Re: Filament Extruder - Convert pellets to filament
Baby's first Arduino!
Managed to blow out port D on the 328p already, whoops.
LCD works great though! Not bad for $0.79.
You are not logged in. Please login or register.
SoliForum - 3D Printing Community → Filastruder → Filament Extruder - Convert pellets to filament
Baby's first Arduino!
Managed to blow out port D on the 328p already, whoops.
LCD works great though! Not bad for $0.79.
Most of the patent is about FDM, why tight tolerances are important, and how they same up with the diameter and deviation they use (1.77mm +/- .06mm) It's really only the last two (of 19) pages that talk about the method. The novelty of their method is using a pair of lasers set at 90 degrees to watch the diameter of the filment, and using that measurement to adjust the speed of the puller. They list the alternative of setting the laser micrometer near the die to reduce the lag time between adjustment and measurement.
They also specifiy cooling with chilled air rather than water, using an air knife near the die to halt the draw down, and then more general air cooling on the way to the roller. The roller could be serrated (like the hobbed bolt) or urethane.
They specifiy a drawdown ratio of 1.05:1 to 1.75:1, which is pretty broad, but less than what I had been doing. I was already getting the idea that I should probably go with a smaller ratio than the 1.8:1 I had been using.
The only thing about the whole patent that is different from the usual process of extruding filament is the use of a laser micrometer to automatically adjust tension to achieve the desired diameter.
$0.79 FiLAStrudEr -- the low cost version... LoVE it! LOL
Baby's first Arduino!
Managed to blow out port D on the 328p already, whoops.
LCD works great though! Not bad for $0.79.
Very Nice
Tim,
I just bought my first arduino uno r3 yesterday and printed this little guy for it.
http://www.thingiverse.com/thing:28971
Where can I get a 79 cent lcd?!
Tim,
I just bought my first arduino uno r3 yesterday and printed this little guy for it.
http://www.thingiverse.com/thing:28971Where can I get a 79 cent lcd?!
http://www.ebay.com/itm/NEW-ARDUINO-12- … 2a27b326b7
Shipping breaks the bank.
There's another source for them at $0.76 qty 25+. That's where I got the $0.79 from, i'd misremembered $0.76.
Interesting little issue - it takes about 100mS to service the LCD. While this is happening, I can't generate pulses to the stepper motor. In 100mS, 1mm of filament is extruded. I guess it'll be fine, but you'll definitely want to optimize the code running on the microcontroller controlling the stepper.
Interesting little issue - it takes about 100mS to service the LCD. While this is happening, I can't generate pulses to the stepper motor. In 100mS, 1mm of filament is extruded. I guess it'll be fine, but you'll definitely want to optimize the code running on the microcontroller controlling the stepper.
I am sure that it is possible to do it all in parallel. I used a trick in a 68HC11 controller for a machine many years ago. I made the timer rate to send pulses to the stepper (or a multiple of it) as my main clock and ran everything else as part of that one big loop. It went to sleep at the end of each loop. The stepper always received evenly spaced steps with no software "jitter", as it was the first thing that happened after waking up from the timer interrupt. It worked perfectly.
Perhaps this idea will spark something.
Yeah, there's got to be ways to do it using timers and interrupts, but that's not how the arduino libraries handle it by default. I just thought it was interesting, and mostly posted it here so I'd remember to work on it. The thread is more a blog than anything else.
The answer I've gotten is timer interrupts. I haven't been able to find an actual example, but I'm guessing that rather than do a loop with a step followed by a 2000 microsecond delay, you have an interrupt that stops everything every 2000 microseconds to make a step. I'm not sure how that would work if you need the program to do something else that takes more than 2000 microseconds to complete.
I tried the PTFE nozzle where I set the bolt into the PTFE and then drilled through the bolt into the PTFE so they lined up. Still didn't work. The PEEK should be waiting for me when I get home, so I will try threading it into the plug. I have a feeling lining up a hole in brass with a hole in plastic cold end will always be problematic. It may be best to drill and tap the plug all the way through at 1/4" and let the PEEK be the inside face of the die. At 175-195C it should be fine with temp, the only concern would be holding up to the pressure.
The answer I've gotten is timer interrupts. I haven't been able to find an actual example, but I'm guessing that rather than do a loop with a step followed by a 2000 microsecond delay, you have an interrupt that stops everything every 2000 microseconds to make a step. I'm not sure how that would work if you need the program to do something else that takes more than 2000 microseconds to complete.
I originally tried this. Interrupts do not happen with the same delay between the interrupting event and the first instruction of the interrupt routine. However, if you put the CPU to sleep, it is always in the same state and wakes up in the same amount of time. Driving a stepper motor realtime is like a DSP functionality. Of course, I was not driving a step and direction control, but driving each motor phase directly.
If a process takes longer than 2000 uS to complete, it has to be broken up into two or more phases (and it can not be part of the calculations needed to decide the next stepper motor step). The calculations for the stepper motor limit the maximun stepper motor rate. Don't worry, it is only rocket science... LOL
I've been looking for a printable spool that requires no non printed parts that uses the least amount of plastic and I think this one just posted to thingiverse could be almost perfect except it would probably need a large diameter hole in the middle for our PVC pipe.
Nick- I use one posted here in the forum. I'm on my phone so it's tough to find it, but it should be searchable
Nick- I use one posted here in the forum. I'm on my phone so it's tough to find it, but it should be searchable
http://www.soliforum.com/topic/589/fila … oolsreels/
Do you the honeycombed would be better then the thingiverse one I posted? I'm interested in finding/designing the most efficient spool from a plastic/print time standpoint that cannot fail.
The one you posted could break if the coil gets wound too tight. I've had it happen with a version of that one. The concept would be good for the filastruder if the hooks could be hinged. It would be nice to be able to flip them open and pull the coil straight off the side. On the other hand a spool with a fully round center would be better just for maintaining a constant tension going into the spool.
Fully printable is convenient, but I finally just bought a bag of 608 bearings from Amazon and some 8mm threaded rod and 8mm nuts to have on hand. So many spools and other projects seem to use those things.
Probably depends on how your planning to spool. My concern with the one you posted is it could tangle easily if you're not careful. I can bang out the "rich man"'s spool in 2.5 hours. Plastic is cheap since I print them with filastruded plastic.
Peek threaded into the plug didn't work-
This was simple, the plug is tapped at 1/4" all of the way through, with the PEEK threaded into it. I didn't want to try drilling partway through the plug at 1.6mm and have the PEEK meet up with it halfway, because I don't think I can get the holes to line up no matter what. I don't have a way to insure they are drilled exactly in the center, exactly perpendicular to the face. Even if I drill the plug first, then run the bit through the plug to drill the PEEK, it might not hit the rod in the center. Then the position of the hole will rotate as the PEEK is threaded into the plug and there is no way to insure the openings are aligned inside the plug. I made a jig to hold the PEEK straight up and down and used a nail to make a starter hole in the center, but it still wasn't centered at the other end, 35mm later.
I think the best bet still is to drill a bolt, thread that into the plug, and then use the bolt to mount a cold end with an ID that is much larger than the die opening, 2.5mm. It might be possible with a lathe or something to perfectly line up the ID of a brass barrel with a PEEK insulator like it does with printer hot ends, but I'm not getting there with a drill press. There is too much runout with <2mm bits. I can't drill more than about 10mm into a brass bolt without breaking one.
So at this point I'm working on two options-
1. Stabilize the output of the extruder where there is not haul-off, only the pressure from the feed, so the filament can be handled and guided without kinking
2. Build a puller that can draw down the filament and maintain tolerance.
I'm not having much luck with #1 so far. Later today I will see how well I can get the puller to perform if I crank the idler bearing hard against the hobbed bolt and disregard the toothmarks. I have a plug drilled to 1.94 so I can see if running a slow speed with a small draw down ratio helps. If that configuration works, I'll work on replacing the hobbed bolt with a urethane roller that won't damage the filament. I'm also trying to decide if it is work making a puller where the upper and lower rollers are both powered, rather than a 3D printer extruder that uses a spring loaded idler.
If you use a haul-off with a sprung bearing idler, you could use that as a thickness gauge. If you extend the lever your bearing swivels on, it should be quite sensitive.
If you use a haul-off with a sprung bearing idler, you could use that as a thickness gauge. If you extend the lever your bearing swivels on, it should be quite sensitive.
The problem to overcome is that the measurement needs to be made at the point where the material is still pliable -- the point that is actually being stretched. Otherwise, what you are measuring may not have any relation to what you are currently controlling. There must be as small a delay in time/distance between the point of measurement and point of control.
It wouldn't be suitable for on the fly control, but it would certainly be a good gauge of tolerance, to see if it is wiggling more than it should. A laser would be a good way to extend the lever. I'm not sure how much the real time control would really be needed, I think if the micrometer were positioned at the nozzle, it would still be playing catchup. You would need some sort of PID or something that can anticipate that a swing in size will go beyond tolerance and make an adjustment in advance.
If all the parts are consistent, the diameter should be consistent as well. The measurement makes it easier to dial in the speed and temperature at the beginning. Once the diameter is correct, you should be able to let it run without alteration.
DePartedPrinter wrote:Is there a procedure for auto-tuning the PID?
I don't have it in front of me to double check, but I think you press and hold set 5 seconds,, tap it a few times until ATU comes up, change to 0001, keep hitting set until it dumps you back to normal screen, AT led will blink during the process. I think it took ~5 minutes or so.
Manual:
I went ahead and auto-tuned the PID and no longer have any temperature oscillation.
Nice! Filament diameter should be more consistent now.
I ran autotune this morning as well, hopefully it will help with the puller. I'm confident that the puller is consistent, so the trouble I'm having is at the extruder end. I was still working with the 3mm die, so it's possible that the large size is inherently less consistent. I'll have to turn the puller off and let it run some 3.5mm filament and see what tolerance it gets there. I almost wonder if there is some factor that tends to balance inconsistencies in feed from the extruder when the filament isn't getting hauled off. The haul off then adds a constraint which forces the variance to get applied to the filament diameter.
Taping a laser to the idler is working really well. It showed me that there is a .02mm bump in the hobbed bolt. I can easily see the difference in tolerance between store bought filament and pulled Filastruder filament. I need to print an idler with an integrated laser mount to keep it consistent, but it is giving what I wanted from the laser sensor idea. It would be simple to attach a ruler to the back of the motor support for the laser to shine on.
Ian- can you measure the motor voltage and current, and record whar the PID setpoint was on the 3.5mm die? Next time you run the 1.6mm die, do the same. I have some thoughts I'll expand upon once I get to a computer.
SoliForum - 3D Printing Community → Filastruder → Filament Extruder - Convert pellets to filament
Powered by PunBB, supported by Informer Technologies, Inc.