Thursday, April 15, 2010

 

Hydra's First Print!

So after working ~30 hours on this project over the last 3 days we finally printed a part on the Hydra platform! Some details on the print:

Printing speed: 40 IPM
Stepper driver movement method: Half-stepping
Resulting resolution: 0.00125"
Printing medium: ABS plastic
Nozzle temperature: 240C
Print bed material: plexiglass
Firmware: Hydra-MMM v1.4 (to be released to the public soon)
Host Software: Hydra-MMM v1.3 GUI
Gcode generation software: Skeinforge
Total time for print: 9 minutes

And without further ado, here is a picture of the test print in progress

A YouTube video of the print is here: http://www.youtube.com/watch?v=V-_Qu8GxbtI

We were actually able to extrude the top layer of the dodecahedron just fine which was surprising from past experience. I think our extruder speed is a bit on the slow side so we are actually stretching the filament a bit as the machine moves from point A to point B. This stretching effect puts it in tension and allowed to to span the void across the top of the object without drooping down into the interior of the part. I suspect that voids up to ~1 inch would be possible to span at the moment with further gains possible using active cooling or from further tweaking of the print speeds. You can also see from the video that we have a heated platform setup, but it was not used for this print as we are still waiting for our 1" wide roll of Kapton tape to arrive.

First impressions of the part were very impressive. The machine has incredible accuracy and rigidity compared to my first repstrap I built using a belt drive mechanism with a threaded rod frame. We still need to figure out how to get Skeinforge to vary the start point of each layer. For this print, each layer started in the exact same spot so we have a relatively prominent seam on the side of the part. I know the setting exists, but Skeinforge is a little cryptic and every time we think we found it, we export the gcode and the problem still persists. We got it to vary the start point on the base layers, but the walls all started at the same location. We will keep investigating, but if anyone knows a solution feel free to shout out in the comments! Even with this problem, the surface accuracy was astounding, but should probably be expected with a machine as accurate as Hydra. I can't wait to add some microstepping drivers and see if we can improve the already impressive 1.25 mil accuracy that we have with half-stepping.

As far as the software and firmware, this was the first real test of the Hydra-MMM firmware and it's handling of the stepper-driven extruder. Up until this point, most of the testing centered around X, Y, and Z moves. For prototyping, the extruder stepper motor is added and needs to be controlled by the firmware. The Hydra v1.4 firmware has the ability to independently move the extruder motor, which is useful at the start of a layer when you want to get the filament flowing before moving and at the end of the layer when you want to pull the filament back and stop the ooze. Before starting each layer we moved the extruder stepper forward a few steps to get the filament to start coming out of the nozzle before beginning the XY moves of that layer. This helped ensure that we didn't have any gaps in the part where the plastic wasn't being deposited. Currently, the extruder motor speed is defined by a variable E_STEPS_PER_INCH, which was first found theoretically and then slightly tweaked to match the real work results. We simply varied the number until 1" of movement resulted in 1" of filament being printed. This number could be decreased to stretch the filament (smaller filament) or increased if you wanted a thicker extrusion. Changing the feedrate of the XY moves will change the speed of the extruder motor to match this velocity, but the relation between these two remains constant. Later versions of the firmware may allow for variable speeds of the extruder motor, but at the moment the results are good enough that I am not sure this is required.

We will hopefully be getting the heated bed up and going tomorrow so we may have some better prints to show off, but in closing, we also printed a traditional test object on the machine before shutting it down for the night. Cheers!


Labels: , , , , , , , ,


Comments:
Very nice machine! I'm doing a new CNC. This time I'll do a extruder too, so I wanna know if your firmware Hydra-MMM v1.4 fit and run into a arduino with ATMEGA8 (or wich? 168, 328)?
 
At this point, the firmware is very robust and cannot fit on a standard Arduino 168/328. I have been trying to optimize it and shrink the space requirements as much as possible and I MIGHT be able to get it small enough that it can fit on a Arduino 328, but temperature control and the internal stepping would have to be disabled. I would suggest getting an Arduino Mega or similar clone if you want to use the firmware in its entirety. The Reprap/Makerbot platforms have both moved to this larger microcontroller so you would be future-proofing yourself on these fronts as well. It's only about $25 more to go from the standard Arduino to a cheap Arduino Mega clone so its not too bad. And for the number of additional pins and RAM you get, it is totally worth it IMO
 
You say that the resolution is o.oo125, but isn't that the positional accuracy of the machine, which doesn't take into account the thickness of the deposited plastic?

Your machine is awesome regardless.
 
Tom, you are absolutely correct, however, both measurements have relevance. The thickness of the filament (~0.45mm in our case with a 0.5mm nozzle) represents the size of each Z-layer and is a big factor in the accuracy of complex 3D shapes. If you are printing parts with lots of overhangs and profiles that change significantly over the height of the part, this parameter will affect your results greatly. On the other hand, for parts that have relatively uniform cross sections, the thickness of the filament is not that important. For example, when printing the minimug or a box, or most of the parts required for Mendel, the cross section of the part does not change much through all the layers. In this case, the positional accuracy (the 0.00125" number) would be the big factor. This would be what determines how much a circle would truly look like a circle or how smooth diagonal lines would be. The chord length of the triangles used in the STL export as well as several other factors will also effect this, but I can tell you from experience that this machine performs MUCH better than my previous attempts, all of which had the exact same nozzle and filament thickness.
 
So, If I understand correctly, the main differences between the hydra and a normal reprap is that yours is much more rigid and driven by a screw instead of the belt.

The results are awesome. Maybe it is worth modifying a Mendel to be screw driven.
 
@Brian Korsedal
and that it doesn't try to make its own parts :P
 
Correct. We designed the entire machine from the ground up trying to find a happy medium between the fast movement required for FDM prototyping and the slow, accurate movements required for CNC milling. It also had to be much more rigid as we are hoping to cut plastic and light aluminum with it as well. I think there is still one more big difference and that is how the firmware works. The Hydra firmware and host software was written completely from scratch and designed to support multiple toolheads as well as all the commands necessary to do CNC machining as well as rapid prototyping. The firmware is what is going to allow us to do some really neat things down the road in a couple weeks :)
 
and that it doesn't try to make its own parts :P

Well, it can cut its own parts using a spindle, or it can print high precision molds, for many of its parts.

And, inasmuch as it engraves and drills its pcbs better than a Mendel, it is arguably 'more RepRappish" than a Mendel. Arguing about names and definitions is fun, and occasionally very important, but so is just recognizing that good hardware is good hardware. :p

Nice bit of work, Clayton. (And to your teammates as well.) Nice bit of work.
 
Good point Sebastien. All of the parts used in the Hydra prototype were made on a standard 3 axis mill, with only a few requiring a lathe. We have already seen several users working on the Reprap project who found ways to use a drill and makeshift stand to simulate a lathe so maybe that requirement can be eliminated as well. Or maybe we can make an attachment for a 4th axis. But either way, it should be completely possible to make the majority of the parts with the machine itself. There are a few such as the table itself which are a little too big, but they could potentially be done by hand as well as the hole positioning is not too critical there.
 
I'm interested as I own the business http://www.streamline-power.com/ if this is a green product.
 
@Pat - at this point it is probably not the most energy efficient due to our low-end stepper motor drivers, but we hope to improve that aspect over the next week. It is, however, capable of using biodegradable as well as recyclable plastics. We were originally looking to use PE as our printing medium because it can be found in common milk cartons, but we eventually went with ABS just because of it's low cost and availability. Are there other environmental aspects besides those that you were concerned with?
 
Post a Comment

Links to this post:

Create a Link



<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]