Adding a build platform to the Stratasys

With the support nozzle on the Stratasys still acting up (it likes to stick in the ‘down’ position once the head gets up to temperature), I thought I’d try running without support material for a time so I can at least be back to making parts. While I could technically deposit ABS right onto the foam base (as I did with one of my very early prints when I first struggled with feeding support filament), this doesn’t make for nice part bottoms, as the ABS fuses vigorously to the glass foam, embedding the foam into the base of the part. Starts making a mess of the foam itself, too. So, I needed to make a regular build platform just like all modern FDM machines use, RepRap included (Stratasys stopped using foam bases years ago and now uses plastic build platforms instead – I should really find out what polymer they’re now using).

I decided to make a modular platform that I could swap into one of the carriers in place of a piece of foam. A piece of 10.125″x10.125″x0.5″ polycarbonate sheet serves as the ‘base’, into which I drilled holes to accept the pins that normally secure the foam. I drilled and tapped the 4 corners of the polycarbonate for 10-32 screws, and then drilled and counterbored matching holes in a 12″x12″ plate of Garolite from McMaster-Carr. Using plastic tubing as standoffs, 4 button head screws then secured the Garolite plate to the polycarbonate base plate. I wanted a good solid platform, as I knew that the Garolite itself would probably be a poor surface to print onto, and my intent was to clamp other surfaces to the plate with binder clips just as nophead does with his heated build platform.

The cap screw heads stick past the upper surface of the Garolite by just a tiny amount - I sure hope the nozzles will always clear. I still need to make some clearance cutouts in the Garolite so that the platform won't hit the nozzles during the Z homing portion of the cycle.

After trying to level the platform as best I could by tightening down the screws for adjustment (using a feeler gauge between the platform and ABS nozzle), I found that the Garolite is bowed downward slightly in the center. Hopefully this shouldn’t cause any major issues, though – I haven’t measured the amount, but I’m sure it’s less than 10 thou, which is the default layer thickness. This does raise a commonly overlooked use of support material, which is that the support raft helps to ensure that the bottom of the part is flat. Without a support layer, you’re at the mercy of the platform flatness itself. When a raft can be laid down, the first few raft layers can droop or smear to conform to the platform, as the flatness improves with each subsequent layer. Of course, this flatness can be destroyed by part warping, but that’s a separate issue.

With the platform reasonably level, I thought I’d heat up the machine and see if the generic ABS would adhere to it in some manner. I used an alcohol wipe to clean off the platform, knowing that my fingerprints and other gunk would make adhesion much harder. I let the chamber get up to around 55° C before I got too anxious and started extruding (I don’t think the extra 15° C to hit normal temperature would have done much). The generic ABS appeared to have zero affinity for the surface of the Garolite when dropped from perhaps 2mm above, so I brought the nozzle down far enough that it would extrude right onto the surface. With the filament feeding through the nozzle at a 20% feedrate, I jogged the head around. The ABS actually appeared to stick to the surface, but when I opened the door to see how much adhesion there was, I found that there was hardly any. Note that I’m using the Garolite surface as it came – I wonder if sanding/blasting/grinding the gloss layer would improve adhesion any.

For the next test, I placed a silicone sheet (McMaster-Carr part number 8632K62) on the platform to see if that would have any adhesion. None at all, it turns out – the ABS practically bounces off the sheet (which made a lot of sense in hindsight, as silicone sprays are commonly used for release agents). A pity, as the high heat resistance and flexibility would have made for a great surface.

Finally, I thought I’d try what has become quite popular among RepRap users and even UP! printer users – masking tape (specifically blue 3M painter’s tape). I thought this was an entirely ridiculous idea when I first read about it some time ago, but so many people were using it that it wasn’t likely to be a running joke. I applied a few strips of tape to the platform, let the chamber heat up once more, and gave a small Mendel part a shot. I was amazed at the results – the ABS sticks to the tape beautifully, yet separates cleanly once cooled. I still had significant corner lifting, however.

First layer of a Mendel x-vert-drive-nut-trap_4off

With a decent platform, I thought I’d try my luck once more with running a plate of parts over the Thanksgiving break.  I dialed the extruder temperature down to 245° C and let the machine hammer away.  I stuck around for a layer and a half, and things were looking okay, so I figured I’d check back in 50 hours when the print should be done.  Unfortunately, the print started failing at some point within the next 24 hours, as a coworker who had stopped in at the office noticed that the machine “appeared to be dispensing low-grade dental floss” rather than doing anything productive.  Sure enough, I had another chamber full of ABS vomit.  The post-mortem points towards massive warping and corner lifting as a likely culprit – one of the parts looked like the hot nozzle had rammed into a lifted side, halting further movement of the head (and causing lost steps as a result).  There were also various small tears in the blue tape, so obviously a struggle had ensued.

Annoyingly, this sight is becoming increasingly common.

During all of this, I also thought I’d check to see what sort of die swell I’m getting with the NIP ABS versus the Stratasys ABS.  Quite surprisingly, the NIP ABS only swells to 0.013″ (out of a 0.012″ nozzle), while the Stratasys ABS swells up to a whopping 0.017″.  The significance of this finally dawned on me a week later – no wonder I was seeing drooping filaments on crosshatch infill, and a ‘bunched up’ looking filament on the first layer.  The same volume applied to a smaller diameter filament means that the filament will have to have a longer length!  It wasn’t a matter of too much temperature after all (though the generic ABS remains much more finicky than the OEM ABS).

The only way to make the generic ABS work in any tolerable manner will be to modify parameters in Quickslice to account for the significantly reduced die swell when compared to the OEM ABS.  However, since properties of the OEM material are very tightly controlled, there is no way in the software to adjust such settings – the only thing that can be modified is the ‘road width’, which may not help me much.  Tinkering with the definition file may be the ultimate method of customization. Unfortunately, the file’s format (though sprinkled with some comments) isn’t documented anywhere that I’ve found.  This definition file is for a specific machine (FDM 1600), running a specific material (P400 ABS), with a specific nozzle (T12, which has a 0.012″ orifice), at a specific slice height (0.010″).  In all, Quickslice 6.4 has has 24 different definition files just for the FDM 1600:

  • ICW6 material (an investment casting wax) with a T16 nozzle at slice heights of 0.007″, 0.010″ and 0.014″
  • ICW6 material with a T25 nozzle at slice heights of 0.010″ and 0.014″
  • ICW6R support material (for the ICW6 build material) with a T16 nozzle at slice heights of 0.007″, 0.010″ and 0.014″
  • ICW6R support material with a T25 nozzle at slice heights of 0.010″ and 0.014″
  • P301 material (a Nylon formulation) with a T12 nozzle at a slice height of 0.010″
  • P301 material with a T25 nozzle at slice heights of 0.010″ and 0.014″
  • P301R support material (for the P301 build material) with a T12 nozzle at a slice height of 0.010″
  • P301R support material with a T25 nozzle at slice heights of 0.010″ and 0.014″
  • P400 material (ABS) with a T12 nozzle at slice heights of 0.007″ and 0.010″
  • P400 material with a T25 nozzle at slice heights of 0.010″ and 0.014″
  • P400R support material (for the P400 build material) with a T12 nozzle at slice heights of 0.007″ and 0.010″
  • P400R support material with a T25 nozzle at slice heights of 0.010″ and 0.014″

The largest portion of the definition file is the flow curves, which is a table comprising a list of entries as follows:

# PDMM START ————————————————————–#

#—— CURVES ————————————————————#
# Flow  D   Pre   Pre  Start Start Shut Roll
# Area  O   Delay Flow Flow  Dis.  Off  Back Speed
# xxxx  xxx .xxx  xxx  xxx   xxx   xxx  xxx  xxxxx
PDMM
30     2  .026   79    7    60    41   143   800
31     4  .028   79    9    60    42   143   800
32     6  .030   79    9    60    43   145   800
33     8  .032   79    9    60    44   145   800
34    10  .034   79   11    60    45   147   800
35    12  .036   79   11    60    46   147   800
36    14  .038   79   11    60    47   149   800
37    16  .040   79   11    60    48   149   800
38    18  .042   79   11    60    49   149   800
40    20  .044   79   11    60    50   149   800
41    22  .046   79   13    60    51   149   800
42    24  .048   79   13    60    51   149   800
44    26  .050   79   13    60    51   151   800
45    28  .052   79   13    60    52   151   800
46    30  .054   79   13    60    52   151   800
48    32  .056   79   15    60    52   153   800
49    34  .058   79   15    60    53   153   800
51    36  .060   79   15    60    53   155   800
53    38  .062   79   15    60    53   155   800
54    40  .064   79   15    60    54   155   800
56    42  .066   79   15    60    54   157   800
58    44  .068   79   15    60    54   157   800
60    46  .070   79   15    60    55   157   800
62    48  .072   79   17    60    55   159   800
64    50  .072   79   17    60    56   159   800
66    52  .073   79   17    60    57   159   800
68    54  .074   79   17    60    57   159   800
70    56  .075   79   17    60    58   161   800
72    58  .076   79   17    60    58   161   800
74    60  .077   79   17    60    58   161   800
77    62  .078   79   17    60    59   161   800
79    64  .079   79   17    60    59   163   800
82    66  .080   79   17    60    60   163   800
85    68  .081   79   17    60    60   163   800
87    70  .082   79   17    60    60   163   800
90    72  .083   79   17    60    61   163   800
93    74  .084   79   17    60    61   165   800
96    76  .086   79   17    60    62   165   800
99    78  .087   79   17    60    62   165   800
102    80  .088   79   17    60    62   165   800
106    82  .089   79   19    60    63   165   800
109    84  .091   79   19    60    63   167   800
113    86  .092   79   19    60    64   167   800
116    88  .093   79   19    60    64   167   800
120    90  .094   79   19    60    64   167   800
124    92  .096   79   19    60    64   169   800
128    94  .097   79   19    60    64   169   800
132    96  .098   79   19    60    64   169   800
136    98  .100   79   21    60    65   169   800
141   100  .101   79   21    60    65   171   800
146   102  .102   79   21    60    66   171   800
150   104  .104   79   21    60    67   171   800
155   106  .106   79   21    60    67   171   800
160   108  .107   79   21    60    67   171   800
166   110  .109   79   23    60    68   173   800
171   112  .111   79   23    60    68   173   800
177   114  .112   79   23    60    69   173   800
182   116  .114   79   23    60    69   175   800
188   118  .115   79   25    60    70   175   800
195   120  .117   79   25    60    70   175   800
201   122  .118   79   25    60    70   175   800
208   124  .120   79   25    60    71   175   800
214   126  .122   79   27    60    71   177   800
221   128  .124   79   27    60    72   177   800
229   130  .125   79   27    60    72   177   800
236   132  .127   79   29    60    73   179   800
244   134  .129   79   29    60    73   179   800
252   136  .130   79   29    60    74   179   800
260   138  .132   79   29    60    74   179   800
269   140  .134   79   31    60    75   179   800
277   142  .136   79   31    60    75   181   800
287   144  .138   79   31    60    76   181   800
296   146  .140   79   33    60    77   181   800
306   148  .142   79   33    60    78   181   800
316   150  .144   79   33    60    78   183   800
326   152  .146   79   35    60    79   183   800
337   154  .148   79   35    60    80   183   800
348   156  .150   79   35    60    81   185   800
359   158  .152   79   35    60    81   185   800
371   160  .154   79   37    60    82   185   800
383   162  .156   79   37    60    83   185   800
396   164  .158   79   39    60    84   187   800
409   166  .160   79   39    60    85   187   800
422   168  .162   79   41    60    86   187   800
436   170  .165   79   41    60    87   189   800
450   172  .167   79   41    60    88   189   800
465   174  .169   79   43    60    89   189   800
481   176  .171   79   43    60    90   189   800
496   178  .174   79   45    60    92   191   800
513   180  .176   79   45    60    93   191   800
530   182  .178   79   47    60    94   191   800
547   184  .181   79   47    60    96   193   800
565   186  .184   79   47    60    96   193   800
584   188  .187   79   49    60    97   193   800
603   190  .190   79   49    60    97   195   800
623   192  .192   79   49    60    98   195   800
643   194  .194   79   51    60    98   195   800
664   196  .196   79   51    60    99   197   800
686   198  .199   79   51    60    99   197   800
709   200  .201   79   53    60   100   197   800
732   202  .203   79   53    60   100   199   800
756   204  .205   79   55    60   102   199   800
781   206  .207   79   57    60   104   199   800
807   208  .210   79   59    60   105   201   800
833   210  .213   79   61    60   107   201   800
861   212  .216   79   61    60   108   201   800
889   214  .219   79   63    60   109   201   800
918   216  .222   79   63    60   110   203   800
949   218  .225   79   65    60   111   203   800
980   220  .228   79   67    60   112   205   800
1012  222  .231   79   67    60   113   205   800
1046  224  .234   79   69    60   114   205   800
1080  226  .237   79   69    60   115   207   800
1116  228  .240   79   71    60   116   207   800
1153  230  .242   79   71    60   117   207   800
1191  232  .245   79   73    60   119   209   800
1230  234  .248   79   75    60   120   209   800
1270  236  .251   79   75    60   121   209   800
1312  238  .254   79   75    60   122   211   800
1356  240  .257   79   77    60   123   211   800
1400  242  .260   79   77    60   124   211   800
1446  244  .263   79   79    60   125   213   800
1494  246  .266   79   79    60   126   213   800
1543  248  .269   79   81    60   127   213   800
1594  250  .272   79   81    60   129   215   800
1647  252  .275   79   83    60   130   215   800
1701  254  .278   79   83    60   131   215   800
END PDMM

# PDMM END —————————————————————-#

After poking at the definition file with a text editor for a very long while, I realized two things:

  1. I have spent waaay too much time on researching Stratasys machines, styrenic polymers, and all points of intersection.
  2. The second column of the flow curves chart is comprised entirely of binary values from 2 to 254.

It seemed a reasonable guess that ‘DO’ means ‘digital output’, and looking through generated .SML files, I saw various PD and MM commands (hence the ‘PDMM’ block of data).  The significance of the table only having even values from 2-254 hit me when I saw that bit 0 of the Asymtek’s 8-bit digital output is toggled when switching between the model and support nozzles – bits 1-7 must then be dedicated to controlling the motor speed.  This made even more sense when I consulted the Asymtek ACL programming reference and found that PD allows for a Pre-Delay between the digital outputs being set and the start of motion (allowing the filament to start extruding before the head starts moving) and MM allows for the digital outputs to be changed while the system is in the Middle of a Move (so that the feed rollers can be turned off just before the head comes to the end of extruding a path, allowing the ‘post-flow’ to extrude the tail end of the plastic).

The best way to see if modifying the definition file would have an effect was to try generating a test .SML file.  I created a .STL file consisting of a block measuring 0.25″x0.25″x0.1″ and generated a test SML file with the default settings and no supports.  Here’s a snippet from the file, where the machine lays down the outline for the first layer (note the MA lines – these are Move Absolute commands to X,Y locations, and you’ll see that the moves do indeed make a square).

# Z = 0.0090 S = 0.00900 T = 00:00:00 ### BEGIN FIRST SLICE ###
MA342,342;
XD209;VS100,1;#FC IDX1
MZ-52;
# obj:0 set:Part type:Perim matl:main width:0.0200 Z:0.0090 S:0.0090 (skipfill)
SR800;
PD.115,79;MM;MM0,25;MM60,118;MM-70,175;
AS1;VM4;BC;
MA342,342;
MA111,342;
MA111,111;
MA342,111;
MA342,342;
EC;VM3;MA241,342;# Exit

I knew that the SR800 was a speed setting (Step Rate) – the last column in the flow curve table.  Hmmm, I wonder if any of those values in the next line match up with a line of table values…

188   118  .115   79   25    60    70   175   800

Yep, that line looks like a dead ringer.  What happens if we change the 800 speed on that line to 801 in the machine definition and generate a new .SML file?

# Z = 0.0090 S = 0.00900 T = 00:00:00  ### BEGIN FIRST SLICE ###
MA342,342;
XD209;VS100,1;#FC IDX1
MZ-52;
# obj:0 set:Part type:Perim matl:main width:0.0201 Z:0.0090 S:0.0090 (skipfill)
SR801;
PD.115,79;MM;MM0,25;MM60,118;MM-70,175;
AS1;VM4;BC;
MA342,342;
MA111,342;
MA111,111;
MA342,111;
MA342,342;
EC;VM3;MA241,342;# Exit

Eureka!  Unsure of what I should actually try next (other than pestering Rick @ MakerGear to hurry up with getting the Experimental Filament Club underway), I thought trying to run parts using much smaller road widths (and not yet actually modifying the definition file) might be a good place to start.  I noticed that using 0.007″ slice heights in Quickslice defaults to using very narrow road widths (0.0137″), so I thought I’d give it a try.  In theory, making up for the reduced die swell by means of a smaller road width should result in not having drooping filaments on crosshatch infill…

Photo adjusted for improved contrast

And indeed, it worked like a champ, even with an extruder temperature of 270° C. and a 70° C chamber.  The crosshatch infill (though still not quite as good as with OEM material, but the best yet with generic material) showed minimal distortion – had I let the part complete (would have taken 2 hours – the FDM 1600 sets no speed records), I’m sure I would have not seen any sign of sagging on the top surface.  I still had significant curling on the part (and the curling that started on the left end of the part appeared to creep along as the build progressed), so the search for better ABS continues.

Another CNC machine?

My Taig CNC mill has served me well for many years, and continues to perform admirably. Perhaps too well, as I always have it fixtured up for one thing or other, and as such I no longer have a CNC machine that I can just hack/play/tinker on.  The foundry class got me thinking about CNC milling foam cores, or perhaps patterns and matchplates out of wood or plastic.  This led me to think that perhaps I should build a CNC router for such work.  I say build rather than buy – gantry type routers are quite simple in construction and building a machine is half the fun anyhow.  There’s many free designs available for CNC routers, so I went looking for one that I liked.  I knew I wanted something based on T-slot extrusion for ease of assembly and straightness (some warping would be almost inevitable if I were to try welding a frame out of square tubing).  After a brief search, I found a promising looking design done by cncrouterparts.com and Fine Line Automation. A 24″ x 36″ working area sounded big and roomy, and I was most impressed by the use of cold rolled steel and skate bearings for linear motion (in a similar fashion to a RepRap) rather than much more expensive linear bearings. The last time I looked at building my own CNC, I realized that the least expensive route would be to buy surplus linear bearings and rails from Ebay in whatever sizes I could manage, and then design the machine around those components. But in this case, I could follow the plans more-or-less as published – a good thing, as once I start redesigning something, I never really stop the design process, and whatever I was working on winds up with a severe case of kitchen sink syndrome.

While Fine Line Automation has kits available, I knew it would be much less expensive for me to simply get the raw materials myself and do the requisite cutting on my own. I did look around for cheaper alternatives to the 80/20 T-slot extrusion that the plans called for (many companies offer ‘aluminum structural framing’), but eventually found that 80/20’s garage sale Ebay store had the best prices, and they actually had all the components that the bill of material called out.

Rough bandsawed extrusion awaiting cleanup on the mill

I chopped the pieces to rough length on the big horizontal bandsaw at work, and then took them home to machine the ends to the final precise lengths.  After tapping the holes in the ends to 5/16″-18 and drilling access holes in strategic locations (all detailed in the prints contained in the set of plans available on Fine Line’s site), the extrusion pieces were complete. A trip to Speedy Metals netted me the cold rolled steel, which I also drilled out on the mill. With stepper motors on the way from Keling, bearings from VXB, ballnuts and ballscrews from McMaster-Carr and couplers from Enco, it seemed like a good time to at least start assembling the base.

Definitely larger than I had envisioned - I'll need to build a special cart for it

Many of the builds that I’ve seen of this design include additional bracing in corners.  The framing itself is quite stout and I don’t intend to do any metal cutting with the machine, but more rigidity never hurts on a machine tool, so I used some aluminum angle on the inside corners.  I ran out of screws, so the base isn’t yet quite fully assembled, but at least I have a sign of progress.

Categories: CNC

Smells like foundry

I had been eager to pick up a class again this semester at MATC, and when browsing through the classes this past summer, I figured I’d give the foundry class a try.  We did a little bit of foundry work way back in high school shop class (I’m sure those days are long gone thanks to a lawsuit-happy society), and I’ve seen a number of Rick Chownyk’s backyard metal casting demos during past CNC Workshops.

As it turned out, the class isn’t so much a class as it is an open workshop – most of the students have been taking the class for years (two of them started taking it 28 years ago) as a way to easily make parts for their own projects (one of the fellows is a live steam locomotive enthusiast, and always has something interesting that he’s molding).  Just the sort of environment I was eventually hoping to find!  I quickly found that as with so many opportunities, I really didn’t have a clue as to what I wanted to make – I had signed up with the intent to learn, not to do. Sure, there was a bit of learning, but weeks later, I’ve only been able to figure out a single thing that I’d actually like to cast (a fixture block for machining an upcoming project). As such, every class has started with me pawing through the cabinet of patterns, wondering what to try this week. Not that this hasn’t been helpful – I managed to screw up 2 weeks in a row by focusing so much on forming a good parting line around a complex pattern that I forgot to actually remove the pattern from the mold before the aluminum was poured in. Fortunately, the temperature difference is enough between the aluminum pattern and the molten incoming aluminum that the two didn’t fuse (my tendency to create very narrow runners also helped in this regard).  Here’s a brief photo collection of some of my successes and failures.

For those of us new to the class, we started out with flat patterns - items flat on one side that were easy to make molds of. I grabbed a sunflower and horsehead from the cabinet of patterns. Rubbing the patterns down with graphite helps them to release nicely from the mold.
After ramming the sand in the cope and pulling it off the base, the cope is then flipped upside-down. The other half of the flask is placed on top and parting compound is dusted over the surface so that the mold halves will separate cleanly.
The drag is then rammed on top of the cope. After this, I used a brass tube to cut gates (holes for the incoming metal to flow down), a spoon to cut runners (channels that connect the gates to the mold cavity itself), and a piece of coathanger wire to poke vents (to allow air to escape ahead of the incoming metal) in the drag (which is the top half of the mold assembly).
This is the actual crucible of molten aluminum in the induction furnace. Due to the IR sensitivity of the camera, it really doesn't look this pink to the human eye. When the crucible is full of aluminum and the furnace is on high power, there is a very pronounced meniscus of molten metal writhing around due to the induction - it's actually quite reminiscent of the T-1000 from Terminator 2.
The mold, freshly poured. The casting sand is bonded together with oil, so flames are quite often seen lazily burning away as the mold cools. Steel plates or other heavy objects are sometimes placed on top of a mold so that the top half of the mold doesn't actually float up on the molten metal, ruining the cast.
Once the mold halves are pulled apart I get to examine my handiwork.
The horse head came out great, but the sunflower didn't fare so well. Since I tried putting the gate (the hole in the sand through which the molten metal enters) right behind the part itself, I had pushed the sand out around the gate just enough to keep the cavity from filling properly.
A few classes later, I thought I'd try doing a matchplate mold. A matchplate is simply an insert plate between the cope and drag halves of the mold that consists of the both the part cavity as well as the runners to the parts. You can ram both sides of the mold without having to remove the pattern, as the matchplate makes its own parting line. When making a matchplate mold, you only have to worry about cutting the gates and risers, so moldmaking can be done much more quickly.
I had been wanting to try doing a lost foam casting, and I realized that the fixture block I needed for an upcoming project would be a great item to make with this technique (I wasn't finding any scrap pieces of aluminum large enough at work, anyway). I hacked a piece of pink styrofoam insulation into the rough shape, superglued a thin piece of foam onto the end to act as a sprue, and then went to work covering it in a mixture of thinned drywall compound and playground sand.
After sticking the foam core into a bucket filled with sand (with the sprue sticking out the top), I placed an empty can over the end of the sprue for the metal pour.
After letting the metal cool for about 10 minutes, I started pouring off excess sand.
Once fully cooled with the drywall compund chipped away, it didn't look too bad. Not great (lots of sand inclusions), but it should be just fine for fixturing.
Hitting all the sides on a belt sander removed some of the harder and more abrasive deposits prior to near-shape milling.
I used a crummy cheap endmill to clean up all 6 faces of the block, not wanting to subject a good endmill to the abrasiveness of sand inclusions. I'll fully true it up once I get my fly cutter in. The black line is where I'll make a bandsaw cut later on - the fixture block will wind up just being a wedge, but I wanted to leave a 'tail' on it so that I could hold it in the vise more easily. The part still has a gritty feel at this state, so maybe I'll need to remove a bit more of the surface in order to hit better quality metal. When milling the faces, I noticed that the metal wasn't forming into normal chips, but was crumbling (much like machining cast iron). A low quality casting to be sure, but for a first attempt at lost foam, I'll run with it.

Not all HIPS is created equal, either

I managed to get a decent pin vise at last and drilled out the support nozzle with a 0.011″ drill.  The other day I got around to testing the cleared nozzle to determine if buildup/blockage might be the root cause of my difficulties in running the high-impact polystyrene from New Image Plastics.  I powered up the Stratasys, let the support nozzle come up to temperature, and started feeding it the NIP HIPS.  I noted that the extruded plastic was still curling right around and contacting the exterior of the nozzle – re-boring a 0.012″ nozzle with a 0.011″ drill bit apparently does zilch in this regard (makes sense, as I wasn’t actually removing any metal), but at least I could be fairly sure that any possible blockage was gone.

After perhaps 2 inches of feedstock, the HIPS filament buckled once again between the feed rollers and the liquifier entrance.  I struggled with this for some time, and found that increasing the support temperature to 275° C (10 degrees higher than the normal setpoint) and reducing the feed to 60% (normally at 100%) improved things greatly.  Improved, but not fixed – I could still expect the HIPS to kink very roughly once in every 18 inches or so.  Additionally, I noted very faint wisps of smoke emerging from the nozzle area – I’m pretty sure I was witnessing outgassing of the HIPS as it emerged hot out of the nozzle.  I had never seen this with the OEM material.

I had a little bit of OEM support material left and figured I should try running it to see if it would also buckle – perhaps I still had some sort of buildup inside the liquifier, and maybe the OEM material would act the same way.  I dialed the temperature back down to the OEM setting and bumped the feed back up to 100%.  It ran just as happily as it had done prior to my mad experimentation with alternate materials.  With that in mind, I started to do some light research on what exactly high-impact polystyrene consists of.  Styrene, as Wikipedia points out, is an oily liquid – many styrene molecules need to be linked together in order to create polystyrene, which is the clear plastic used for car windshields and airplane canopies in plastic model kits (modelers hardly ever say ‘polystyrene’, referring to such material as simply ‘styrene’) .  Polystyrene (sometimes abbreviated as GPPS, for General Purpose Polystrene) is itself rather brittle, so to toughen it up, a bit of polybutadiene rubber can be added to improve the impact resistance and we get High Impact Polystyrene (HIPS).  The polybutadiene ruins the transparency of the polystyrene, so uncolored HIPS is a milky white color.  Perhaps this particular HIPS had enough polybutadiene in it that made it buckle more easily than the OEM material?  A brief side note – there are apparently a LOT of ways to mix styrene and butadiene.  I discovered SBS (styrene-butadiene-styrene), which Chris DeArmitt did an excellent job of contrasting with HIPS.  There is also SBR rubber which shares the same CAS # (9003-55-8) as HIPS, even though the two materials seem quite dissimilar (I am not a chemist, so this may all make perfect sense if one is versed in the ways of polymer science). And if that isn’t confusing enough, have a dose of CAS # 91261-65-3, styrene-butadiene block copolymer. Yeah, my eyes glazed over long ago as well, and now my brain hurts.

In looking through various MSDS sheets for different grades of HIPS, I noticed that many formulations contain a few percent of mineral oil.  I wondered if perhaps vaporization of the oil might account for the faint smoke I was seeing – the 265° C nozzle is hot enough to boil some grades of mineral oil.  Once again, NIP was kind enough to provide me with material specs for their HIPS, which turns out to be Styron 487.  The MSDS sheet indicates that less than 5% of the composition is not in fact mineral oil, but is a ‘copolymer mixture’ marked as being a trade secret. Furthermore, the sheet only cautions against temperatures over 300° C as causing decomposition.

I went back to playing with running filament through the Stratasys, remembering that the machine has a torque monitor setting.  If the filament jams in the extruder, the machine will detect that the drive rollers are pushing too hard and pause the system.  I’m not sure that I’ve ever actually seen this torque limit tripped, but when feeding filament through, you can hit the torque button for a realtime display of how hard the feed motor is working.  Comparing the torque when feeding the OEM support material versus the HIPS should tell me…  well, something.  Setting the support extruder for 265° C and 100% feed, I ran the OEM support material through and the torque was around 65 (I’m pretty sure this is a unitless number) – interestingly, I couldn’t see any sign of smoke or vapor with the OEM material.  I then ran the NIP HIPS through with the same settings (as expected, faint wisps of smoke/vapor were visible once more), and it was a whopping…  What?  Only around 60?  What the heck is going on?  I noticed that both materials were extruding much more nicely this time – they would still curve a bit after exiting the nozzle, but not as much as before.  I ran about 6 feet of HIPS through without a hitch before getting tired of waiting for it to kink, which it never did.  As best I can guess, there must have been a little bit of buildup or blockage that the OEM material flushed out.  Could there be some magic cleaning additive in the OEM support material?  Possible, but unlikely – the P400 support MSDS sheet notes the only ingredient as being approximately 99% styrene-butadiene copolymer.  Eh, maybe I was just lucky.

As long as I was looking at torque values, I thought I’d see how the OEM ABS compares to the NIP ABS.  At 275° C and 100% feed, the OEM ABS showed a torque of about 70.  The NIP ABS at the same settings was much higher – about 110.  I also noted that again, faint smoke/vapor was present with the NIP material, but not with the OEM. The MSDS sheet for the Polylac PA-747 that is the NIP ABS notes that additives comprise less than 2% of the material, but it doesn’t say what those additives might be. One possibility is that NIP may be adding some sort of chemical to their batches in order to improve processing, but this is only a wild guess – I don’t know much about the extrusion process and would have to ask Jim @ NIP if they add any sort of special sauce.

More Stratasys disassembly

My use of generic ABS and HIPS in the Stratasys finally has caused some annoying issues (more than just drooping filaments on infill).  The photo of what awaited me after after the first build weekend with generic materials should speak for itself.  It looks like the machine literally barfed all over.

I wondered if perhaps the support filament was to blame, as it had jumped off of the spool and wrapped around the spool shaft in testing a few times (due to me winding the HIPS onto the spool all the way to the rim).  I looked in the dry box, but all was well.  I then opened up the head for a look inside, and it looked like an accident at a spaghetti factory.  HIPS was stuffed all around, and required removal of the motor blocks, so I figured this was a good time to photograph details of the head (the only peek inside a Stratasys head that I’ve found online is at Bouke’s blog) for others to see.  Plus, a good cleaning was in order – the generic ABS and HIPS appear to have a great deal of volatile compounds, and there was a good amount of soot and burnt plastic sticking to the nozzles (plus, there was enough melted plastic gunk between the two nozzle rings that the support nozzle wasn’t able to travel up and down freely).  This is a big concern when looking for what materials to run through the Stratasys – I had no issues with the OEM material (even with the material left cooking for half a day), but the generics from NIP like to stick to the nozzles and char (note all the brown lumps in the photo).

After cleanup and re-assembly the ABS nozzle is still extruding very nicely, but the HIPS nozzle may very well have some buildup inside.  I could only extrude perhaps 10 inches or so of HIPS before the filament would buckle between the feed rollers and liquifier entrance, which is what caused the impressive birdsnest in the head in the first place.  I haven’t tried extruding more OEM support material through however, so maybe I’m just hitting a limitation of the material itself.  Still, I think the support nozzle has some sort of buildup, as the material is not coming out straight – as soon as it exits the nozzle, it curls back around and sticks to the nozzle, after which some sort of mess is inevitable on a long build.  Nophead notes in his latest blog entry that he needed to clear out one of his nozzles with a drill bit to restore proper extrusion. I ordered a 0.011″ drill bit from McMaster-Carr, and then found that the pin vise I have can’t properly grip such a tiny bit, so I await a better pin vise before I can see if this re-boring fixes the issue.

On to the photos!

After the head is removed, this is what you see.  There’s three vacuum cleaner hoses that run to the rear of the block – the large center one carries cooling air (which gets expelled from the central nozzle) as well as the two filaments (which poke out on either size of the nozzle through adjustable grommets).  The two smaller hoses at the top are for the air return.  The brass brush at the lower right helps keep the nozzles clean – after every 2 build layers, the head zig-zags the nozzles over the brush to wipe off excess filament that has oozed out.  Of course, this only really works with the OEM filament, as it doesn’t adhere and melt onto the hot nozzle surface like the generic ABS and HIPS likes to do.

This is a top view of the head – there’s a locating pin at the center front and rear to make sure that it is perfectly aligned when the latches are secured.  The front cover protects the solenoid.

This is what the rear of the head looks like, with one of the motor blocks removed.  Each filament passes first through the black plastic guide bushing, then through the pinch wheels, then finally into the entrance of the liquifier.  I’m not sure what type of plastic the reddish-brown liquifier end caps are, but it’s obviously a high temperature material.  Right between the two liquifier entrances is the point at which the cool air is directed – it’s important to keep the filament solid before getting to the liquifier.

Here’s a close look at the motor block itself.  I’m guessing the amber colored insulator plate is the same plastic used on the top of the head.  The toothed roller (appears to be black anodized aluminum) is the driven one.  The MicroMo gearmotor label reads as follows:

1624T012S123 X0520

16/7 415:1K703

HEM1624T16 KW 45/96

It turns out that these 3 lines specify the motor, gearhead and encoder.

This is what the underside looks like after the nozzles and protective rings are removed.  Note that the black ring on the nozzle is actually a seal (though I’m not sure what material – Kalrez or other perfluoroelastomer, perhaps).  Yuck, look at all that black crud.  I tried cleaning the rings and nozzles by soaking them in acetone, but it really didn’t help much.  I’m assuming that the relatively volatile styrene (which acetone dissolves) had been cooked out already (again, why the OEM material doesn’t degrade in this manner is still a mystery).  The rings look to protect the bottom of the foil and insulation wrap.  The picture really doesn’t show it, but the heating elements come all the way down to where the foil ends.

Here we have the real guts of the head.  Each liquifier (build material on the left, support material on the right) has a thinwall stainless tube at its core (according to Stratasys patents, anyhow – I’m not about to start unwrapping insulation to find out).  It looks like there must be some sort of other material around this core, over which the heaters are spiral wrapped and then covered with a layer of what appears to be fiberglass and foil.  The cylindrical spring-ended parts that flank the liquifiers are the RTDs that actually measure the temperature. The cylindrical caps at the top of each liquifier are Klixon thermal circuit breakers. If the temperature controllers are improperly set (easy to do, and the manual warns that seeing ‘100’ with the ‘M’ LED on does not mean 100 degrees, but rather 100% output), the circuit breakers should keep the liquifiers from overheating. I don’t know what this cutoff temperature is, however – the circuit breakers don’t appear to have any markings other than ‘Klixon’.  Note the rear of the right liquifier, specifically the two aluminum blocks on either side.  These are actually pivot blocks – there’s a pin on either side of the support liquifier to allow the whole unit to tilt downwards by perhaps 1/16″ or so.  You can just make out the extension spring (and pin that the spring attaches to) at the front of the liquifier that keeps it in the ‘up’ position when not in use.  And what moves it downward, you ask?

A Lisk push-type solenoid pushes down on a paddle connected to the bottom of the support liquifier.  Note the hex nut at the bottom – this locks in place the set screw that adjusts the downward travel of the paddle.  Upward travel is adjusted via a screw attached to the cover that normally protects the solenoid (the solenoid core actually contacts the screw at the top of the solenoid’s travel).

In short, the head is pretty straightforward in terms of design and construction.  Would I want to scratchbuild one myself should this one become irreparably damaged?  Heck no, but it would certainly be possible.

More fun and games with plastic filament

After my previous post on alternative ABS use on the Stratasys, I wound up with a pile of informative comments. Jeff directed me towards US patent application 2009/0295032, which contained this gem:

The extrusion runs of Examples 1-12 were performed with a modified ABS material commercially available under the trade designation “CYCOLAC” MG94-NA1000 ABS from General Electrics Co., Pittsfield, Mass. The extrusion runs of Comparative examples A-D were performed with a standard ABS copolymer commercially available under the trade designation “AG700 ABS” from The Dow Chemical Company, Midland, Mich.

We surmised that Dow AG700 ABS may likely be the standard Stratasys ABS, and given the date of filing and material properties, the GE MG94-1000 ABS was likely the newer ABS-M30 material.  I started looking into the AG700 resin first – while I was able to find something of a datasheet for the product, the fact that Dow now only sells to the automotive market (in North America, at any rate) pretty much quashes any possibility of acquiring the resin. Still, we now have a set of material properties to use for comparison, and I’m wondering which properties specifically make for an ideal Stratasys/RepRap/FDM feedstock. The “high flow” and “low gloss” aspects jump out at me in particular – obviously high flow is needed given the small orifice sizes, but I wonder if the glossiness of the extruded filament is indicative of its surface energy, and hence the amount of attraction that two layers will have for each other (thus potentially causing warp).

I had much more luck with the MG94-1000NA resin – GE sold their plastics division in 2007 to SABIC, so I dug around on their site (after needing to create an account, grrr) and pulled up the datasheet for the material. It turns out that the ‘-1000NA’ is simply the color code – they have dozens of colors available, and -1000NA is the plain old uncolored ‘NA’tural one. After a call to sales, I found that the smallest quantity I could order was 55lbs. at a whopping $30.61 per pound. Yowza. However, the price decreases drastically with quantity, and 330lbs. would only be $7.69 per pound. I don’t exactly have $2500 burning a hole in my pocket, but the quantity and pricing certainly isn’t out of reach for a few dedicated hobbyists to try. There’s a few other possibilities to research before falling back to that, however.

Recently there’s been a bit of buzz on the RepRap forums about a low-cost, very RepRap-ish 3D printer from China. What caught my eye the most was that this printer is actually using Stratasys sized 0.070″ feedstock rather than the 3mm RepRap standard. They note an ABS price of $50/kg, which comes out to $22.69/lb. Not as cheap as from NIP or other sources, but if the plastic acts just like Stratasys ABS, it would be worth it.

Speaking of NIP, I called up Donna to see what other possibilities there were for ABS. She and Jim were kind enough to provide me with the datasheet for the specific ABS they use, which is Chi Mei Polylac PA-747. They’re able to get other Chi Mei resins as well, so I took a look at their high flow offerings. The highest flow formulation, PA-756H, looks promising – the low impact strength is one of the properties that stuck out on the AG700 resin. I’ve emailed Donna to see if they can acquire this material.

So much for material musings – I have 5 pounds of HIPS that I need to wind onto spools! I’m not about to wind a half mile of filament all by hand again, so I needed to figure out a good automated system. I originally figured I’d just chuck the empty spool on the lathe and wind it that way, but then I decided to use the mill instead, as the head is variable speed on-the-fly. How to hold the empty spool, though? An expanding collet would be great, but I don’t have any as big as the 2″ bore in the Stratasys spools. A little bit of digging through scrap bins, and I found a Delrin cylinder that would work perfectly for making a pair of bullnose centers mounted on a bolt. Here’s what the assembly with spool looks like on the mill:

I clamped a live center into the mill vise to support the spindle from underneath

What to do on the unwind side was a bit trickier.  Based on my winding by hand of the ABS, I knew that keeping the coil of filament in a rather static shape was very important, so I opted to make a simple spool that could be assembled around the coil itself.  I used Masonite for the sides and drilled holes  through each piece to allow for joining screws.  Multiple sets of holes were drilled so that I could adjust the screw locations to perfectly fill out the center of the coil.

The coil of HIPS filament as it arrived from NIP, still bound with shrink-wrap, is placed onto one of the spool sides. The screws that will attach the other side are placed as close to the interior diameter of the coil as possible. 2" pieces of pneumatic tubing are slipped over each screw to serve as spacers.
The fully assembled spool, mounted and ready to unwind. The radial lines on the top piece of Masonite are a result of using an angle grinder to deburr the holes. It may sound crazy, but it worked far better than the countersink I tried first.

I clamped a piece of 3/4″ rod in the bench vise and slipped the spool over it (a collar on the rod kept the spool at an appropriate height).  It rotated pretty freely, so I took the loose end of filament and tied it to the inside of the empty spool.  I set the mill spindle to the lowest speed, and hit the power.  60 rpm is perfectly fast for winding, though I cranked the speed up to around twice that once I had a good start on the winding.  I held the filament in one hand (with a rag so that I’d actually have skin left) to provide continuous tension, and the material unwound from my adjustable spool just as nicely as I had hoped, with no snarls.  About a half hour later I had two spools fully refilled.

This beats the heck out of doing the winding by hand.

I gave the NIP HIPS a try in the Stratasys, noting that the Stratasys support material is slightly more brittle than the NIP HIPS.  I ran a small test part with NIP ABS at 250° C, and the HIPS at 265° C.  The HIPS performed admirably, and had good adhesion to the NIP ABS.

The part separated from the base cleanly. I'm at a loss as to why the top crosshatch fill turned out nice and tight. The only reason I can think of is that perhaps the drive rollers were beginning to slip, resulting in a diminished flow rate.

The Stratasys is currently running a large plate of parts, and we’ll see if I have less warping this time.

A splash of color, part 2

After my first successes with anodizing I became a bit more adventurous with techniques. The results weren’t as good, but it was still fun to try. Here’s the rundown of samples.

1) I recall reading somewhere that hot glue worked for a masking agent.  I dyed the piece turquoise, then threw on some stripes of glue from a hot glue gun.  I then dunked the piece into the pickle bath (where the glue actually seemed to do okay as a mask), and then the piece went into the sealing bath (just boiling water).  That may have been my failing – it probably would have been better to try and remove the glue first.  After boiling, the glue’s grip on the aluminum became even more tenacious, and I really didn’t have a chance of removing it.  You can see all the remnants that remain as the whitish areas.

2) I wanted to try doing a droplet of pickling solution onto a piece, hoping for a splash pattern like Edgerton’s legendary high speed photos.  You can barely make out some of the radial splash pattern, but it’s not nearly as marked as I had hoped.

3) I believe I had come across mention on Caswell’s forum about the use of a baby medicine syringe to squirt/spray on rubber cement for a nice chaotic looking splash pattern.  This sounded like a great idea, so I picked up a baby medicine syringe at Toys ‘R’ Us, and gave it a shot.  Try as I might, I simply could not get the cement to ‘spray’ out of the nozzle, no matter how I attempted it.  What happened was a giant loogie of rubber cement blobbed onto the  orange dyed piece.  I shrugged, pickled the piece and threw it into the electric blue dye.  I’m guessing you could probably build a sprayer to get the effect I was trying for – maybe creating a big external mix airbrush that feeds rubber cement rather than paint?

4) An attempt at doing an ‘acid wash’ finish.  After dying, I used a sponge to dab on the pickle solution, then washed it off.

5) Another attempt at the same technique.  You can see the rectangular impressions of the sponge block.  The acid wash technique apparently takes a good deal of practice to master!

6) For this piece, I tried ‘spraying’ on the pickle solution onto an olive drab dyed piece by running my thumb over the head of a toothbrush loaded with the solution.  I let it sit for a while, then dyed it in bronze.  The result is a barely noticeable color combination.

7) I had wanted to try a toner transfer technique I had read about (use a laser printer or copier to print to a piece of transparency, then use an iron to transfer to toner to the piece as a mask).  Frankie had an even better solution, which was to use Press-n-Peel Blue, which is used for etching PCB boards.  We tried this little cupcake guy on a piece of flourescent yellow dyed aluminum, then put in the pickle and dyed bordeaux red.  After wiping off the mask with acetone, the results looked pretty good.

8) I had also read that Sharpie magic markers worked for anodizing, as you could simply draw onto the surface and then seal the anno.  This worked like a champ, although you can’t really scribble over areas to fill them (you can see where strokes overlapped, just like on paper, though the blue marker didn’t really exhibit this).  As an aside, many many years ago in Action Pursuit Games, there was an article on some of Carter Machine’s custom paintguns. One that I recall was called the “Lil’ Pounder” – it was a standard Nelson style gun that had been so heavily machined that it weighed a whopping 16 ounces as a result. The caption had noted that “anodizing pens” has been used to accent some of the milling cuts, and I spent years looking for these mythical “anodizing pens”. In the end, I’m betting that Earon Carter used nothing more than a red Sharpie!

9) This was done simply by using pieces of electrical tape to mask off areas.  With custom vinyl stickers, you can create fairly complex hard-edged patterns.

10) I wanted to try something a little different with this one – specifically, I wanted to have an effect that looked like a sponge pattern had been brushed on.  I dyed the piece in fluorescent yellow, sponged on Krylon matte clear spray, pickled and dyed orange.  I then removed the masking with acetone, and brushed on rubber cement.  After pickling and dying red bordeaux, I got the result shown.  It’s not as convincing as I would have liked – the sponging still shows through the red.

11) This was pretty simple – dyed bronze, sponged on Krylon mask, pickled, and dyed green (I forget which specific shade).

12) Similar to (10), I did a dual sponge and brush effect.  But in this case , I did both the sponge and brush at the same time (violet was the base color, with red bordeaux after pickling).

We hold these truths to be self-evident, that not all ABS is created equal

My limited knowledge of plastics has bitten me, as I had half expected (hoped) that I could just run the NIP ABS through the Stratasys and have things ‘just work’.  As noted previously, I had used the exact same extruder temperature on the ABS from New Image Plastics as with the Stratasys ABS, but had sagging filaments as a result. Nophead (who is easily one of the most experienced RepRappers out there) guessed that I likely was using too high of a temperature, which made sense. But shouldn’t ABS run just like ABS? Not a chance, as I found out – I had mistakenly assumed that ‘ABS‘ referred to a specific polymer composition, when in fact you can tweak the ratios of the components (Acrylonitrile, Butadiene and Styrene) to achieve certain properties.  Obviously, NIP is using a different formulation than Stratasys is.  Additionally, Erik de Bruijn noted that he had seen different colors of ABS require different temperatures for optimal extrusion.  Stratasys is apparently doing a great deal of work to make all of their P400 ABS colors act identically at 270° C.  Out of interest, here’s the sticker from a Stratasys reel that lists recommended temperatures for various materials:

E20/E20R is an elastomer (and corresponding support material), ICW06/ICW06R is an investment casting wax and support, P400R (high-impact polystyrene) is the support for P400 ABS, and P400SR is a soluble support material.  It’s interesting that the 0.010″ tip suggests a higher temperature (for one material, anyhow) than the 0.012″ or 0.016″ tips, but I suppose this makes sense – you’d need a less viscous fluid when pumping through a smaller orifice in order to maintain the same linear flowrate.

My latest plate of Mendel parts completed with the NIP ABS, but the results were not as good as with the Stratasys ABS, I’m sorry to say.  I had severe warping on the larger parts, though this could be due in part to me dropping the extrusion temperature down by 15° C early on in the build as a result of nophead’s suggestion.  On the plus side, the NIP peels off of the support material just beautifully (perhaps a little too well, as one of the drive-pulley_3off parts became detached from the raft near the end of the build).  I’m wondering if the lower adhesion between the two materials may have also contributed to the warping – keeping the corners held down more securely may be part of the secret of getting more accurate prints.

I also notice a lot of fine feathery filaments with the NIP ABS at the end of an extrusion path.  Whereas the Stratasys ABS acts like a microscale toothpaste, the NIP ABS acts like a microscale silicone caulk.  The NIP ABS also shows its displeasure at the higher temperatures by turning brown after sticking to the nozzle for a while (meanwhile the Stratasys ABS would simply lose a bit of color, but not appear to actually be charring).  So, time for some testing to see if there is a magic temperature at which the NIP ABS acts like the Stratasys ABS.

I adjusted the white balance on this photo to better capture the filament detail – the parts really do not look this grubby, although it’s interesting to note that the highest temperature pass does look a bit browner than the others.  I created a box with the 0.15″ crosshatch fill in Quickslice, and ran it at ever decreasing temperatures on the Stratasys.  At the recommended 270° C for Stratasys ABS, the NIP ABS is practically dripping off of lower layers.  As the extrusion temperature drops, the filaments droop less, but even at 240° C there is still a reasonable amount of droop.  At this point I started having slippage on the drive wheels (I think this was due more to buildup of tiny ABS fragments than due to lowering the temperature), and more importantly, I was entirely out of Stratasys support material.  As such, the next phase will be to play with the HIPS from New Image Plastics and see how it differs from the Stratasys support material.

A splash of color, part 1

A few weekends ago was Frankie’s aluminum anodizing class at UWM. I had done a little anodizing during the open session last summer, but that was limited to single colors and I wanted to play with multicolor effects this time around. I’ve purchased a video on DIY anodizing as well as the MoonLite and Caswell anodizing manuals, and have done a lot of web research on the process. While these resources cover the basics of anodizing, they say almost nothing about how to create multicolor effects (the MoonLite manual does discuss a few techniques, however). The more complex techniques appear to be closely guarded secrets, and companies like PK Selective (who actually invented splash anodizing – a pair of good examples are here) are not keen on discussing how finishes such as their legendary ‘thunderstruck‘ anno is created.  Piers Wiggett was a true master, and once he left PK, I seem to recall that their prices jumped noticeably – like magicians, anodizing artists are generally not known for revealing their secrets!

Still, there’s plenty than can be experimented with, and lots of tidbits exist on Caswell’s anodizing forum.  By masking with vinyl tape, you can apply different colors to different areas.  More complex and organic looking masking can be done by applying a fast-drying liquid by brush or sponge.  Parts can be dipped into bleach or a sulfuric acid solution (which is what we used) to bleach the dyed color back out, and by slowly dipping into the dye bath or bleaching bath, you can achieve a color gradient.

The actual anodizing bath - a combination of sulfuric acid and electricity. Obviously a bit of caution is in order. The pieces in the bath are the 2"x2" square aluminum plates we used for samples.

Anodizing actually ‘grows’ a very thin (perhaps a thousandth of an inch) layer of oxide out of the base aluminum (unlike plating, in which a dissolved metal is deposited out of solution onto the substrate).  This oxide layer is porous and can absorb certain dyes, hence the basis for the class.  The process of dyeing/masking/bleaching is not unlike reverse glass painting – what you put down first will be ‘on top’, so for complicated patterns, you have to think through the procedure in reverse and plan out your steps in advance.

Various samples on display to illustrate some of the colors and surface treatments that are possible
Steve Krueger was on hand to assist us with our work - I really should have picked his brain a lot more about the techniques he used for all of his samples
Taste the rainbow - Frankie had every single color from the Caswell catalog on hand for us to try

Here’s some of the samples I made during class – photographs just can’t fully capture the look of an anodized piece, unfortunately:

1) Dyed the piece in fluorescent yellow, then brushed on rubber cement, and did a slow dip in the pickle (the weak sulfuric acid bath we used for bleaching) for a color gradient.  I’m not sure why the base color darkened towards the top.

2) Dyed yellow 4A, then lightly sponged on Krylon clear matte spray finish, pickled and then dyed electric blue.  I wish it was possible to get a resulting blue as vibrant as the pieces look right when you pull them out of the dye – the color is so deep and saturated that it really defies description.

3) This was actually my first attempt at a gold/blue scheme.  I did a sponge mask on the undyed aluminum, then I taped a scrubbing pad to the piece.  I then dripped yellow 4A and blue 4A (thickened with corn starch) through the pad, hoping for a fairly sharp color delineation, but the corn starch didn’t make for a very solid paste, so the colors faded together.  Still an interesting effect, though.

4) Brushed on rubber cement over the undyed metal, and then dripped on electric blue and fluorescent pink (again mixed with corn starch).  I tried a few pieces with fluorescent pink, but just could never get it to take well – it always leached out when I rinsed it after the dye bath.

5) Dyed fluorescent yellow, brushed on rubber cement, then dyed deep red.  I didn’t do an intermediate pickle step, assuming that the red would still overwhelm the yellow, but this didn’t really happen.

6) I sandblasted the top half and dyed the piece turquoise.  I’m not really a big fan of blasted finishes on anodizing, as they’re very rarely done well and generally come out far too coarse (in the paintball gun world, at any rate).  From what I recall hearing, the secret to getting a really nice ‘frosted’ type finish is to fully polish the base metal and then very lightly blast it with glass beads, not sand.

7) I ran the metal through a rolling mill with a piece of plastic mesh bag wrapped around it, which provides a neat reptilian skin look.  Dyed bronze, brushed on rubber cement, pickled and then dyed copper.

8) Dyed turquoise, sponged on Krylon matte clear spray, pickled and then dyed electric blue.

9) Dyed bronze, sponged on Krylon matte clear spray, pickled and then dyed olive drab.  The olive drab dyes very quickly, probably the fastest of any of the colors I used.

10) I used another plastic mesh to provide a surface finish with the rolling mill.  Dyed violet DS, and then I used a cotton swab dipped in pickle to draw on the surface.

11) This one went through the rolling mill with a torn piece of paper – it really didn’t apply a noticeable effect unless you hold the piece up to the light.  Dyed in olive drab, brushed on rubber cement, pickled, then dyed in yellow 4A.

12) This one was a pleasant surprise, as I really wasn’t sure of what I was trying to achieve.  Dyed red bordeaux, brushed on rubber cement, pickled, brushed on another layer of rubber cement, pickled, and dyed yellow 4A.  The colors came out nice and rich.

My ABS is all droopy

I ran some more parts with the New Image Plastics ABS, and noticed something odd:

No, not the horrific brightness/contrast I needed to apply (natural ABS does not photograph well if you’re trying to capture detail)  – the sunken, spongelike top surfaces of the parts.  I had been using a crosshatch fill pattern for parts within the Quickslice software which worked just fine with the Stratasys ABS, but the NIP ABS acts a little differently.  Here’s what the internal crosshatch fill should look like:

The pattern is about 0.15″ square – the important part is that the filaments are entirely straight and do not sag.  Unfortunately, this isn’t what happened with the NIP ABS:

In this case, the extruded filament did not stay taut, and the insides of the parts resembled the world’s smallest Golden Gate Bridge convention.  When the final top layers were laid down, they draped over the peaks in the internal fill, leaving a lumpy final top surface (though the outer contours were just fine).  None of the process parameters were changed from the Stratasys material, so I’ve started to wonder what the difference in formulation or processing might be?

In any case, I’m not about to lose sleep over it – the order of magnitude difference in price between Stratasys and NIP materials means that I can dispense with the crosshatch interior fill and use the standard ‘fast’ fill.  Sure, it takes longer to build, but the parts are notably stiffer with the extra density, and the Stratasys has proven to operate reliably when run unattended.

Perhaps tweaking the parameters will help some, but if I have to run with standard fast fill instead of full crosshatch, I can certainly survive.  It’s not as if I currently have a need for ultra low density parts.