Fyberpunk

As teased in the previous post, a few weeks ago, I acquired a Singer 15-91 in a № 40 Queen Anne cabinet. I rewired it, oiled everything that moves, and put it in my Necchi cabinet (with apologies to the BU Mira), but that's all standard work that's been documented a thousand times. Heck, I was following the adjuster's manual from Singer, why would I need to reproduce what they wrote back when the machine was new? No, I'm here to write about why the machine is currently in pieces after getting it all working perfectly.

And boy do I mean perfectly. I can easily sew through a few layers of heavy fabric by turning the handwheel with just my pinky finger! I could've just used the machine as-is, but as with the Necchi Supernova, I just can't let enough be enough. The geared motor puts very little drag on the machine compared to a typical belt setup, which is both a blessing and a curse. It makes the machine more efficient, but also means there's very little braking force when I take my foot off the pedal, so the machine coasts a long time. Not great for stopping at corners.

Terminology note

The motor in the 15-91 and 201-2 is often affectionately called a “potted” motor by the community. Coming from the world of electronics though, I'm going to use that term just this once. That's because to me, potting means encapsulating with a solid or tarry substance, which you'd never do to a motor that you still want to turn, haha.

Servos to the rescue

So how can I add some braking force to the motor? Shorting the leads when it's disconnected from power would work, but there's another problem, which is that I just don't love universal motors for sewing machines. They're usually so weak that you can stop them with just a finger on the handwheel, and they pierce badly at low speed because they lose torque.

So I'd like to try something a little more modern on it. Brushless DC motors are common, cheap, and efficient. Projects like ODrive have made controlling them with feedback pretty affordable. So, a brushless DC servo as a replacement motor for the 15-91, then!

New features

What benefits will this BLDC servo bring?

• True speed control, so the machine won't speed up or slow down as it goes from piercing leather to chiffon and back again
• Stopping on a dime when I remove my foot from the pedal
• Needle synchronization, e.g. always stopping with the needle out of the fabric, for a little more work
• Control by other electronics

The last two points deserve a little more discussion. The 15-91 has a gear ratio of 10∶51, so for every 5.1 motor revolutions, the machine makes exactly one stitch. I could set it up so that you always turn the machine on with the needle in a known position (e.g. all the way down, just coming up, or something like that), and then it could keep track of where the needle is as the machine runs. Or, as I'm planning to do, I could mount an angle sensor by the handwheel to read the needle position directly, which will account for the spring linkage between the drive gear and the main shaft. I figure I can mount the angle sensor to the original power terminal boss, and make it so it's easily removable for winding bobbins.

Why would I want other electronics to control the machine? My ultimate plan for this 15-91 is to make it into an open source embroidery machine. For this, I'll need an embroidery hoop mounted on an X-Y table so it can move under the needle, and I'll need a way to tell the machine to make a stitch. Converting it to have a servo motor will achieve the latter by giving me position feedback and a digital command interface; effectively, I'll be able to send it a message meaning “run the motor 5.1 turns” to make a single stitch. More efficiently, I can use the same setup to keep the machine running continuously, and only move the embroidery hoop when the needle is out of the fabric.

Status

So that's my plan. I've been modeling the necessary parts to make it a reality, and I'm about ready to start ordering stuff. Here's a render of the new motor assembly.

Most of the parts are available either off the shelf, and a couple of them can be 3D printed. The shaft on the right is the one pickle I'm in though, as it needs to be turned on a lathe just for the project. I can order them to be CNC machined, where pricing is just hilarious. One shaft for $200, ten for $400, or a hundred for $500. So basically, I need to figure out how many people will be interested in reproducing the project before I order, since I don't want to keep a box of ninety-nine shafts under my bed in perpetuity if I don't have to. I might see about getting a prototype made some cheaper way first, just to make sure the dimensions are right before placing a big order.

⁓Clara