Wednesday, September 29, 2010

Filament Drive



Today I worked some on the filament drive system.  I chose to do things a little differently than what I have seen others do.  I strongly dislike live shafts in mechanisms, and I use dead shafts whenever I can get away with it.  I also wanted some reduction, but didn't like the backlash present in the lower cost options for gearing.  Timing belts and pulleys are relatively cheap, and can be very compact.  These are MXL pitched pulleys.
I use a .25" shoulder bolt for an axle.  They come reasonably precise, and with a hardened, ground shoulder perfect for running the needle roller bearings on.  The drive roller itself is a piece of 360 brass .625" in diameter.  The center was drilled and reamed to the bearings OD, .4375", for a nice press fit.  I then bored the pulley out to fit this assembly.  This direct coupling of drive roller to reduction avoids having to fasten each one to a live shaft.


Here's a mock up of the whole setup, minus the shaft that the idler rides on.  The idler is a rubber idler roller, with a bronze bearing in the center.  I still have to make the frame for it, I'm waiting on some hardware to come in from perennial favorite McMaster.

Monday, September 27, 2010

Heatsink

Progress is slow, delays in the fabrication of this part.  I had moved the CNC control to a laptop machine, and neglected to disable some power saving functions.  When the computer turned the display off after 15 minutes of inactivity, it also was somehow able to put a hiccup (or perhaps shut down completely) the USB port driving the CNC.  The mill would then freeze, lose all of it's offsets (these tell the machine where the part it's working on is located)  and need to be rebooted.  15 Minutes into a 45 minute program.  Several times before I figured out the problem and corrected it.

Cutting the fins of a heatsink is generally a tricky one.  You want a lot of surface area, and that means deep slots, and thin fins.  There are a couple ways to do it, none of them fast.  It would be simpler to mount pre-made extrusions to the sides of an aluminum block, but I like machining things so I came up with this design.  It is sized to fit a 40mm DC fan directly.

 

To start the process of cutting the fins, it's best to remove as much material as possible.  Deep, narrow slots have issues with chip clearance, and the less material there the better.  I drilled these holes, the same pattern from both sides.  While I could have packed things a little closer together, I didn't want to risk busting through a wall, and messing things up.  Both sides were used because again, at much over 8-10 times the drills diameter deep, things start to get tricky.  These did do a good job for their intended purpose.


The depth of the part again poses a problem when trying to mill something like this.  The slots are .095" wide, so the most generally available "deep" endmill is .0625".  That will cut to ~.4625" deep, almost enough to get halfway through the block.  A shallow pocket on the face allowed just enough clearance to get the end mill halfway through.  A little more could be removed, but if this doesn't provide enough cooling as is, it's time for a totally new design.  


The rear of the block is slotted for clamping action on the stainless thermal barrier tube.  You can see the counterbored holes on the side which will facilitate the clamping action.  A little breakthrough into the fins was unavoidable with the size of these parts.


The last op was to put this hole pattern into the top.  The mount for this extruder will be of a 3 pointed design.  The tripod will provide equal contact, positive location, allow the head to move in case of a crash, and even electrically indicate such a crash.  It also could serve as a 3D probe, more on this later.
These holes match the pattern in this mount arrangement.  


Friday, September 24, 2010

Nozzle Part 2


Work continues on the nozzles,  these are almost complete.  All that remains to be done is to thread the base to 1/4"-28.  I had made 8 blanks, so I drilled 3 to .5mm (0.0197") and 3 to .3mm (0.0118"), leaving 2 "blank".  I think I'll try a little smaller at some point, perhaps .25mm or .2mm but I want to get the larger sizes working reliably first.  I might also want to attempt to get a smaller filament to make such a small nozzle size easier.


These are the thermal break tubes.  303 stainless steel was used because it machines so well while retaining much of the qualities of stainless steel, such as a relatively poor heat conductivity.  The diameter is reduced right after it connects to the heater block.  This will hopefully further reduce the amount of heat transferred upwards.  The wall thickness here is ~.025".  The bore was reamed to size for the smoothest finish I could get.  The upper portion will contact the aluminum heatsink. The only thing left to do on these is to cut the wrench flats on the largest diameter section.  


These are the blanks for the heatsinks.  1" thick, and 40mm square to match the fan.  There are a bunch of operations to do on these before they are ready to use as a heatsink.  The altered slightly from pictured earlier.  Instead of a separate duct part, I've made the heatsink match the fan size, one less part, and it makes the whole thing smaller.  I'm still not 100% that this arrangement will provide enough cooling, but it's worth a shot.  With any luck I can get these and the heater blocks done by the end of the weekend.  PLA has been ordered, but has not yet arrived, and I have pretty much finalized the extruder drive design.  

As is often the case when building things, I get ideas for how I can make it better.  Thinking about the construction of the nozzle, it occurs to me that given the right sequence of draw dies, one could easily draw cartridge brass to the right shape, without machining.  The wall would be no thicker than the original material, so the reduction would be very small.  A flange could also be molded on the end to attach with a flare style nut, perhaps even using an off the shelf fitting.  Something to think about for next time.

Wednesday, September 22, 2010

Nozzle Part 1

Today I started on the extruder nozzles.  These are made from 360 Brass, which is super easy to machine.  I'm saving the threading for later, as I want to grab what will wind up being the threaded section in a collet to turn the tip.  I'm aiming for everything around the actual orifice being no thicker than .020".  I used a straight flute carbide bit to get the holes very smooth for lower friction, and with a known angle at the bottom (135deg).   

More goodies continue to arrive.  I decided that the non linearity and tiny size/delicate wires of the thermistor were going to be a problem.  Above is the AD595CQ which will go with the bolt-on thermocouple behind it.  I still have to come up with a board for this, I am going to try to come up with a single sided shield for the Ardunio to hold this, the FET's for power driving, and stepper output terminals.  

Monday, September 20, 2010

Gathering Supplies


Supplies have begun to roll in.  Shown here are a single NEMA 17 from Keling and a Gecko 251.  I really like these drives, I use them in a slightly different form in the mill.  Four of these boards are bundled together along with an opto-isolating breakout board, which makes a lot of things easier and safer.  For new people to CNC the G540 is a great way to get into things, without having to do a ton of wiring.
Here the G251 will be driven by one of the Ardunio boards.  I got a little crazy at Sparkfun and wound up with a couple with USB, one non-USB variant, and a Lilly Pad for other projects.  Also some conductive thread and sew in LED's.  Rounding things out are an LCD to display critical info and some button input boards.
I ordered the wire wound resistors, fans, and switching MOSFET's from Digikey.  I also picked up some Thermistors, but after some consideration have decided to go with a Thermocouple for temperature sensing.  I've read up a bit on the non linearity within the temperature range that I'm trying to measure, and I think it's best to avoid the workarounds to the problem.  Although I'm not sure exactly how critical a role constant temperature plays, getting a linear temp reading makes maintaining it easier.
The first of several McMaster orders is a away, and I'm closing in on a design that I'm prepared to machine.

Wednesday, September 15, 2010

It begins

Recently I have become aware of the existence of open source 3D printing technologies, most notably www.reprap.org.  A bunch of reading of that site, and other associated blogs, and I am determined to build my own printer.  It's not my intent to make a direct copy of the Reprap design, or of several commercial designs out there, but rather to build the best printer I can make for a reasonable budget.  Self replication is not one of my prime movers.  The RepRap project has helped me an enormous amount, so I do hope that by documenting my build, some folks working on that project might gain some insight or ideas.

I do have a lot of tools, both hardware and software to use to do this.  Over the past year and a half or so, I've put together a pretty cool 4 Axis CNC machine.


The major structural components are made from 3"x1.25" 6061 bar stock aluminum.  At my last job there was about half a pallet of cut off ends of 20' sticks, maybe 22" long.  The fact that this material was free to me was a major factor in the construction of this machine.  Most of the work done building this machine was done with a HAAS TM1, which is a small toolroom/prototyping CNC mill.  Each axis has two frame rails, which are connected by cross bars.  These cross bars are doweled into the rails, and bolted.  The sections were then machined flat to accept the recirculating ball linear guides.  These are a relatively cheap variety, but almost certainly overkill given the envelope, overall speed of the machine, and spindle horsepower.  They do provide a very low friction, as well as slop free motion.


To get things moving ball screws were chosen.  Two nuts on each screw with Bellville springs between them take up any backlash between the screw and nuts, and dual angular contact bearings at the driven end eliminates it from that area as well.  An old cast iron machine table was "recovered" from a surplus place, and I've reused the head from my old Sherline mill.  It's not an ideal solution, one of the current projects involves replacing it with a Sherline "industrial" cartridge spindle, and reworking the head of the mill a bit.


Driving the system are a set of NEMA 23 sized steppers driven by Gecko drives.  I really like these drives, as they have a good feature set at a relatively reasonable price.  Good service and support as well.  Control of the system is a bit of a compromise.  Due the the vagaries of legacy hardware support,  coupled with newer tech such as power management systems, modern parallel port implementations can frequently be lacking.  Hours of testing revealed problems at higher step rates that I needed to get the precision/speed I was looking for.  There are relatively niche motion control boards out there, but they are pretty expensive.  After testing several options including EMC2, I found the most cost effective way to use the computing hardware I had at hand was to pick up a Smoothstepper USB motion control board.  It does force me to use the Windows program Mach3 for control, along with several other minor limitations.  Although annoying, living with them is a small price to pay for a machine that cuts parts accurately and without missed steps.  A friend had a trunk full of surplus 8020 so I built a stand for the machine.

After the big pieces were squared away I converted the Sherline rotary table to CNC by building a table mount for it and it's stepper.  4 Axis programing is "interesting", either with or without CAM.  It has enabled me to make a bunch of cool parts that otherwise would have been impossible.


I'm sure anyone who is familiar with the RepRap and RepStrap is thinking, "Why doesn't he just make a thermoplastic extruder, and slap it on the side?"  The short answer is that these mechanics were optimized to move a heavy table and vise, and cut metal and plastics at moderate feeds.  At higher feed rates, the increased mass needed to handle the stresses of machining becomes a liability as you have to exert more energy to move it.

The plan is to build a thermoplastic extruder first, testing it with these mechanics.  The cartesian bot is relatively trivial to build, especially with the experience of having built the mill.  It will be optimized to move lighter loads faster, while maintaining good accuracy.


Extruder design is based heavily on nophead's work.  Here is a model of the current design, nothing cut yet, as I still have a few things to button up.


The heater block will be made of 7075 Aluminum, with a split clamping design to grab the resistor.   7075 is chosen over other grades because it machines very nicely, is very strong, and I happen to have an appropriately sized piece of stock handy.   Nozzle is threaded male and made of brass.  The intent with both the nozzle and the thermal break is to have the flat bottom of the part bottom out on the machined flat bottom of the hole.  Threads on the nozzle and thermal break are relived from the end to eliminate tapping to the bottom of the holes.  The thermal break will be made of stainless steel.  I'm using Alloy 303, as it is considered the most machinable of commonly available grades.  It will thread, drill and ream easier which is a great help when dealing with stainless.  The thermal break will be turned down for a section to lower the ability to transfer heat upwards, and the heat sink at the top will hopefully cool the top of the thermal break enough.  The black portion of the diagram is the mount/duct for a 40mm fan.  If this arrangement does not suit, I plan to investigate water cooling.  The only PTFE used is to insulate the heater block, and for wire insulation, none under mechanical stress.  There is also no adhesive used in this assembly, making rebuilding, adjusting, and modifying easier.