After which I inevitably put together a completely unusable cutterhead prototype.

My intent all along was to use the same tattoo gun brass cam as on the tiny motors, but last weekend I couldn’t find a tiny hex key for the set screw to remove it from the previous prototype and affix it to the BLDC motor. It being a very short drive from Missingtherighttoolville to neighboring Badideatown, I printed a plastic cam and new connecting rod to fit a standard skate bearing. Let me tell you, turning on that cutter was like holding a powerful vibrating thing with a razor-sharp blade in your hand.

It didn’t cut the fabric particularly well, either. It had plenty of torque but lacked either sufficient speed or travel to cut unclamped fabric; it just shoved it out of the way. (It cut just fine when the fabric was held in tension, but that’s not the objective.) The eccentric mass of that heavy skate bearing did not motivate me to turn it up faster, particularly when I already had the other cam in mind.

Since I was missing the hex key for the brass tattoo cam’s set screw and since somehow I’ve never had more than a few random hex keys, I ordered a cheap collection. When they arrived the next day, I removed the cam from the tiny gearmotor and discovered that the BLDC motor doesn’t have the 3-mm shaft it looks like it does. No, it’s 3.175 mm, aka why is the world manufacturing new high-tech products with tapped metric holes and a 1/8″ axle???

I ordered more stuff — a 1/8″ end mill bit to drill out the cam again and a pack of 1/8″ ID bearings to support the motor axle near the tip, since the eccentric load was going to be pretty far from the motor’s internal bearing and the 3-mm-ID bearings I’d just bought were going to require too much force to get onto that shaft.

This morning I finished printing new mounting parts, did a sloppy job of milling out the cam on the Dremel drill press (you’d think I’d get something better-suited for precision work), and fitted everything together. Unfortunately, the tip of the motor axle is filleted and the cam can’t be drilled very deep; and the 1.5-mm hex key is poorly manufactured and can’t tighten the set screws well; so even after borrowing a second set screw from my other cam, the cam works itself loose from the motor shaft before I can get the assembly up to a high speed or do more than the briefest test cut. To that point, it was vibrating less; but more less would still be preferable.

I’ve now ordered a set of small Wiha hex keys (I’m done messing around with the cheap ones), some spare set screws, and a counterbalanced tattoo cam.

Quickly swapped into the same prototype cutterhead, with predictable results: It has a higher no-load speed but bogs down in the cut. This style of gearmotor won’t be my solution.

YouTuber How To Mechatronics has a stellar video How Brushless Motor and ESC Work and How To Control them using Arduino explaining all the theoretical and operational details of BLDC motors and their electronic speed controllers (ESCs). If you want to know (pretty much) everything about them, go watch! Or for a quick overview, keep reading.

BLDC motors resemble steppers in that they have coils that must be energized in sequence to move the rotor. Unlike (typical) steppers, hobby BLDC motors have three (sets of) coils with a common node connecting one end of each coil (set of coils). Passing current in sequence through two at a time of the three wires energizes:

• V+ → A coil forward → common → B coil reverse → V-
• V+ → B coil forward → common → C coil reverse → V-
• V+ → C coil forward → common → A coil reverse → V-

with the effect of using one coil (set of coils) to pull the rotor’s permanent magnet(s) toward it/them and another coil (set of coils) to push the rotor magnet(s) away.

And the unused wire during each step is used to sense back EMF, to close the feedback loop for managing acceleration. Slick!

Such an arrangement requires a driver; and because these motors are targeted at (RC) hobby use, the BLDC drivers (electronic speed controllers) are commoditized, configurable, and convenient. The ESC has an XT-60 LiPo battery plug for power, bullet connectors to feed the motor … and a servo input for control.

That’s right — you inform the ESC what speed you want by feeding it a servo control signal, with short pulses being interpreted as off and longer pulses being interpreted as faster. Moreover, the user guide for my ESC describes what I gather is a common process of powering up the ESC with the servo signal at max, waiting for the ESC to beep (by sending audio-frequency signals to the motor, using it as a speaker), and turning the servo signal to min to calibrate the ESC’s expectations for this particular servo signal transmitter’s range of pulse lengths. That’s only one of many parameters configurable through the simple servo control interface; it’s really well thought-out.

The only things left were power — the rackmount power supply it’s sitting on — and a servo control signal. Younger me with more free time would have programmed an Arduino; but recent me bought a servo tester, which even displays the (alleged) pulse width in μs. Patched together, the knob controls the motor speed and with the addition of a 3D-printed base, I can test the motor!

Worth noting is that the visible housing is part of the rotor — the stator coils surround the axle and the permanent magnets are affixed to the inside of the housing. That’s disconcerting to handle and is the reason for the grippable base.

One more note: Hobby BLDC motors have a “Kv” rating representing the maximum rpm per volt of the power supply. The same size package could have more turns of fine wire for lower current and higher torque or fewer turns of coarse wire for higher current and higher speed, given the same load. Rotor Drone Pro has a nice overview of Kv and motor and propellor selection for aircraft.

Expenses

Everything was about the cheapest I could find on Amazon:

• 850 Kv BLDC motor: $14.99
• ESC: $16.49
• servo tester: $11.99

And now I remember how I stumbled across the idea to use a tattoo gun motor — Nick Poole’s SparkFun post on building a tattoo gun from one of their cute widdle gearmotors, which are available in a wide range of gear ratios and output speeds. So when the tattoo motor didn’t have enough torque, I ordered two different speeds of the micro gearmotors from SparkFun, following Nick’s path of innovation.

For the crank to transfer power from the motor to the connecting rod, I’m not going to beat the convenience or price of the tattoo “cam” with the bearing and pin already in it — that’s exactly the type of commodity component I’m hoping to be able to use. So I ordered a couple more of those, bearing in mind a caveat — these “cams” are made for tattoo-motor 1.4-mm shafts and my gearmotors have 3-mm shafts. A little drilling was in order; so I ordered a 3-mm end mill because flat-bottomed hole, you make the cutter cam go ’round.

Not being a machinist, I don’t have cool tools for work-holding, so I printed an enclosure to increase the odds of actually drilling the brass cam.

I’d have liked to clamp it down; but even having this to keep my fingers away from the bit worked out quite well. I got it drilled this weekend; Dremel drill press FTW.

Fitted together, it looks like a tiny model of some steam-era mechanism.

With the “cam” prepared, I next had to model and print a new motor mount.

These shots turned out so nicely, I can’t choose,

so I’ll just use all of them.

Predictably, this version had the opposite problem from the ungeared tattoo motor. It showed no inclination to stall in the foam spoilboard; but the blade was moving slowly enough that it wouldn’t cut its way in from the edge of the fabric — I had to put tension on the fabric and make a plunge cut to get started — and the fabric tended to cling to the blade, oscillate with it, and push in front of it.

Looks like I should try a gearmotor with a lower gear reduction and a faster output — the fact that I got a super-easy cut with the tattoo motor and I got no proclivity to stall with this gearmotor means there’s still a chance there’s a sweet spot between the two. Unfortunately the 900-RPM motor that I tried was the faster of the two I’d ordered; the 4900-RPM motor that Nick used is sold out at SparkFun and I can’t find it on Amazon; and the 2700-RPM (next fastest) motor from SparkFun would cost me about $25 with shipping, sigh. So I’ve ordered a 4000-RPM motor from AliExpress, which should be here in a couple of months. It should be the same form factor as the one I tried this weekend; so when it arrives, I should be able to swap it right in without having to update the case design.

I’m also thinking about an electric toothbrush motor. Although it vibrates rather than reciprocates, I think it has a fair chance of getting the blade to cut due simply to the speed at which it will move. It’s worth a try; and I’ve ordered one whose Amazon listing boasts that “friends to feel can only buy” and “material thick.”

There, got that out of the way.

If I’m going to try to develop a CNC fabric cutter, at least the prototype and ideally the final version will use all commodity components plus easily-fabbed parts (3D printing, lasercutting, easy woodworking with common shop tools and not requiring a high degree of accuracy). The self-imposed choice of commodity components makes me want to use a readily-available X-ACTO® blade as the knife.

If I’m going to try to develop a CNC fabric cutter, I want to develop a working prototype cutterhead first. It’s the only part of a CNC fabric cutter that’s significantly different than a 3D printer or mill. So if a cutterhead can be made to work, the rest is easy; if it can’t be made to work, the rest is moot.

When I started thinking about this last fall, I hoped to find a very small, lightweight reciprocating power tool that I could use at least to power the blade in the prototype; and if it turned out to be suitable for use in a final product, so much the better. I searched both online and in person looking for options and didn’t come up with anything.

Thinking about doing it myself, I considered three common means of converting the rotary motion of a motor TBD to the reciprocating action of the blade:

• A crank has a connecting rod attached at one end to a pin protruding from a rotating disc and at the other end to the piston; this mechanism is familiar from steam locomotives. The linear action is perpendicular to the axis of rotation.
• A cam uses a disc or other shape eccentric to the axis of rotation to push and pull a cam follower. The linear action is generally perpendicular to the axis of rotation, though the follower of a cylindrical cam may track in a groove to create linear motion parallel to the axis of rotation.
• A swashplate uses a disc concentric with but not perpendicular to the axis of rotation to push and pull a follower. The linear action is parallel to the axis of rotation.

Of these, the crank with connecting rod struck me as the easiest to find and/or fabricate. And then I stumbled across something that many people already know: tattoo guns have tiny motors and crank discs with bearings already in them, quite affordably. I ordered what I lovingly call “Amazon’s best $13 tattoo gun” and disassembled it upon arrival.

I designed and printed a good-enough blade holder and connecting rod and joined the two with a good-enough machine screw (’cause I’m not planning to put a lot of miles on this particular iteration).

The tattoo gun’s brass “cam” (that’s what they call it) has a post press-fitted into a bearing press-fitted into the body and comes with a rubber bushing to couple the original wire needle assembly onto the post, which serves me well for my connecting rod.

My printed housing has spots for the motor terminals and nut traps to attach the piston’s “cylinder” on the working side.

And it works! It actually cuts fabric, laid on top of styrofoam spoilboard.

Pros of this method:

• It cuts fabric!
• It doesn’t seem to be prone to pushing or bunching the fabric; the reciprocating action does its job.
• The distance of travel provided by the tattoo “cam” seems pretty appropriate for cutting fabric, which is convenient.
• It’s exhilarating to hand-hold a small mechanism ending in a razor-sharp blade that’s reciprocating at audio frequencies.

Cons:

• The motor is ungeared and has very low torque, so it stalls easily when the blade is in the styrofoam.

So, proof of concept, yay! Now we need more torque.

Yes, but I don’t know much about it and my acquaintances haven’t worked at places that use it. What I’ve been able to find appears to start around $30-40K for a fairly short table, with extensions available. That entry pricing only makes sense for a fairly high volume of product.

Could we do better, and is there a market for it if we could?

The Need for Low-Volume CNC Fabric Cutting

Nearly all of the attendees at Kathleen Fasanella’s sewn-product manufacturing boot camp have some interest in the sewn-product industry, but I’d say it’s a minority who are already working it as their primary occupation. Most of the attendees are entrepreneurs and would-be entrepreneurs who want to learn more about the production phase; and a high fraction work in or have worked in IT or engineering, which is part of the appeal to me.

Over dinner one evening at last fall’s boot camp, at the nerds table — you know, the one that keeps attracting people from other tables with our loud conversation and raucous laughter, and then they leave after two minutes because they find out we’re talking about characterizing the accuracy of our 3D printers and complaining how the command-line invocation of the open-source CAD software can’t yet accomplish everything that the GUI does — Gillian mentioned her desire for an affordable, low-volume CNC fabric cutter.

Her entrepreneurship goal is made-to-measure business suits for women — a market that is already well-served for men and completely lacking for women. She can do the algorithmically-based pattern drafting from (the right set of) measurements; but to be able to hit a price point to develop the idea as a successful business, the labor, duration, and space requirements for hand-cutting the fabric felt cost-prohibitive; yet the current commercially-available CNC fabric cutters are even more so.

The conversation moved on but that idea really stuck with me and I’ve been thinking about it since. Hobbyist CNC has become popular enough that even the components to build it yourself have become commoditized, so this is a good time in the history of the world to build a hobbyist CNC fabric cutter.

It should have open-source hardware and firmware. If it were to have free-as-in-beer-and-speech plans, and a distributor from whom you could buy your choice of vitamins, vitamins plus printed parts, a complete kit including frame and table, or an assembled product, the range of prices should make it available to entrepreneurs who can’t afford one now. More successful sewn-product designer/manufacturers creates more jobs in the sewn-product industry, providing more stable work across the country and around the world for a predominantly vulnerable population of women of low socio-economic and educational status. We’re not gonna solve the world’s problems; but if we have a chance to make a difference, we should take it.

How Can We Cut Fabric?

Blades or lasers, right?

Lasers are cool, but there are a lot of issues with DIY laser cutting:

• Power.
• Safety.
• Liability.
• Scorches some fabrics; melts others.
• You need sharks to attach them to, and who has sharks?

So let’s just agree that we’re talking about cutting with blades.

How should the blades move through the fabric?

• Drag.
• Scissor.
• Rotate.
• Reciprocate.

The very successful Cricut® desktop CNC cutter uses a drag knife, as does the DIY CNC cutter whose knife is shown above. The knife is mounted to a gimbal with the knife tip slightly eccentric from the gimbal’s axis; as the knife is pulled around the workpiece, it drags behind the gimbal’s axis and the gimbal passively swivels to keep the knife’s cutting edge pointed in the direction of the cut. This works really well for cutting fairly rigid materials or materials attached to a rigid backer, but will tend to push flexible materials in front of the knife.

An animatronic Edward Scissorhands would be fantastic for frightening away intruders; but I don’t want to have to design a CNC scissor mechanism.

A mechanized rotary cutter has potential, but also a couple of drawbacks I’m not keen on. Cutting with the edge of a large circle tangent to the surface you’re cutting on means you tend to have to overcut at corners, which limits how tightly you can pack pieces together. It also requires quite a bit of downforce, which in turn requires a lot of rigidity from the (say) 72″-long X-axis gantry. That rigidity increases the cost and mass of the gantry and the power required to move it, and lowers the Y-axis acceleration to have to sling that mass around, increasing job times.

So I’ve been thinking about a reciprocating knife blade. It should oscillate up and down at a high speed, possibly moving straight or possibly following a tall, narrow elliptical path. The cutting edge should be vertical or canted slightly in the direction of the cut. It will need to be actively aimed in the direction of the cut rather than passively dragged on a gimbal.

A vertically-reciprocating knife blade needs to dip below the lower surface of the fabric. That could be accomplished by having a shallow foot under the fabric to lift the fabric enough to be cut, like the industrial electric knives; but the stand that attaches the foot has to be maneuvered in from the edge of the fabric, so you can’t make plunge cuts in the middle.

An attractive alternative is using a spoilboard under the fabric. Foam board insulation is inexpensive and available at any home supply store … though it remains to be seen whether it gets torn up so quickly that the entire idea is impractical.

In the garment industry, “made to measure” refers to clothing whose pattern is made from measurements of the individual wearer and then either drafted algorithmically or pieced together from library parts for each possible value of each collected measurement.

In product development and in made-to-measure production, a number of factors of fabric cutting are different than in mass production:

• The manufacturer or producer may be making a single quantity, certainly only a small quantity.
• The manufacturer or producer is unlikely to be making multiple sizes at the same time.
• The above factors rule out the long markers used in mass production.
• The above factors rule out lay-ups of many plies of fabric.
• The manufacturer or producer may not even have a marker — pattern pieces might be hand-drafted and on separate pieces of paper or oaktag (card stock).

I can’t speak definitely about every possible cutting scenario in product development and made-to-measure production, but I can cover common cases.

Even for a single item (product development or made-to-measure order), the patternmaker may still make a marker with all of the pieces in their desired position and orientation for cutting out of the fabric, particularly if the pattern is being developed in patternmaking CAD software. If a marker is used, the marker for one item will of course be much shorter than the marker for multiple iterations of multiple sizes of the item.

Or if the pattern was drafted or altered by hand, or for a variety of other reasons, it may be in individual pieces on regular or heavy paper. In that case, the pattern pieces will be laid out on the fabric and traced in chalk, china marker, or something else that shows up on whatever color fabric is being used.

Then in either case, the pieces are likely to be cut out by hand. That might be done with shears or it might be done with a smaller electric cutter, like the one that my friend Raquel is using to cut hiking pants samples for her fledgling business.

Both hand-operated shears and small rotary or scissor-action electric cutters can cut a few layers of fabric; so they could be used to cut, for example, four plies at a time for an order of four dress shirts.

I volunteer in the sorting room at my local thrift store about once a week and a few years ago, one of the guys who processes electronics and appliances showed me this thing out on the sales floor and asked whether I knew what it was. Pizza cutter? Linoleum cutter? Don’t know; but it’s a beautiful piece of machinery, so I bought it.

By the time I left the store that night, we’d figured out together that it was a fabric cutter. And as it happens, by now I know a lot more about the topic than I did then.

Kathleen Fasanella and Fashion Incubator

Kathleen Fasanella is a sewn-product industry veteran who has decades of experience and who wrote the book on sewn-product manufacturing. She’s been blogging about the industry since 2005 and gives away a wealth of information on the business, including insightful articles on lean manufacturing written in the context of her industry but applicable to others as well.

Twice a year (except not right now), she conducts a manufacturing boot camp, taking twenty-five participants into her factory for a long holiday weekend and guiding them in cranking out a run of around a hundred of a product to donate to a local charity, starting with raw fabric and ending with a product at a slightly higher price point (that is, a slightly nicer good) than what you’ll see at your local store. This sounds insane and impossible, but it’s not actually … impossible.

I signed up thinking I’d get to see some lean in action (particularly pods) and hoping it would spur ideas I could apply in my IT work, the way The Phoenix Project adapts the lessons and storytelling style of The Goal from physical manufacturing to the software development lifecycle. I’ve returned to and volunteered at her boot camp because of the great group of people getting together with enthusiasm and energy to learn and to make something that benefits society.

Industry Fabric Cutting

In the sewn-product industry, fabric for mass-produced goods is generally cut the same way regardless of whether it’s for garments, messenger bags, or tents. (The primary exception is heavy goods like felt and leather, which are press-cut with a die.) (If you’re an industry expert and take such exception to my generalization that you feel compelled to comment about it, be sure to include why the failures of my generalization are relevant to the topic I’m discussing here.)

A custom cutting guide called a marker is drafted (probably in CAD) and printed to lay on top of the fabric. The marker contains the pattern for all of the pieces of all of the sizes to be cut from that type of fabric, packed as closely together as possible to reduce waste, hence cost. In fact, the marker typically contains multiple iterations of each size, because no matter how well you pack the pieces, most markers will end up with a somewhat jagged end; and the unused fabric at that jagged end is waste. The higher the count you put into a single marker and the longer it gets, the more repetitions you have (multiplied) before that jagged end of waste (constant) and the higher the overall yield. Kathleen reports that her two 6′ x 80′ cutting tables are mid-sized for industry.

You unroll the fabric onto the cutting table, getting it perfectly straight and aligned, stacked in as many plies as you need for your production run, up to many inches thick. You then put the marker on top and cut with an electric knife, like this:

On this very small production batch for boot camp, that’s Kathleen and Mr. Fashion-Incubator all safety-ed up to run the knives, which oscillate a 10″-long razor blade up and down at something like 3000 cycles per minute, IIRC. Just stop and ponder that for a bit.

The knives are great; but watching them in operation, it’s easy to tell how much skill and experience they take to use. With the pieces packed tightly together, go a little past then end of a cut and you’ve nicked the next piece. You may have to trim a tiny sliver off an edge because two slightly differing curves were packed side by side and you can’t cut both at the same time. The stack has some give and you want it to remain perfectly vertical so all the pieces in the stack are identical.

But in exchange for the skill and safety considerations, you get a huge yield, fast.

My older cutter at the top is a slitting machine, apparently for making long cuts to take fabric down to a narrower width. You can tell by the blade shape that it wouldn’t navigate curves and corners well.