Engineering Roundtable - The SpoolMaster 9000 with Paul

I don't want to sound like I'm ragging on you, because that's a neat rig.  However a big part of my job is programming winders for sheet steel up to 3/4" thick, so here's a few concepts you might be interested in - "Dancers" and "Torque Control."  A dancer is a feedback mechanism that monitors the tension of your web.  I typically use inclinometers, but you can use encoders, reactors, differential transformers, or just potentiometers.  Let the dancer "ride" on the web, and then use it's feedback as an input to a PID loop controlling the torque reference of your winder or unwinder.
Now, to use torque control you can't use hobby-servos which all run in position control mode via PWM.  You can use "proper" servomotors, which are expensive, or you can use DC motors with a PWM driven motor driver board.  Use the current feedback as the setpoint variable in your PID, and you're golden.
And never drive two steppers on the same winder, that's rule one.

Oh, BTW, I like your idea of using the servo to distribute the thread, I'd been working on something similar and wasn't sure what I was going to do about that part, now I think I'll pinch your idea if that's OK :)

You forgot the most important part, the knob that goes up to 11 !

Impressive project!
Since the thread movement against the flat spindle somewhat replicates the motion of a sewing needle on a sewing machine, using a pre-tensioner mechanism similar to what's used on sewing machines might reduce jamming and make the operation smoother/faster. The pre-tensioner on a sewing machine is the arm just above the needle that bobs up and down with the thread looped through the end. Also, a grayscale optical encoder on one of the steppers could quickly establish jam detection if it deviates from the expected bit pattern, and provide a way to alert that it needs "assistance".

I gotta tell you, Paul... That's a really clean layout for the tool. The wiring is nice and simply, redundant parts are there when they make sense for simplicity and reliability. I wish I could have had a cool job like that when I was15!

I love the way you used 2 steppers. Completely simplifies the hardware at the cost of a trivial duplication of motor drivers. I find giving it the "9000" moniker a bit premature, though... what about when you add a better tensioner? You can't just START at 9000... that causes the "It's over 9000! That's impossible!" dilemma when you add a simple feature (Like brass trim.) and call it the "9001"

It needs some brass trim. Nothing says "Leet construction skillz" like wood with brass trim.

Think the Time Machine H.G. Wells built in "Time After Time" (Best time travel film ever!) It was all kinds of fancy and ornate!

I've been toying with the idea of making a programmable coil winder, but wasn't sure how to control the back-and-forth movement of the wire. A servo is just the thing, thanks for the idea!

Couple of questions:

1) it seems that if the thread jammed, the torque from the steppers would be sufficient to end up bending the cardboard. Do you do current sensing on the steppers to try and figure out a jam ? 2) Can the cardboard slide out of the two holders, or is the interference fit enough to hold it, you were able to slide it in pretty easily.

If you do a lot of this type of stuff, maybe a new title, toolmaker, http://en.wikipedia.org/wiki/Tool_and_die_maker . My dad was one, but he basically made fixtures and jigs out of steel.

My math says 60 turns times 5 inches per turn is 25 feet. Did I miss something?

Wow, this is awesome. I'm currently a MechE student, with a side interest in programming/electronics/mechatronics, so it's really cool to see things like this combining all of those different areas in a way that can solve a very tangible problem. I look forward to seeing more videos and designs of these types of things as they are made! It's really an inspiration to keep going with the studies.

I made something similar for my wife a long time ago when she had a project that required dividing skeins of yarn into equal thirds. No 3d printing or laser cutting, just hardware store parts. No dual stepper motors, just hand cranked. No arduinos, just a decade counter/7-segment decoder IC. Did use a hall effect sensor though. Kids these days.

Now, if version 2.0 could change the spool without operator intervention, I might forgive the overkill :-)

Why 2 stepper motors? Was more torque required?

Just out of curiosity, why not count the number of revolutions by storing the number of 'steps' transmitted to the motor, then dividing by the number of steps per rotation, instead of using an external hall-effect? I suppose it would be more prone to error...

When I saw 2 arduino, I immediately said "WTF 2 arduino!!". Then I thought about it. 2 arduino = it's simple and it works (KISS). Sometimes you need to get sh * t done! That is something I just recently realise. You can do something that is elegant, well thought out and your proud of it or something totally trash but works like the elegant version. I also like the saying "if you do something, do it right the first time" Like Paul said, he could of done that with one arduino. You think about it for 10 min and then code it. That would of require a little bit more time and debugging.(depending of skill) In my case, I would of tie the hall effect and e-stop to both arduino interrupts. The e-stop raise a flag and the half effect ++ counter.

Nice. And varying the speed of the main motor could almost eliminate the pulsating draw of the thread.