXLMaster it's been suggested many times and I have attempted it. It hasn't worked for me. The edge of the scratch is kind of flaky and fractured, and the crack doesn't follow it.

[@rodjsy6363]

Having spent most of my life working in the glass trade the most accurately cut circle, especially in thicker glass is to cut a rough circle 4 inches bigger then cut your accurate circle right in the middle of that circle. This equals the stresses and works perfectly with a tidy edge every time.

[@fozbstudios]

Hope James sees this. Seems doable

[@antonw8134]

From the amateur astronomers I’ve seen on KZbin, most cut (or mold) their circular glass blanks oversize, then wet diamond grind the blanks to the correct diameter. I think you’ve achieved the best results that may be achieved with the process you’re using.

[@UncleKennysPlace]

Yep. For mirrors, I just use a circular mask rather than make the blank perfectly round.

[@ericmorriscompany9648]

My experience from a project where I built 1400 doors with glass and mirrors is that getting precise dimensions on glass is done by grinding after cutting. Since glass is an amorphous solid and not crystalline it’s not going to break exactly where you want it. The grinding serves to give a safe edge as well.

[@MachiningandMicrowaves]

For tiny boring bars, I use Simtek Simturn AX carbide bars that can cut OK down to about 2mm diameter. They do much smaller ones that go down to 0.3mm and they are reliable at 7-9 x depth. I use them to machine tapers from 4.5 down to 2.0 mm inside antenna feedhorns. I made a holder using a cross-pin and setscrew to set the orientation against the angled rear face because the factory holders are $$$$$$$. I "invested" in a full set of three-point bore mics and ring standards from 6mm to 50 mm, mostly from ebay purchases, and I've hardly used my old Starrett telescopic snap gauges since then.

[@matthewlee8917]

As others have said I think grinding is the only way you’re going to get more dialed but I’m actually impressed with how accurate you method is.

[@PennerFab]

As someone who was a Glazier for most of my adult life, glass is one thing that you will almost never get to run without some sort of flaring 🤷‍♂️ water jet would be the most consistent thing you could do.

[@BrianBoniMakes]

Used to get glass inspection plates for measuring color on moving materials, basically an optically pure sight glass. The glass shop Proscience would cut the flat sheets round then wet grind them to the exact size. Looked like a lathe with a huge steady rest and suction cups for a chuck. Did wonderful work, they were always perfect.

[@rickshoop2063]

The next time you run the part, mark the orientation of the part and see if the high/low dimension corresponds to where the axis reverses as it goes around the circle.

[@Chris-bg8mk]

That. Also to check if the larger dimension is consistently in X or Y, because you could then purposefully try to make an oval to compensate.

[@justus1995]

maybe also rotate the glass 90° between two cuts to see if it´s a material or cnc based problem

[@milosrankovic8952]

If you cut two circles, mark their X and Y orientation and stack them so that X of first is on Y of second, you will see how much you are out of the circle or how much your gantry is off.

[@DavePalmer]

Great vid, my guess about the deviation is that the tip of the cutter is not concentric with the bearing. As it rotates it deviates.

[@JohanDegraeveAanscharius]

I think tou are absolutely right, to much points of play.

[@0RDClark]

Might be worth looking at the glass through polarised light, just in case there's any residual stress that's causing the circle to deform when broken away from the parent material. I would also, as others suggested, mark datum points on both the stock and circle to see if the variations are plane specific, and related to any internal stresses.

[@johnmatthewson6733]

As others have said driving the crack around uniformly is the problem. As someone else suggested, cut from an oversized circle to make the boundary conditions uniform. Driving the crack by pressing downs with your fingers in a few places is not a great way. You might be able to correlate the out of roundness to where you pressed. However to drive the crack uniformly is I think the key. Once scribed you might be able the drive the crack with a flame. That might well move thermal expansion the wrong way. Try touching the crack with a cotton bud dipped in liquid nitrogen then once initiated, push the crack with the cotton bud keeping it wet with lN2. I also saw strange shapes cut into glass without scoring (except for a small initial crack.) The equipment had a head with three tubes side by side. If I remember correctly cold air was blown out of the middle tube and very hot air through the outer tubes. Apply that behind the crack tip and the crack will grow away from the cold tip and perpendicular the the line of the two hot tips. If the tips are moved and rotated in the right way, the crack can be steered around through quite tight radii. I’m pushing my faulty memory a bit far but I think it was used to cut precision display panels. And I just remembered that in the old days glass cutters would rub spit into the score - this lowers the stress required to drive the crack due to stress corrosion. This means the crack grows more slowly and is less likely to wander off the score line.

[@ChazzC]

This is reaching back into the deep recesses of my memory, but I think that the glass blower at UVA's Chemistry Department in the early 70's used a glass lathe & grinding wheels to make precision round items. I worked with him on a couple of research projects as part of a Summer job & my thesis (Chem Engr) and would always spend time just hanging around after he finished working my job or when I was picking up the fully annealed end product because what he could do was fascinating. Somewhere I have a complex mini reactor that he scrapped because it was out of spec (but no one knew it's not correct except him & me, and I guess now a lot of others).

[@vanlife4256]

I’m very pleased with your workflow and the pursuit for accuracy and precision!

[@wdwerker]

Probably 20 years ago a relative got a job running a CNC water jet cutter for a stained glass shop. Complex shapes were breaking unpredictably in expensive glass. They didn’t need super tight tolerances but the labor and material savings let the company finish ahead of the deadline and profit even after buying a very expensive machine back then.

[@samheasmanwhite]

I have cut glass tube and rounded discs by putting them on a lathe and using an angle grinder on it, came out surprisingly accurate from such a crude method. If you have a grinding attachment for a lathe that would help, and I am sure you could use diamond or even just carbide tooling if you set the feed low enough (use a negative rake tool if you can). To be honest I think the easiest option would be to just grind off the high spots, fine carbide grit on a belt sander is very effective, but for adjustments this small you could just do it by hand.

[@markkrick8602]

Not sure how much this will help but I'd guess the variation is due to slop in the carbide cutter axle and it's clearance slot

[@rkalle66]

The glass cutting edge will travel in a sinus wave like way around the ideal path. Like train wheels on a track or wheels of a shopping cart. On a curve the mechanism is pulling the cutting edge not on a tangent to the circle/curve but a secant line. This will introduce side forces to the cutting edge touching the glass.

[@unpopular_mechanics]

The tiny one piece carbide boring bars you get on mcmaster are great for small holes. Some are made by micro100, some are made by another brand that I can't remember.

[@billstrahan4791]

Just need to make a strong suction cup device to hold the glass and go in your lathe with the 4 jaw chuck so you can adjust for runout. Then a strip of sandpaper on a mildly flexible backing that you can hold on the toolpost and slowly feed to hit final diameter. I say all of this mainly because I want to read the safety nazi comments if you do it. ;). As always, I enjoy your videos immensely!

[@lwilton]

Lens makers for 100s of years have stuck blocks of wood and dowels to their glass blanks to act as handles when grinding, usually by using melted rosin or hard wax. With a jig you could presumably get an axle positioned better than you could a suction cup (which is likely to slip sideways as you squeeze the air out), and has the advantage of not being particularly flexible. Once the axle is installed, proceed as you described. :-)

[@chuckinwyoming8526]

Could you clamp the glass on vacuum riser block, find center and grind 0.020" off the edge with a diamond wheel circular G2 or G3 path in the CNC? Should result in a square edge and +- 0.001" circle. The vacuum block could be a box made with couple sheets of plywood, 2x4" edge spacers, a grid of holes in the top with a non slip pad and a large hole in the side for a standard vacuum cleaner. I use this design for milling PC boards.

[@azinfidel6461]

The one and only time I ever had a round glass table top made, they had a plywood table with a lazy Susan on top of it on which they placed the glass disc. Attached to the table was essentially a Disc Sander on its side. They just slowly turned the glass disc against the sander to finish the edge. Confident you can come up with the high-tech Precision version of this...

[@markfletcher6511]

I loved your video, for detail, thought process, and entertainment value. I was wondering if you were able to measure the accuracy of the scribe line prior to running the crack. That would take into account any error in the cnc machine and tracking of the cutting wheel around its pivot. Keep up the great videos.

[@cullenpurkis4593]

To bring the glass into round, try using a sanding jig like wood workers do with a pivot a fixed distance from a sanding belt. Sandwiching the glass between two pieces of wood may stabilize the setup as well. Good luck!

[@bobluthier3031]

Cool video. I think Stefan has a video where he mills glass with a regular carbide endmill. Pretty sure its one of his shop talk series

[@grendel1960a]

the issue with concentricity on the original tool may have been made worse by that shim, that would have pushed out one side of the bearing, meaning the bearing centre was off the line of the tool, and as you rotated around the centre the shim would have thrown the tip off by its thickness as you rotated

[@elanman608]

Just a thought I assume that the bevel on the edge varies according to where you press. I have seen a trick used by the guys who custom cut windscreens they pour alcohol along the cut line and light it the differential heating cracks the glass instantly. They do identical cuts on either face of a laminated screen close enough that they cut the laminating film with a razor blade.

[@geofferyshanen7758]

Howdy great video. I know you probably already thought about this, but I use my Avid Cnc to cut vinyl and gasket material using a Dog River drag knife that functions almost exactly like your CNC glass cutter. One thing I did not hear you mention in your video is the drag knife blade offset. When I use the drag knife I use Vetric Aspire for the cad and cam work. Within Aspire there is a specific tool setup for a drag knife operation in the Gadgets drop down. One of the important settings required is to set the blade offset, this offset places the cutting edge precisely on the cut line. Once you set the dimension in the CAD, you do not try to chase the perfect diameter, you adjust if needed by changing the blade offset. I don’t know if this helps out, but I thought you might be interested. Great Machine work by the way!!!! Thanks

[@stephensackett8920]

Try using the diamond engraver to score the glass instead of the caster, it might improve the precision by eliminating the bearing run-out and wobbling of the caster.

[@anmafab]

I was thinking this too. Is it more likely to crack the glass than the scoring wheel?

[@nathanliebespeck8088]

The diamond engraver might actually work quite well. When I need to score glass I just use a carbide scribe. Similar to what you would use for marking steel. I would say there is a learning curve for getting the pressure right, but I would also say it's not that steep of a curve.

[@stephensackett8920]

@@anmafab I'm no expert at glass cutting but it couldn't hurt to try, after all what's more fun than wasting material in the pursuit of perfection?

[@stephensackett8920]

@@nathanliebespeck8088 James' engraver is spring loaded so tuning it may be as simple as adjusting the spring rate or compression. If that doesn't work next I would try to dampen the spring to keep the tool from vibrating.

[@davidapp3730]

With a grinding wheel made for stained glass setup in your CNC setup and a mist coolant you should be able to take off the few thousandths you need to refine the size/shape.

[@gawaic]

Hello Even in automated glass cutting processes, the guys use a petrol based lubricant/cooling placed on the path to cut before to run the carbide wheel. It seems to improve the cutting. Another thing : As it's a drag carbide wheel, it means it is slightly off centered and your cnc drives the trajectory of the center of your tool. Not the wheel. Could this lead to some discrepencies in the final dimensions?

[@TravisFabel]

@@johncoops6897 I guess you didnt read his comment, where he clearly said "on the path" and not "a few drops on the glass"... but more importantly he nailed the biggest problem with this... the cutting blade pivots on the center of the tool, and doesnt follow it.

[@gawaic]

@@TravisFabel Thank you john. You got my point! In car glass industry, they automate the cutting of glass parts and machines have a continousely fed flow of petrol lubricant ahead of the wheel.

[@Gamerock82]

@@johncoops6897 Yes but it shouldn't be oil. Turpentine works best. As much to lubricate the wheel as to ensure a good cut.

[@tonyray91]

Mineral spirits also works well

[@g.tucker8682]

I worked briefly as an assistant to a stained glass artist, on a large church job. (He used baby oil as a cutting oil, btw). There were several unusually large segments in each of the works, about 30" in some cases. We knew the glass would move some after cutting, but found it was much greater than expected, and the came (aka 'leading') would not cover the error. We re-cut the smaller parts that adjoined the large elements. The next day when we started to cut the lead, we found the glass had continued to move overnight. We ended up cutting all the large pieces for all the windows, and giving them several rest days to settle down. Anyway, glass can move after cutting, a lot more than I would have believed before I saw it.

[@MichaelMoore-or2fi]

Good morning man your problems with the sizes is the steps in your machine if you re-calibrate your machine on the steps it’ll go correctly for you. I’ve been building CNC machines for about six years and this is just a simple configuration problem and your machine will be cutting perfectly on your measurements, just check and see which axis is different. For example, if the Y axis is out of measurement, you’re going to want to re-calibrate only the Y axis, but you have done an excellent job man.

[@andrewhall2554]

I once talked to a man who made stained glass windows and told me he had good results cutting glass with a water jet CNC. Unfortunately, I don't know what kind of tolerances he was able to achieve, but I understand that this process can cut thin materials to better the .005".

[@JohanDegraeveAanscharius]

0.01 mm with waterjet = 0,0004" In Europe we also work with thou... sands of mm... change to decimal and you will work more accurate....

[@matspatpc]

Whilst (usually) not round, the aquarium that are made on industrial scales are ground on the edges - I'm sure that's to some degree cosmetic, but it also ensures a better size. I have quite a few hand-made aqarium too, and the edges aren't quite the same straightness. I'm pretty sure, like Anton W says, the only way to guarantee precision roundness/dimension is to grind the edges. Because when the glass breaks, it will vary a little out or in from the score.

[@gregfeneis609]

Really cool stuff, James. Perhaps you'll be able to get rounder parts with some elliptical pre-emphasis to the round cutting path? You'd have to get a way to index the part to the rectangular blank to the work holding fixture. Maybe the out-of-roundness is somewhat repeatable to a location in your machine. I hope you can see your way clear to tighter tolerances.

[@geofferyshanen7758]

Great video as always! Just wondering, Garrett Frome has a great video on cutting perfect circles in wood using Vetric Software. His principal point is controlling the tool offset, not the circle diameter. He says you should never change the circle in the software, but true it up using the offset… Just a thought.

[@MachiningandMicrowaves]

Excellent video James! Is there any correlation to the error orientation with the clamping positions? I wonder if viewing under polarised light might show a grain or internal stress pattern in the glass? Perhaps then try cutting it with the "grain" at two orthogonal orientations relative to the bed to see if it's about internal stresses? If you can measure the diameter more precisely at multiple angles, is the shape an ellipse or just randomly not round? Is the draft angle different where the area of waste is larger, or do you have a consistent angle all round? Perhaps cutting the square sheet to an oversize circle first? How about using a hot iron to propagate the crack slowly like when cutting stained glass rather than using a bending moment, where the crack velocity is uncontrolled?

[@rbyt2010]

Just curious... did you try just scribing the glass with the drag diamond? This would be sooo much faster than my diamond ring saw (and of course more accurate).

[@moki123g]

I use a diamond scribe to cut glass. The engraver probably would work great.

[@shauncrosby7652]

Was coming to ask the exact same thing.

[@jimurrata6785]

x3

[@v8Mercury]

I also came to find this comment.

[@lustenaderj]

First thing that popped into my mind watching this too.

[@Oberkaptain]

Have you thought about putting a small flat on the 1/4in bar to assist with the grub screws? Awesome little tool fix.

[@jimmurphy6095]

Filling the bearings with a much more viscous grease might take some of the "wiggle" out of the track, improving it a bit. It seems the bearings were slightly too free to move around.

[@jeffreyhallam5517]

I have an idea that might help you out. Use your welding table as a jig for an over sized lathe. The fixturing holes should be precise and rigid enough to machine a spindle mount to hold oversized stock off of the side of the table. I came up with this while trying to come up with a way to cut large gears that didn’t fit in my lathe. It would only be good for low speed and low pressure operations but I bet it would work a treat. Plus you have a waaaaay better welding table that I do. Machine a vacuum spindle and an X/Y tool holder that index to the holes in your welding table and you could have precisely ground glass in the home shop. Heck I bet you could leave a drip trey under the glass and run coolant too.

[@gregdittrich1206]

Love your videos James. You deserve so many more subscribers.

[@j.dietrich]

Cool project and potentially useful for cutting flexible materials like leather or vinyl film. FYI Stefan Gotteswinter and Ben Krasnow have done videos on machining glass.

[@heirik2012]

Nice video, while I personally don't have experience with cutting glass there is an inherent issue with using a cartesian CNC to cut perfect circles. In my experience, you will almost have some deviation from the circle due to the axial nature of the cut. If you have the option I would oversize the cut slightly and then refind the edge on a lathe at very slow speeds.

[@Fed3d]

I think its totally okay to grind last 0.02-0.03 after. Maybe some rotating fixture to standard stained glass grinder.

[@msmith2961]

CNC a workholding fixture for the glass. Cut one side of the glass then flip and do the other side to reduce the bevel created when you chase the crack around the score lines.

[@jimj.9106]

James, It looks like all the clamping is done from one end. Relocating the clamps to equalize pressure may help.

[@Clough42]

The surface the glass is on is as flat as the glass, so the clamping doesn't introduce much, if any stress. The ten failed attempts I showed at the beginning of the video were done by hand with no clamping, and they have the same issue. It's a property of glass that makes cutting below that level of precision impractical without grinding.

[@BigDogCNC]

We have a waterjet cutter and regularly cut glass to within .003.

[@think2hard422]

Put something like your microscope camera pointing at the wheel as it goes around. You should see something flexing. Either that or the wheel doesn't stay exactly perpendicular to the cut. My guess is the cutting wheel/shaft has some axial play in it. The equivalent of a Belville washer might help.

[@alfrond]

i suspect along your lines here that a cutting wheel oriented off-bearing-center would consistently alter the cutting diameter, whereas axial play in the cutter could result in path instability so as to steer the cutter back and forth across the intended path.

[@leysmetalmecanica1885]

Hello James, I think that the inaccuracy in the diameter could be due to the fact that the fracture in the glass is not generated perpendicular to the plane of the glass in the entire circumference, even I do not think it has a certain pattern. Could this be modified by marking the glass on both sides? it's just an idea to force the fracture in a certain direction. I clarify that I don't know how to work with glass, I even find it admirable how you separated the circle from the rest. As always, thanks for the great content on your channel.

[@fourtwo7612]

Or thinner glass?

[@Gamerock82]

Not an option, I'm afraid. Glass will immediately break along the first cut when the pressure of the second cut engages. That said, if the identical track was followed, on the opposite by a pressure wheel (not a cutter)... the break might be perfectly perpendicular. Glass is only slightly predictable with regards to break direction but one thing is for sure, it breaks away from a score line. With score on both sides there will likely be some splintering. I may be wrong as not tested but I was taught only score one side and never go over a score line, the tiniest angle change in the cutter can result in a splintery snap as opposed to a crisp cut.

[@GBWM_CNC]

Great as always! Really smart tool.

[@DavidLindes]

Might be interesting to get some slow-motion footage of the roller during travel... I wonder if it's wobbling any. OTOH, I could easily imagine it just being variations in the run-out of the cut, too... Hmmm, perhaps you could set up a "lathe" for the sanding stage??? Somehow center the glass between two suction cups or something, with something driving them to turn..... Not sure if that could be reasonable. Anyway, interesting stuff, and hey, 17 thou isn't too bad. :)

[@DavidLindes]

oh, something else to try: the diamond engraver??!? Does that have a nice sharp point to it? It could maybe score the glass, too... maybe glass dust would build up and cause problems as it moved, but I know tubing cutters (or at least one style of them) don't have a rolling part, so... maybe you could get more positional consistency there, if (like others have also suggested) the problem is just that the wheel is wobbling about a bit (for any of several possible reasons)?

[@JasonTHutchinson]

You could try using a spiral leadin toolpath to preload the cutter so the start angle of the cutter is correct relative to the circle. Some of that error could be caused by the swivel not lining up perfectly at the beginning of the cut. You might be better off with a diamond or carbide insert opposed to a cutting wheel. The swivel can introduce quite a bit of error.

[@Clough42]

A curved segment to line the cutter up before moving to the circle is a good idea. But coming in to the circle tangent to it is a bad idea. Where the score lines intersect, the glass breaks off and leaves a peak. I tried the diamond point first, and while it scratched the glass, the glass refused to break along the line. I'm not totally sure why.

[@JasonTHutchinson]

@@Clough42 you could try a different technique in breaking the glass along the line. Instead of pressing down, you can use the end of a glass tool to tap around the edges. One common way is to slide the edge of the glass off the edge of the table and gently tap around the circumference from the bottom and then finish up by doing the same on the top. This will ensure that the glass is going to break at a 90 degree angle rather than being tapered.

[@adam207321]

I also wonder if the non squareness of the edges is an effect of you putting the pressure on it. Maybe you could try thermally abusing the glass instead of putting pressure on it. Could possibly yield a better edge, or a more chaotic one...dont know just an idea.

[@VoidedWarranty]

interesting. from my experimenting with a vinyl knife on a cnc router, I was expecting it to cut very slightly undersize due to the trail/caster angle of the wheel. Though that mostly showed up for me in tight radiuses

[@foxabilo]

Thing to check that's caught me out few times is that darn carbide wheel can have some side to side slop in it's mounting/bearings and last time it caught me out it was about 10thou either way of slop.

[@Bob_Adkins]

I think the problem is with small irregularities around the edges, making the roundness measure more inconsistent than it really is. Hand sand a little longer and it will get closer than you think, faster than you think!

[@JohanDegraeveAanscharius]

Yes also, but one should suck it on a turntable with a fixed sanding device.(or let the router do the work, if it needs to be so perfect)

[@Bob_Adkins]

@@JohanDegraeveAanscharius Yep. I think it's just a 1 off, but if not, he'll figure out a fixture to true it up.

[@ChiralSymmetry]

There might be some backlash in the cutter head. Also, when the CNC moves the cutter quickly, it might slightly torque the direction of the cutter due to acceleration. Maybe keep the acceleration really small and gentle right after the initial straight line?

[@trapiturette]

hey James, hope not repeating (have not red all comments...) but to me the play here is on the scribing wheel that shift a little on its axis, perhaps a little shim washer would narrow the movement? I can remember when I used to cut glass for art frames, the cutting/scribing wheel was quite loose.... and thanks for your great videos!

[@squelchstuff]

The swivel has caster, so it would be interesting to see how the orientation affects the dimensions. Perhaps you cam make some XY witness marks prior to snapping the circle out James? My feeling is the caster does not quite follow the spindle centre, thereby introducing a concentricity error. Interesting experiment all the same. [EDIT TO CLARIFY] Is the cutting wheel caster truly concentric with the spindle?

[@nils1953]

It shouldn't be concentric, otherwise there would be no turning force to orient the wheel to it's cutting direction.

[@squelchstuff]

​@@nils1953 Agreed, it should trail behind the centreline, but my poorly worded comment tries to point out that the scribing wheel may also be offset from the centreline, or even non perpendicular (camber) to the surface of the glass. Lateral play in the scribe wheel's axle may even introduce the errors observed. This is speculation based purely on the noted build quality of the original cutter of course, and I'm sure the inaccuracies could be characterised and mitigated against if they are found to be consistent in each of the XY planes. The snipers in the movies make it look so easy cutting a murder hole. Even on tempered laminate high rise building glazing no less.

[@DavidLindes]

@@squelchstuff I don't think the "murder hole" needs this level of precision... ;)

[@GT40Nut]

James, it wasn't fast but it did work. My brother used a rubber mask and a sand blaster to make very intricate 3D engravings and cuts in glass. I'm sure I'm not the only one waiting to see how your new CNC router cuts something like maybe aluminum? Also what wireless controller are you using with your new machine? Keep it up, inquiring minds want to know.

[@offgridgarageUSA]

Have you ever run a Ballbar test on your X-Y axes to see if it's actually able to move in a true circle. It will also tell you if you have backlash in either axis. I'm figuring that your machine uses stepper motors instead of true servo motors. True servos are much more accurate. Ballbars are quite expensive tools but well worth it for diagnosing this stuff.

[@phoolb7326]

Very interesting. Is there a tendency for error based on the glass direction with respect to the sheet when it was purchased? Could be internal stresses that developed during cooling that affect the bias of the break. You could correct this by just cutting a properly corrected ellipse if the axis is repeatable. I also initially thought that a misaligned bearing would be the issue, but that would just move the center of the glass you cut with respect to where you thought the center would be by a few thousands. Now I'll loose a few nights sleep until I (or You) figure it out.

[@scottallen3660]

I was wondering the same thing. Could be tested by marking the short and long axis inside the circle before cutting. See if there is a correlation.

[@Gamerock82]

I would love to see the results but as far as I am aware there should be no directional differences, in stresses or thickness, etc. If this is regular float glass, that is. I would imagine these factors would change with toughening or annealing. Toughened, for example cannot be cut this way at all.

[@phoolb7326]

​@@Gamerock82 I agree with float glass I wouldn't expect much stress, but I am assuming that this is plain cheap borosilicate glass which I believe is rolled in production with a relatively quick cooling time. I certainly could be wrong though... A good day is when I'm right at least once. lol

[@gaboonviper85]

I think you'll need to grind the OD to a finished dimension by maybe sandwiching the glass between two fixture plates so you can spin the disks in the lathe and grind the od of the glass via tool post mounted grinding attachment.....

[@Mobile_Dom]

could you not use the diamond drag engraved for glass as well? i feel like if carbide is good a diamond tip would be better, no?

[@johnsheppard7388]

Was thinking the same thing, maybe the wheel is wandering slightly.

[@jobersudyobodou9362]

Me too.

[@Oberkaptain]

@@johnsheppard7388 or since it is oval it could be a slight backlash issue on the ball screws

[@chrisquinn3751]

For a few years I worked in the aircraft windscreen industry ( when they were made of toughened glass! ) All the glass was cut with a diamond and an aluminium template. In your position James, I'd definitely try the diamond drag tool - maybe with slightly increased pressure. I'm not sure how central the cutting point of the diamond is to the centre of the holder so it might be necessary to try cutting a few squares to check that parameter. If it is not acceptably "central" it might be possible to compensate in the tool holder or maybe in the CAD model? Good luck - glass can be pretty tricky, maybe because it is not a real solid?

[@ikbendusan]

@@Oberkaptain 20 thou backlash in a ball screw is anything but a "slight" problem. even accounting for pitch error + backlash, you shouldn't be seeing anything more than 0.003", maybe 0.004" (with a single non-preloaded nut; different story for a double preloaded nut)

[@MrLeerivett]

after twenty years as a glazier the only tip i can give you for a cleaner edge is to chase the cut when you up open up the score line. by this a mean apply pressure 1 or 2 inches behind the leading point of the crack as you work your way around the circle. it looks like you are applying pressure on top of it. the worst point will always be where you initially start to open the cut. but this technique should help with a clean straight cut the rest of the way around. hope this helps

[@MrLeerivett]

after reading the other comments. scoring on the opposite side will not work at all it will only create more tolerance errors. the diamond engraver cant be lubricated properly and will not leave a clean score line, there is a reason all glass cutters are wheels. heating the glass will result in a wavy cut when you open it. best is room temp and glass has no more movement than steel

[@Clough42]

Thanks for the benefit of your experience. As you noted, there is a lot of advice here from people who thought about it for a few minutes, but have never actually done it. It seemed clear to me (someone who has also thought about it and not done it) that scoring both sides would not work out because the fracture lines would never align. I'm able to run the crack ahead of the pressure point pretty easily, until it gets near the edge of the sheet. At that point, the forces change significantly, since I'm just pressing against the support from the towel underneath. Some have suggested cutting a my circle out of a larger circle so these forces would be even. That sounds reasonable, but I'm becoming convinced that the only real solution here is grinding.

[@MrLeerivett]

@@Clough42 your welcome for the advice. Cutting your circle from a larger circle would give even pressure when opening the cut but I’m not sure how much it will help. You are right about the grinding. If you are looking at .000 inches then grinding is the only way. The glass cnc cutting industry is well established and you should be able to buy polishing bits that fit the Collet on your machine. Then you would need a water coolant feed and drainage also a vacuum holding method. I would make a tray with vacuum , drainage and high sides that can be fixed to the machine bed when needed. Water feed can be clipped to side of the spindle. This method would require checking of the tool diameter to adjust for wear and tear on the grinding wheel

[@PacoPR28]

Must be nice for intricate stainglass pieces.

[@samcivi9326]

I wonder if you could get a more consistent crack propagation and edge finish if there was a crack nucleation point of both sides of the pane?

[@johnpelitidis6297]

James, the swiveling scoring wheel looks as though it is not inline with the axis of it's pivot centre. This may be the cause for the difference in roundness as it traverses around the path. i.e. it is acting like the castor wheels on a shopping trolley. Just my thoughts :)

[@ArcAiN6]

if you want to address the beveling of the crack propagation, score both sides of the glass, and when fracturing the score line, do so under water for a cleaner fracture, with less "walking' of the fracture along grain-lines of the glass. As always, make sure the CNC tolerances, steps/mm, and backlash are addressed

[@peirossmallhomemachineshop5364]

Interesting. However, you have not checked if the larger and smaller diameters correspond to the same orientation on the cnc table. Keep also in mind that glass is anisothopic and this implies that the ´molecular’ structure is impredictable. However, I guess that you have achieved a good result for many applications. Nice vidéo as always.

[@dannyschell1095]

great video! there are many good suggestions in the comments. I would think if the orientation of the egg shape is consistent then just compensate it in the CAD before you CAM it... good ole Kentucky windage!

[@knobbymcfeck]

As is tradition! You said the phrase!

[@toastrecon]

I wonder if you could run a spiral toolpath into the desired circular path? It might make a mess. Something like 20mm on the outside and then come in by 5mm in radius until coming to a tangent and then the desired circle?

[@Sergiodj453]

I’m more impressed with the adapter, nice work all around!

[@robertsternlieb5532]

Question: could you offset the 13mm bearing bore so the cutter does land exactly under the spindle? it would align itself as soon as the gantry moves.

[@mistr_clean7557]

use dowel pins to set your reference square and make a front and back cut on the glass

[@michaelbush2905]

I've been making glass products and for 35 years. when you cut a circle and need it to be exact , the drop has to be the same around the diameter . depending on the size the drop is a least 1 to 2in for a 90 degree break .. some dressing may be required ..

[@bkoholliston]

Always fun videos! Why not try your diamond drag cutter? Do you think you have lateral runout in cutting wheel on its axle? At least the diamond drag would not have any play.

[@MikeRatcliffe24]

Excellent video, I skipped ahead to the results and though I was being smart by wondering if the non uniform differences arose from having two rigid materials [work-piece/tool] . A spring loaded tool being a solution you had already implemented humbled me quite some. One thing I have noticed from general making of things from cheap components, if its a small/cheap bearing, its a C3 clearance and dont be surprised when you take it to bits and find they carrier doesn't have a full set of balls, whilst im not sure what bearings you salvaged from the hand tool [or if it even had ball bearings], its worth a check to see if the bearing is only half populated or of loose tolerance, with the leverage effect of the head/cutting_disc, this could be a major part of the lack of repeatability?

[@brandonthebuildertx]

Now you need to make a glass lathe. I'm envisioning something like a record player to spin the circle with an oscillating spindle sander on the edge to both clean the cut and true the size.

[@bstanga]

Great job! I wonder if you could use the diamond drag to score the glass?

[@Clough42]

Tried it. Failed. The crack doesn't follow the score line.

[@matt-scher]

Any glass supply company stocks carborundum stones that can be used to remove the sharp edges after snapping the glass on the cut. This is far safer than using emory cloth. For lubrication try kerosene. That is a traditional glass cutting lubricant.

[@stephenbow3292]

Hi James, A thought, Have you tried to run the crack using alcohol and a flame? so as to not exert 10 digit pressure around the fracture and near the edge of the sheet. I have watched a windscreen mate do this on LAM glass. Sorry if someone has already mentioned this. Long time viewer first time comment.

[@jtwhite2084]

Have you measured how much the cutter wheel can shift from side to side on its axis of rotation?

[@SR-ml4dn]

I Think the problem is here, the cutting tool is wobbling around. You could also Cut Them On oversized measure and mount the CNC with air driven grinding tool and let the CNC router make the circle again.

[@weaves7]

Is your out of round consistent with the orientation of the glass on the machine in a particular axis/axes? If so you could either adjust your machine to be true if that is the issue or if it is the cutter maybe you could program the CAD/CAM to cut an ellipse that accounts for the error.

[@justinlabarge8178]

His machine isn't the problem, it's the process of breaking the glass.

[@bradley3549]

@@justinlabarge8178 I agree with you, but tracking the orientation would have been wise none the less. I think the way glass breaks close to an edge is different than near the center. The leverage forces seem to change the way the glass fractures in strange ways.

[@jbkunk9611]

I've been using a water jet to cut glass almost perfect diameters. Stock is expensive 40/60 beam splitting glass .06 thick. Biggest problem was cracking from the glass flexing but once I got some new Rino board and better way of securing I was almost there. I found penetrating it was also a chance to crack so I set lead in for the amount I need to cut from and edge about .100" then start my circle. I feel your pain brother!

[@richard2121]

Could you flip the glass and score the other side. Might need to drill some holes for some locating pins.

[@ifeelcnc4891]

You can try milling glass with carbide endmill just like Stefan Gotteswinter showed in one of his video. I think it is worth a try but it wood be a bit mesy but you can use vacum cleaner.

[@Clough42]

Yeah, I considered it. A diamond burr and flood coolant also works well as shown on Applied Physics. But it's also super slow and messy.

[@Andrew_Fernie]

@@Clough42 Applied Science. Ben has a the Advanced Patience & Tenacity plugin installed.

[@rontidd6551]

I suppose you checked the bearing and the play on the cutting edge. But i might blue a bit of metal and run concentric circle marks - pitty it wont cut sharp corners, or you could try a 3d printer trick with inscribed square-circle-triangle.

[@InRegardsToMetal]

This might be a bad idea, but if you had a consistent setup, maybe you could score the glass on both faces? Perhaps that could prevent the cracks from propagating with a taper.

[@criggie]

What if the CNC machine did several laps of the circle rather than just one? Could you make it repeatable and flip the piece, and score the other side too? Does the carbide leg angle stay consistent after the first cut - might need to slow that transition speed? I guess the final option is to build a lathe with a short 50mm bed length, a selection of big suction cups for a chuck and another one in the tailstock, and capable of spinning the whole workpiece in a bath, like a wet grinder? POC the idea by putting a pair of 8" toilet plungers in the tailstock and chuck with a large plastic bucket resting on the ways. You'd only want ~10 RPM or so, and a toolpost mounted grinder. It could be messy and silica dust is noone's friend.

[@bruinflight]

brilliant work!

[@roughlyEnforcing]

dude this is fantastic

[@shodanxx]

What about drilling two dowel pin holes for registration Place glass down, score top surface flip the piece, score the other side in the exact same place. Also, what about using a 3D printer for holding the tool ?

[@dnwheeler]

Because the cutter is dragging slightly behind the center of the tool, the circle will always be slightly smaller than intended. A little math should be able to fix that. For the "oval-ness" problem, does it matter where you start the break (i.e., is the oval aligned with, or perpendicular to, the starting point)? Also does is matter if you start the break by applying pressure to the inside vs. the outside of the cut? I suspect the break is applying angular pressure, causing the slight bevel. Maybe a large circular ring could apply even pressure to break it evenly.

[@VorpalGun]

Would it help to do this on the lathe? You wouldn't need the swivel mechanism then, and if there is any play in that part it would become a non-factor. EDIT: Also since the amount of "ovalness" is pretty consistent, are you sure the CNC machine is properly calibrated between the x and y axes?

[@Skyliner_369]

the only way to really go perfectly to dimension and circular is to grind it in. the nature of the way glass fractures won't allow it to go as precise as you'd like. However, now I'm curious if you can get the scoring front and back to line up perfectly, because if you can, it might help a tiny bit, make the deviation about half, by directing the stress paths. If that makes any sense. --edit: wrong word

[@rokk8041]

Stefan Goteswinter describes how you could CNC mill plain glass using "normal "end-mills in one of his videos, there was a NASA paper written on it. Might give that a try.