Ok...
The Problem:
1.75mm filament needs to be cut into pellets for a friend's manufacturing processes. I won't be addressing the "Why" in this post.
The History:
Years ago I built a little machine that would save human time by cutting pellets. Here's a photo of the machine in action.
Considering it's been in operation for so long, the concept is solid enough to run with. It's essentially a stepper motor pushing filament through a filament sized hole. Another motor maneuvers a steel arm down to cut it. It does this about once a second, and currently runs a little faster since upgrading the stepper driver to a Gecko 201x, but that's really only a Band-aide on a rather large stream of issues.
The amount of pellets in the photo is about 2-3 hours of work. Not only that, but as you can also see, they are statically charged. So not only is it days to get enough material for any real use, it's sticky and hard to handle.
Design Alterations:
Over the past year or so I've come up with a few ideas I thought were worth looking into. They all have a few things in common; they're supposed to run fast, they're small designs in keeping with the old design. Lets run down the list.
Here's the design we've seen, with annotations in case it gets confusing. As it stands, maintaining the electronics is sketchy, and the computer/driver/wires are all stuffed in that little box under the feed motor. It gets warm in there. How it works is there's a little button on top that gets pushed in by the cutter bar, which indicates to the logic when to push more filament through. This is not a flexible system, especially considering part wear. The linkage system connecting the motor to the cutting bar is not very lined up, and is either too tight or too loose, causing partially cut pellets and a loud squeal because of the motor current chopping attributes. It's been modded to use a fancy stepper driver which will get it running for another few years, but it's currently a huge bottleneck if the end user needs enough material within the next hour.
This design was always in the back of my mind. It's very similar to the v1 design, but with a different power train for maximizing the cutting force and speed. It never made it into the prototyping stage because it lacked an increase in speed.
This design uses an electromagnet to force the cutting bar down. The idea is speed and force. The electromagnet was made, but thinking about it further kept me from pursuing this iteration.
Here's that electromagnet I was testing. I used an online calculator to figure out how many winds of which gauge I'd need.
There was another design involving interlocking gear teeth , but it was absolutely inflexible with pellet size and screamed 'wearable parts'. The only reason it was considered is I had an old stick blender motor lying around. Can't beat raw AC power...which brings us to our next design.
The Solution (v2)
This is a machine that solves a problem, right? It's got to run fast, no excuses. So let's replace that Nema 23 motor with a 1/2 hp motor, and let the feed motor run as fast as it can (reliably).
It's not exactly small anymore, let's call it medium. The plan is to give it all the features. The wearable parts are off the shelf, and it can accommodate changes in case the supplier of said part is no longer in business. Let's look at some more photos.
Using corrugated plastic as a footprint, I began figuring out where things should go. My plan is to have the high power stuff on the right, with 12v over on the left. I have no idea how this much electrical noise will affect my Arduino based circuitry, so I'm playing it safe. There's a noise filter right before the motor, so that should help protect other sensitive electronics on the grid. I didn't cheap out on the cutting blade coupler, and initial testing showed no worrisome vibrations. It's a 1,700 RPM continuous duty motor, so I think we're going to be okay.
Here's the (fancy) hardwood plywood that I use as a base. The motor still has grounding, however. It's 3/4 inch thick, which offers very little deflection.
After the metal plate is cut and assembled, I can begin final design of the feed motor and actually do some practical testing.
With the metal plate removed, here's the testing layout. The feed motor is sitting on the desk on the red plastic bracket. It was 3D printed in ABS and is beefy enough to take a beating. The funny white plastic object with wheels is my "Fila-meter". I did several iterations on this, some using compliant devices printed in ABS. It's designed to detect if the machine is receiving filament, and how much. I've already given my class an earful on that, so I don't feel it necessary to go on that tangent. The final version uses bearings, rubber coated surfaces, and metal bushings, so filament running through doesn't wear down the printed parts.
For the circuitry, my first thought was to do a solder-trace protoboard like all my other projects. However, after watching an encouraging episode of EEVblog, I spent a few hours learning KiCad, and ordered my first PCB a few days later. The image to right is actually the second revision. It adds diodes to transistors, rework the traces, move some things around, and added a Bluetooth module. Bluetooth and the Arduino based boards are going to be daughter boards, so my board is simply a breakout for the sensors and power. Both the old and new versions of the board work well, so no problems there. If you're even remotely interested in PCB design, watch a KiCad tutorial. Turns out it's pretty easy and lots of fun.
One thing, I wanted a large capacitor on the main power line to keep the logic running when the stepper driver is taking up all the juice. 47uf is probably way too small to do that job. However, it doesn't actually need it anyway, but everything was left as is.
This is the final control board assembly. The connectors point up to maximize the footprint. They also allow the wires to screw to the terminals before plugging in for ease of maintenance. The modules are easily swap-able in case something happens (I'm not an EE after all).
Now that everything is figured out, it's time to build the case that covers everything. I used my laser cutter to make the guard and electronics box out of 3/16 acrylic. It's not the strongest material, and if this were any other application the box would be metal and grounded.
Final build:
At this point I tested everything again, added some extra safeties to the code, and built the app (MIT App Inventor) that will talk to the machine. Let's go over the features:
-Bluetooth to optionally receive warnings or simply check in on the progress. The machine is fully functional regardless of Bluetooth connection
-LED flashes depending on the error, and reports this error in plain text to the Bluetooth app (vibrates phone + alarm if something goes wrong)
-Immediately shut down if the blade guards are removed (two magnetic sensors on the blade guard)
-Immediately shut down if the lid is opened
-Immediately shut down if the vibration sensor trips (drops off a table, something falls on it, motor becomes unbalanced, human flailing, etc.)
-Immediately shut down if any of the sensor wires are cut
-If any errors are present, the motor never turns on to begin with
-Internal resettable overheat fuse in the motor, resettable fuse on the side, and circuit breaker to prevent short circuits. If the relay welds to itself and doesn't shut down, the circuit breaker can be flipped off like a 30 amp light switch.
-Fila-meter to ensure filament is being processed in an expected way. If it's too short or stops detecting a feed altogether, the machine stops
-Hidden ion generator that pumps ionized air into the cutting chamber to neutralize static
-The switch that turns on the machine is isolated from the AC line, only 12v shock hazard if everything goes wrong
-Operates 70-80 times faster than the first machine
-Adjustable pellet size from 1-2.5mm via potentiometer.
Really, it's a machine designed to turn off.
Cheers
