Last week the box with my R2 came after a long 6 month wait from pre-order.  And I was not disappointed, but for one weird thing.  First, it is quite a massive machine, larger than an Ultimaker 2.  The 5″ touchscreen is half the front lower panel.  There is a front door with a magnetic latch, unlike the C2 which has no door at all.  Mechanically the R2 shares the general design layout and setup of the C2, but there are some differences.  For one, there is a webcam installed on the left inner panel (looking from the front of the unit).   And the heated build plate is magnetically attached but leveling screws are present which must be manually adjusted, though there is a guided process from the menu.

Unpacking was routine, and the dreaded inner door scratches people reported in early shipments were not present because there was a nice piece of plastic film protecting it.

As I completed unpacking it, I noticed that at the top of the Z screw a bearing was missing.  I checked the packaging for it and could not find it.  To be sure, I even searched in the electronics compartment.  Thankfully this is just a press-in-place 608 skate bearing and I happen to have plenty of those around of good quality. Its purpose is to fix the end of the Z screw so it will not wobble.  If you read my review of the C2 you will recall that I discovered the same bearing there to be dislodged after several hours of use.

Powering it up and after connecting to my WiFi it reported a firmware update was available so I immediately did that and waited for the process to complete.  I then did the bed level adjustment and the Z height offset.  Eager to get it going, I loaded the blue PLA filament it came with and started printing.  And remarkably, everything worked as it should have!

So straight out of the box, plus one firmware update and it worked perfectly.  I’ll get some close up pictures of the machine itself as well as some prints to show you how it does, but the default settings in Cura for PLA give fantastic results.

I purchased a Robo C2 from Amazon 6 weeks ago, and figured it was time to collect my experiences to date. It came packed very well for shipping, lots of accessories and a small reel of filament. There apparently are not many out there yet. Let me start with the good: it has WiFi remote control with over-the-air updates using Octoprint (I’ve had at least 6 updates since I started using it), but it also has a USB front port for direct load of jobs, a color touch screen (though more on that later), automatic leveling (but some say not fully implemented yet), a nice aesthetic for design, and a fairly large build envelope for its overall size. There is a smartphone app to control it too. In general, mine seems very well built but there have been some problems out of the box. On the good side, I’ve gotten some fantastic prints out of it, but also some nasty failures.

This truly should have been a machine that would knock the competition out of the park. With some QC and tweaks it might eventually. But if my experience is any indication, as of now it just doesn’t. You expect a learning curve but if we are to have a reliable “appliance” there must be no points of major fail. So where does this machine fall short?

First there is no heated bed.  OK, not a show stopper – without a heated bed there are still loads of filament options.  I’ve tried PLA (good), TPU (limited success), PVB – Polymaker’s new alcohol polished filament -(OK but not consistent).  Obviously ABS is out.  There are grades of enhanced PLA (Polymaker PolyMAX PLA) which also have worked well.  For now though my best machine for TPU has been my Monoprice Makerselect.

I’ll start with a minor annoyance. The first thing you notice after you turn it on is the head cooling fan.  And by the way, my power switch is “backwards.” The machine turns on when the circle is depressed but the convention is for the line to be “on.” In IEC 5007 standard, the power on (line) symbol, appears on one end of a toggle switch that the control places the equipment into a fully powered state. It comes from the binary system (1 or | means on) – thus “0” means off… The electronics are neatly out of the way in the base of the machine. There is a Raspberry Pi board on the left side and a Robo control board on the right.

So, that extruder fan runs at full blast all the time even if the extruder is cold…  I have not checked but either the fan is a sleeve bearing design or maybe just noisier that others of its type, but that has not been the biggest issue. Now you might ask “why is fan noise an issue?” Answer: this is meant to live in a room like a classroom or other living space and who wants a thresher running at all times? Somewhere I thought there was a control for that fan but I have not found it in the C2’s menus just yet…

But a much bigger concern is this: If you look at the various support boards you’ll see that getting the Z offset right has been a big gap in this machine (sorry for the bad pun…). The build plate has 4 magnets that stick it to 4 acorn nuts. Two of my magnets were loose (clearly they were force fit at the factory, but with time the plastic creeped – so with a little superglue I got them fixed. On first use you manually level the bed by using the “Wizard” for leveling. So far so good. It calls for a piece of paper to be the gapping tool. However, if you use a single sheet of copy paper (at least for my unit) the head will go crashing into the build plate. What works for a gap tool is the heavy, single-layer cardboard you find backing a 9V battery package. Yes, that is many times thicker than a piece of copy paper… but for me it works. I destroyed several sheets of the black painter tape pads that cover the plastic bed before landing on that solution. To date I’ve not seen this addressed, and I also gouged out the build plate too. You can adjust the temperature, feed rate and fan % on the fly but not the Z offset. I would like to be able to dial in a change to the Z offset from the front panel, but if there is an option for that I did not find it. And in a recent update, the Z offset values after calibration have shifted from around 5mm to 14mm with no explanation as to why. As you might suspect, getting the Z offset exactly right is critical to getting a successful print.

Which takes me to the next issue. The Z stage design is typical – supported by two large rods with the Z screw in the middle of them. There is a bearing at the top of the Z screw to stabilize the end. That bearing came loose – again just force fitted like the build plate magnets and over time it popped out of the hole. There is nothing to positively keep it in place. Thankfully that is a trivial fix. Gently pushing on the front of the aluminum plate that attaches to the Z axis mechanics, you find that there is considerable mechanical “slop” or twist. It is as if the linear bearings are not correct for the shafts or the mount is not firmly attached to them. The play is at least 1-2mm at the front and you can clearly see the shift at the rods. Try this with a Wanhao D6 and you’ll get nowhere (these machines are the same price and incredibly rigid).

As for the touch screen, it is resistive – meaning the response is a bit slow. You can’t expect it to behave like your capacitive screen on your smart phone or tablet. I’ve adjusted to it but sometimes you want fast response – like when you are in a hurry cancelling a job when the head crashes into the build plate. The interface is full of features, they are logically arranged, but here and there one is left with some puzzling options. You can remove filament and reload through the Filament Change wizard. But if all you want to do is unload, there is not a direct option. So I ran the wizard up to getting the filament unloaded, then tried to use the “back arrow” to return to the main menu – nope – crash goes the interface and all I could do was shut it off and reboot.

Another thing I’ve noticed is the WiFi connection seems to get lost after a job starts. Restarting the Cura software (a customized version for Robo) usually seems to fix that. I have other 3D printers on my network (MakerBot, Mod-T) that never have this issue. Perhaps this is more a bug with the version of Cura that Robo supplies for this machine (2.3.1).

So I plan to continue testing this machine (of course with filament that does not require a heated bed). On wonders why not a glass plate instead of painter’s tape? If the IR sensor for leveling needs black, surely black glass might work?? Or how about BuildTak(tm)? I’ve created a CAD model of the build plate and will try BuildTak(tm) on it instead of black tape. If it works, look for it on Thingiverse! Thankfully Robo is sending me a new original one on warranty, but at the moment you can’t purchase them for extras.

More details to come and pictures as I get some time in the next few weeks.

UPDATE 5/08/17: I made a request to customer service concerning my crashed build plate and not too long after someone got back to me with a form to submit for a replacement. And a few days later I received not one but two new plates! This time, the magnets were glued in place (a bit of fog on the plastic from the cyanoacrylic glue was evident). In the mean time, I had replaced the tape on my original with BuildTak(tm) custom trimmed to fit. After resetting the gap, it works really well. Because the BuildTak is stiffer and a bit thicker than the painter’s tape pads it comes with, the craters in the top of the surface don’t matter. So I’m saving the new plates for later. I’ve lost count of updates, but things keep getting better. Print quality is great at the default (100 micron) layer thickness and the WiFi connection is not lost as much, though it still does. So it looks like they are serious about perfecting this machine, and I’m getting much more consistent prints. I still need to post some pictures – soon!

UPDATE 7/11/17: Wow, things have come a long way since the release. I lost track of the updates but the most recent one was quite a refresh of the user interface and one that makes sense to me. And the older customized Cura 2.3 client has been replaced with Cura 2.5 customized for the 2 series machines. What a difference! No more dropped WiFi connections for one. And the prints are outstanding at the “default” 100 microns.

So clearly Robo made a huge effort to get this into shape and with all of the updates I can fully recommend this model, and the R2 unit too (just got it and a review is being written this week).

About two weeks ago my long-awaited Tiko mini delta 3D printer arrived. Promised to deliver over 14 months ago, I had high hopes that all of the major design and manufacturing issues had been solved. Not so.  The picture below chronicles a history of failed prints…

Mostly failed prints.

Starting from the upper left corner, that is the print which came attached to the build plate as a sample from the factory.  Next to the right is my local demo print.  Not too bad so far.  But then the next print (the rook) is leaning to one side and the plastic looks very distressed.  Those first 3 prints were with Tiko GEM filament that came with the machine.

I switched to orange PLA from MakerBot.  The Make Robot test piece failed in about the same place both times.  The filament jammed in the Bowden tube and had to be removed manually.  Next is a try with red translucent MakerBot PLA – got further but still no success.  And the rats nest to the right in the middle row is back to Tiko GEM PLA and was supposed to be a rook.  The rotor keychain on the bottom right failed to complete too – this was with M3D Cobalt Blue PLA.  The white part in the middle also failed and was Polymaker PolyMax PLA.  Finally, the last one on the right was back to M3D Cobalt Blue PLA – a larger try of the demo part it started with at the top.

In every case but for the one rook, the filament jammed behind the extruder nozzle.  Removing the filament manually I found in every case that heat creep and softening of the filament 1.5″ – 2″ above the tip caused a plug to form.  The extruder gear then would grind off the filament and the whole thing had shut down – no more extrusion.

This led me to wonder just what the temperature of the nozzle was really. Although I set it to 205C in their WIT control interface, all the evidence pointed to something wrong on the high side.  And now you see why – my FLIR thermal camera is showing the tip temperature when I merely selected “heat nozzle.”  And when printing (set for 205C!)here is what I got – a whopping 270C!  Note that this design uses a PTFE Bowden tube to the head and no true active cooling of the extruder itself.  PTFE softens at 247C or thereabouts, and by 270C is all but useless.  Not to mention that PLA is not going to survive in the heat break area above because eventually it will significantly soften there.

This is exactly the behavior I was getting.  What to do?  Thankfully there is a recovery panel where you can force the machine to recalibrate the heater.  I decided to try this –  and the change in results was immediate.  I was able to print to completion the Make Robot and the demo sample but now a new problem is evident – layer shift.  You can clearly see this in the letter “M” on the Make Robot piece.  Why the shifts?  While not clear at this point, it appears to be a mechanical problem. One theory is that an axis motor gets stuck with the interfering Bowden tube.  It sure seems to be too long.  Another is that the control arms are too flimsy and bend when warm.  They clearly seem too flimsy to me, but some bloggers think this is a deliberate choice.  Deliberate or not, the mechanical integrity of this 3D printer is in question, and many users are reporting this issue.  A third possibility mentioned is the slicer.  I might collect the gcode for the Make Robot and put it into a visualizer to see if indeed the math is wrong somehow and the machine is just doing what a corrupt slicer is saying to do.

BTW the hot-end heater has no thermistor or any means of direct temperature measurement by the controller.  They are relying on the DC resistance of the heater and then controlling the current into it based on some translation in the firmware.   There are no other wires coming from the heated end other than the digital accelerometer (used to determine leveling and mounted on the Bowden tube well above the hot end) and the two wires for the heater.  And in an October Kickstarter update, they discussed “solving” the jams caused by this heater.  This type of design is called a “self-regulating” heater, and these are most often applied where a specific target temperature is to be achieved.  The best example application is the heat tape you use for deicing a downspout or gutter in winter. It is certainly novel for a 3D printer, and from my engineering point of view a poor choice for a low-cost machine.  Temperature regulation needs to be precise and dynamic at the hot end – and this one has a very low mass – thus little “stored” heat to level things out.

So should you preorder one of these at the asking price of $199?  Knowing that I still do not have a truly working unit should help you decide.  Even my best print so far has lots of stringy webs from insufficient retraction. In the picture of the robot you can see this between the arms the and top of the head.  Apparently at least a few folks are getting decent results.  And as of this writing there is a “problem” with batch 4 and beyond but the Tiko folks have not even hinted at what that might be.  And refunds which have been offered and requested due to the long delay in delivery have been frozen…

So I ran two more test parts after the heater recalibration – and the second completed but clearly showed the start of jamming half-way through due to overheating behind the hot end.  So for now my plan is to put it back in the box and wait to see if some update is offered that gets the quality issues solved, assuming that actually can be done!  While there are a number of suggested modifications on the discussion areas, I intend to keep my machine “stock” for now.

UPDATE 1/27/17

There was a promised update for the web interface and the firmware today, and yes indeed, powering it up it got it just fine.  Remarkably, at least for my machine, the update actually made it work tremendously better – it actually got through a long print (3 hrs.) without a layer shift or jam.  Still stringy though…. and yes, now rafts seem a bit harder to get separated as some on their forums have mentioned.  I’m not using the GEM filament that came with it – it is getting discarded… but I’m keeping the spool and refilling it with something of known pedigree that works in other machines.  I have some empty MakerBot Mini spools which fit nicely though they hold perhaps only 35% of the Tiko-supplied full spool.  However they are smoother running because there is less pull needed to move the filament.  I’ll update with some pictures of the new parts soon.

UPDATE 1/28/17

OK, the long overnight print sort of failed… about halfway through the LFS Elephant (http://www.thingiverse.com/thing:257911) there is a clear reduction in extrusion rate, and while it did complete the build, the top-most layers are largely webs instead of solid skins.  I reduced the temperature to 200c for that job (based on how things were looking on the first set at 205c), but I’ve now changed filament again and am running a low-poly squirtle with transparent red pla at 205c.

The gallery above shows the progression of extrusion from a reasonably good layering to the failed region. The first image shows the bottom area of the LFS Elephant.  The second picture shows the clear transition from acceptable to low / loss of extrusion.  The last picture was taken in monochrome at the last and top-most layers.  Clearly the filament is no longer being extruded to match the machine’s movement rate. These pictures were taken with a Celestron FlipView digital microscope at lowest magnification.

UPDATE 2/8/2017

As I hinted at earlier, the signs of business failure were becoming evident. And now if you check the project comments you’ll discover that indeed the Tiko project is in very deep distress. Despite promises, as these things tend to go I do not expect there will be refunds to folks who have preordered one or any more batches to be manufactured and sent – nor warranty support to those who were fortunate enough to be in the three batches that did get delivered (1 for Canada, 2 for US). With Monoprice to offer a $149 delta printer soon with practically the same build volume, touchscreen and WiFi control – the Tiko is simply too little too late. So another Kickstarter 3D printer project goes crashing – the success rate of 3D printer offerings on crowdfunding would be interesting to determine – I’d guess only about 1/3 – 1/2 actually deliver. As far as the Tiko goes, whatever community springs up to be the unofficial support will determine its ultimate usefulness.  For now, I don’t have the time to delve into modifications and improvements. And if all else fails, I’ve got an interesting WiFi controlled desk lamp!

UPDATE 2/13/2017

Could not say it better: Tiko 3d Printer Hard Landing

UPDATE 2/22/2017

The long promised update is now posted. A significant quote from it goes as follows:

“Despite the year and a half of hardware and software setbacks, manufacturing challenges, repeated delays, regulatory hurdles, unending certification requirements, unplanned operating expenses, logistical nightmares, sleepless nights, strained relationships, frustrated suppliers, a disgruntled community, new competitors, and a jaded industry, we simply refused to give up. Day after day, we kept fighting with all our hearts. Evenings. Weekends. More engineers. More developers. More prototypes. More testing. Nothing was ever enough. We did everything we could to turn this around, but by November 2016, we were at a breaking point. With resources running thin, and options running out, we prepared for one final push. We believed software improvements could solve the remaining technical issues, and that pre-orders could solve our financial ones. It was our big chance to get back on our feet.

We gave it everything we had, and we made great progress, but in the end it wasn’t enough. Software had improved, but not enough. Same with hardware and quality control. Pre-orders streamed in, but they too were just not enough. We were right at the finish line, and yet, we were miles away.”

Of course, that status was not reflected in the updates last year.

This morning 3D Systems announced “end of life” for their consumer 3D printer, the Cube 3… Cube 3 Discontinued and while parts and supplies will be available for some time, the refocus of 3D Systems into the higher-end educational and industrial applications is a reflection of the narrow profit margins for low-volume per vendor and lots of competition in the consumer space — not to mention that the expected “3D printer in every home” is not even close to reality.

It has been a hard year for consumer-level additive manufacturing in general.  And despite some record funding, additive manufacturing projects on Kickstarter such as Pirate3D have collapsed in a heap.  Solidoodle is rumored to be bankrupt Solidoodle out of money.  Several others also folded (Phoenix3D, MakiBox, supposedly Zeepro, etc.) and of course as mentioned above, 3D Systems has ended their consumer Cube line in favor of refocusing on educational and industrial markets.

This was as I predicted – the wave of startups brought chaos and fragmentation to the consumer marketplace though there are good choices to be had.  A problem remains — 3D printers, no matter what the make, are not just one more “D” in complexity than ink jet/laser paper printers on the desktop.

However, the educational applications are indeed great.  In the “STEM” field, the ability to create and then “print” designs as part of the overall learning experience is an important enhancement.  In a way, this is nothing unexpected – project-based learning has always been at the core of good teaching.  Learning about an electronic circuit is fine but getting the parts and making it actually work is another level of skill. The same applies to mechanical designs and 3D printing – as part of a larger maker space or even as a community resource.

So what is in store for 2016?  The Consumer Electronics Show in early January in Las Vegas will be one indicator.  I predict we’ll see substantive improvements and useful tweaks in consumer offerings, and I’m hoping that liquid systems, though still fussy, will continue to be refined and reduced in cost.  The “plug-and-play” 3D printer is not here yet and may not be soon, but I’m not giving up on the technology!

And one last word – to keep this blog independent and self-sustaining, a small donation to my “caffeine fund” would be appreciated! Just find the “donate button” on the left side of the web page – even $1 will be gratefully acknowledged! 🙂

With 2014 soon to be done, it is time to write some thoughts on how much further things have developed with consumer 3D printing and scanning.  If you’ve read my past posts, in March I listed the “new crop” of consumer models announced or in progress for sale.  Here is a recap:

Fused Filament Machines (fully assembled)

• MakerBot:   www.makerbot.com
• Cube3D:     www.cubify.com
• DaVinci:      us.xyzprinting.com
• Zeepro Zim:   www.zeepro.com
• CEL Robox:  www.cel-robox.com
• Solidoodle: www.solidoodle.com

Kits:

• Phoenix3D:   www.phoenix3dprinter.com
• Makibox:   www.makibox.com [as of summer 2014, Makibox is bankrupt and did not deliver all orders]
• Printrbot:  www.printrbot.com (and fully assembled)

3D Scanners:

• Matterform:   www.matterandform.net
• Rubicon: www.rubicon3d.com
• MakerBot: www.makerbot.com
  

So how did things turn out?  There were definitely successes and failures, as one would expect.  As for failures, MakiBox was one.  I was fortunate to have an early order that was fulfilled, but many were left with nothing for their $250 investment.  The manufacturing quality was fairly good, the instructions “Ikea” like,  but the cartesian XY system was a bad design and subject to binding or too much “slop.”  I’ve not build anything with it yet, but it does respond to Repetier and should work if I ever chose to.  However, it is in desperate need of a revision to the XY axes system.

MakerBot had tons of problems with their “Smart Extruder” system, continuing to revise it all year long.  I think the latest generation of extruders are doing better – I’m on my third one for a Mini.  Software has gotten more stable and generally works as planned.  However, the Mini and 5th Generation machines are simply overpriced (as is their scanner) compared to other units.  What will they show next month in Las Vegas (the Winter CES)?  Will there be a 6th generation that actually works reliably and allows for an ever wider array of filament options?

DaVinci released their dual nozzle unit and the AIO – a single nozzle machine with an integral scanner.  Blog reports for the AIO suggest the scanner is not much use.  In fact, speaking of scanners, the MakerBot scanner isn’t of much use either – I’ve tried numerous models and the results are rarely usable.  DaVinci has tried to beat the cartridge refill option, although it is still possible to remain on a machine firmware level that will allow this – the web has plenty of hacks that work.  A new touch screen interface for even the 1.0 unit will probably make all the current hacks obsolete, but don’t underestimate the cleverness of the technical folks out there.  The promised release of the Nobel 1.0 SLA didn’t happen as far as I can see, so we will have to wait until next year. Update:  release announced for 3d quarter 2015 at $1499 – which is pretty amazing.

Phoenix3D is more or less dead.  They have very poor reviews of their kits on several blogs.  I’ve asked for a refund of my “deposit” from March 2014 and have not received it as of this writing.  They raised their prices dramatically – probably to try and pay off their refund requests – but they are nowhere competitively.  You can purchase a fully assembled and working Davinci or PrintrBot Simple for less.

Zeepro ZIM continues to fulfill their Kickstarter rewards and also have begun to sell machines on Newegg, including filament supplies.  My machine works but I had an early failure with filament backup around the nozzles – the machine crashed somehow but continued to extrude.  I finally got this cleaned up.  A replacement to the retaining hardware for the Bowden tubes was sent, but my Bowden tubes didn’t fit the new retaining fittings very well and I crushed them too much during reassembly.  I purchased new tubing of a better fitting OD and the fit was perfect.  I need to retry this machine again – there was at least one firmware update while it was down for repairs.  It is very quiet, nicely built, but the main limitation seems to be the web-based interface.  Wish they had gone with a Cubieboard 2 instead of the first-generation model for that part of the system.  Running slic3r on that board is a bit of a challenge – yes, the ZIM has two CPU boards – a traditional RAMPS board adapted to that system, and a web-based server running on a Cubieboard.  On mine at least I can run it on Repetier and it bypasses the internal server completely.

CEL Robox continues to sell machines into the market, but mine had some issues early on.  I actually have now built (or sort of built) some small parts but I would not declare this fixed and ready to go.  There seems to be a Z gapping problem.  What happens is this – you first go through the manual calibration process by using the software and a piece of paper to set the gap.  Once completed, you reinstall the PEI build bed.  Then before each print, the machine measures the Z gap by lowering the head into the plate until a limit switch is tripped. However, my results give a Z gap that is too low – despite confirming the gap as directed.  My first layer is smeared to the table and there are clicking noises from filament jamming – neither good.  So I put in a support ticket over a week ago and so far nothing heard (yes, that includes Christmas but still…).  They appear to have lost their best support tech and there has been little added to the FAQ section (and yes, my problem is not listed.)  While this machine is very well made and has some wonderful features (the host software is actually one of the best I’ve seen), support is slow. Update: got a reply to the ticket on 1/8/15 and they posted a solution.

Solidooodle refreshed their lineup this summer, and the surprise was a new “consumer” unit called the “Press.”  Built in China instead of Brooklyn, the first buyers were offered different tiered pricing depending on quotas.  Well, for some reason, I managed to snag the very first Press order at $349 (it is now $599).  This was promised for fall delivery, but as usual, there were delays.  And as of now, my Press is sitting on a container ship off the port of Long Beach, California, waiting to be unloaded.  The Press even comes with a new software package that appears to be on the same interface level as the DaVinci machines – pretty bare bones but functional for first-time users – the market Solidoodle is trying to reach.  Stay tuned.  I’ll be getting a review out just as soon as it hits my workbench.  This one could be the sleeper of 2015!

Which leaves me to the last one, the Cube third generation machine from 3D Systems.  A beautifully made cast and CNC’d aluminum internal frame with linear sliders all around, this dual nozzle unit is a steal at $999.  And the prints it makes are very impressive – it is just that there is no restock of filament cartridges (they are of course proprietary).  I have 5 cartridges on order from as long ago as July.  I know that they have had manufacturing issues with their plastics supplier and even as late as December I have no idea on when they might ship my orders.  So while this unit might be my every day building choice, one can’t use it without filament!  Lets hope they get this part fixed very soon for the obvious reason – no filament, no printer. Update: cartridge orders are starting to get sent as of 1/16/15

3D Scanners:

As for 3D scanners, there were some interesting developments.  The MakerBot scanner should be completely avoided – it should be priced under $200 for all it does.  The MatterandForm scanner has much more promise and the latest software release is definitely making a big difference.  A few folks seem to have hardware problems, but my unit with the new software outperforms the MakerBot scanner by and far.  As for the Rubicon unit, it is still “in production” but I expect that I should have mine sometime before March 2015.  The software for it seems to be well developed, and I have tried some earlier editions (without a scanner, of course).  But I did receive two other scanners – the Cubify Sense unit and the Occipital iPad add-on.  Both work pretty well, but don’t expect sub-millimeter accuracy.  So long as you are scanning modestly large objects (like people, statues, furniture, etc) you’ll get reasonable results.  On the other hand, engineering scanning with these models is mostly out of the question (things like gears, mechanical parts to copy, etc.).  I will be doing a comparison of these consumer offerings to some commercial scanners early next year.

SLA Machines:

It was thought that 2014 would be the year of the liquid machine.  So far, there has not been a huge outpouring of models, but there are some interesting developments.  I predict that 2015 will see much more in the way of machines that are primarily consumer oriented.  I built a SeeMe CNC “DropLit” as a beta tester – it works, although my main complaint is that there are too many light leaks and there is no lower shutter to shield the resin during idle times.  A LittleRP kit is on its way to me as I write this.  Both use DLP projectors to expose the resin, modified for close focus.  I plan on using a closeup adapter lens instead of modifying my projector.  A new “mini” SLA machine was announced on Kickstarter that uses a UV photodiode array for exposure.  Although small, it most certainly provides a nice intro to the technology.  And the Peachy Printer (another Kickstarter) continues to develop their hardware, but they are way over their initial delivery date.  Hopefully that will be fixed in 2015!

OK, some of my pictures are not optimal, but in the interest of getting some things out there for those waiting or considering purchasing this model, here we go…

This gives you an idea of the size – it is not as deep as it is wide or tall – neatly sitting on a conventional computer desk.  The lighting is bright as you can see, and the color touch screen control panel and power button are all the controls for the entire machine.  Yet it has  a larger build volume than the Cube 1 and 2 units at 6″ in x, y and z.

Cartridges mount on the sides of the unit.  They snap in to the base and the nozzle locks into the head with a 60 degree spin.  You can see the drive pin in the very center of the cartridge tray, and the socket on the cartridge itself.  As far as I know, this is the first use of a remote extruder drive in a consumer 3D printer (the Stratasys Mojo was the first I think on the commercial side).  This design feeds and forces the filament up from the cartridge to the head – there is no motor or other drive at the nozzle end. While the pathway is short and the tube sealed, this design has some potential issues – but on the plus side you get a new extruder drive and nozzle with each cartridge.  

The motor for the filament drive is inside the Cube.  In the Cube 1 and 2 units, the head itself had the drive motor inside, and the filament was drawn from the cartridge.  Yes, these cartridges, like the Cube 1 and 2, are chipped and proprietary, but unlike the opaque shells used earlier, the frosted clear plastic allows a quick visual on the remaining filament.  But I believe they contain as much or more filament than the old cartridges – approximately 500g is on the spool.  And despite what must be a higher cost per cartridge shell, the price is the same as before.

Setup was quick and easy.  The touch screen leads you through with visuals and text to let you know what is next.  The demo build is a small keychain fob and a cursive “Hi!” printed below it.  So after getting things connected, I went on to do some builds of my own.

The one I really wanted to try was the dual color faceted vase that is an included sample.  So I took the plunge and set it up for green on the main part and black on the round side features (in my unit, the starter cartridges were neon green and black PLA). As I made it full size, the total estimated build time came to 36 hours at 70 micron layers.  Things got off to a good start, but somewhere a few hours into the build the green cartridge stopped extruding.  What I found was the machine “air printing” and some of the black that was part of the two-color build was now tossed around the machine (not having the green part printed to the layer where the black would take part in the build).  The problem turned out to be the cartridge was jammed; I could see shards of green filament inside the center of the cartridge hub below the extruder drive. The black cartridge was perfectly fine.   The picture above is how far it got before jamming, but you probably should not judge that build as the best because the cartridge was defective.   However, after this was cleaned up, I printed a black only one in smaller size and in “Draft” mode.  The results in this fastest mode with 200 micron layers is incredible. Yes I’m disappointed that my first build failed due to the green cartridge.  However, a note to customer support got a fast response and a new cartridge is being sent to replace the defective one.  The picture on the left is that build.  I’m not sure I will use the 70 micron layer height much if the 200 micron builds are this good (and remember, this is the fastest setting!).  The model here is about 3.5″ high.

While the Cube3 does not have a webcam to monitor prints, it does have a 2D side view of the part you are printing as a “progress bar” of sorts.  And this can be viewed not only from the unit’s display but also in the apps that are used to sent it files.  Both the android and iOS apps work but neither of these mobile apps allow for the modification of the part before printing (such as scaling) at least for now.  So this is all I’ve been able to print so far, and I rather wait to get the green replacement to try more builds.   The small green circle is another try at the vase in green with the green cartridge on the right side, using the same draft mode.  But alas, the green cartridge is stuck and will do no more…

So am I pleased with this?  Yes, while frustrating to get a bad cartridge out of the box new, the machine itself is lightyears ahead of the previous Cube generations.  There are some things not easily spotted unless you look past covers and slots.  For one, the internal chassis is CNC’d aluminum frames, not pressed steel as in the earlier Cubes.  This makes the unit very light.  Dovetail linear bearings are used for all three axes.  While mine still makes a few squeaks and squawks, I suspect it will get quieter with use.  The waste tanks are hidden in each side but easily removed to empty.  The build plate is a plastic coated aluminum plate with embedded magnets to attach it.  Be careful – they are strong – and not obvious where they are unless you see the mounting pad on the machine where they attach.  There is a fan for the heater assembly in the head and a separate fan for cooling the part top layer.  Both are hidden in the head, which is pretty small to begin with.  Autoleveling and height adjustment are completely automatic and appear to use an optical sensor to determine the adjustments – but I need to investigate this further.

So despite the cartridge problem, I’d highly recommend this unit for someone either new or experienced in 3D printing.  And I’ll be doing more builds and examples soon with comparisons to commercial machines that cost many times more than this does.

UPDATE: Based on early customer experiences, the firmware and calibration options have been improved to adjust for small, but important differences in the two nozzle heights.  This change has greatly improved my results – one nozzle was slightly higher than the other, and if you used that nozzle to build, the slightly lower one would nudge the build layer and cause issues.   As of late November, I’m waiting for new cartridges (which were listed as shipping on 11/25).  I still have some material on my original black and green PLA cartridges, but without refills I’m going to have an idle machine.

Received my unit yesterday – but due to work obligations (work, yes!) I had to just leave it in the box.  However this new generation is amazing and awesome!  Ten minutes out of the box I was printing a test part (keychain fob) and a greeting (Hi!) in PLA.  The small keychain fob is the Cubify icon.

This new 3d printer has some amazing features.  It automatically levels the build plate, and automatically measures the z-height for nozzle clearances.  While there is no live feed camera, there is a very useful icon of the part being built shown on the color touch screen (and remotely via the client app).  As the build progresses, the icon turns from black and white to green like a progress bar on software installations. The nozzles are integral to the cartridge so they are replaced with a new one each time you replace a cartridge.  Users of Cube 1 and 2 machines will recognize the menu structure of the Cube 3, but now the menu top level scrolls across the screen.  Over-the-air firmware updates are now included, so you don’t have to hook it up to your computer for keeping that up to date.

The build plate is now aluminum with a white plastic coating.  Careful!  The coating seems soft but you still use Cube Glue and wow, does it stick!  Be patient and soak your part in warm water to soften the bond.  And by the way, do that in a non-steel container – the magnets are embedded in the build plate this time and they are strong!  I managed to get the build plate stuck in my square metal (iron) cake pan and removing it was not easy.

This machine is also quieter than the Cube 2 yet still makes some noise, but the level is nothing objectionable.  And the light show it puts on during calibration is quite fascinating.  Speed seems fine – a 3″ rook took 6.5 hours at 70 microns.  The result was incredible.  The layers were barely perceptible.

I’ll be updating this in the next few days with some pictures and build samples to show, but based on my first experience, this is definitely the best one out there!

After months past the promised delivery date, my Kickstarter edition of the Robox from CEL arrived this week.  Oh wow, what a nice package and level of build! However, my excitement quickly turned to frustration, and now I have what is a cool looking paperweight that has never made a single part.

What is wrong?  The machine failed to load and purge filament.  The cause?  The head with its motor that flips nozzles and valves from side to side is dead, or perhaps the drive electronics behind it. It heats up, fan runs, lights come on, etc. but there is no extrusion. There is filament to the head.  I’m currently waiting for a response but as it is the weekend already in the UK I don’t expect any help until next week. Will update this if things change, but for now, I’m just using it as a nicely lit (programmable color) paperweight! 

Until more of these are out with users and properly evaluated, I’d avoid purchasing one of these.  At $1400 US plus shipping (post-Kickstarter price) you want something that works first time out of the box. I know one of the Beta testers – he is still waiting over a month for parts to repair his machine.  Again, another problem with the head.  My immediate impression is that this head design will be a challenge to keep working in the long term. So until this is resolved, my CEL-Robox is really a CEL-Robrick…

UPDATE: support finally answered back after the weekend, but it has been slow going.  I’ve been running a lot of diagnostics that don’t solve the (obvious) problem of no extrusion – a failed nozzle shuttle motor  and/or electronics.  I’ve suggested sending back the head for a factory inspection – if it is good (it most probably is not) then a replacement or fix at that end will solve this.   If on the off chance it is working with one of their machines, then I’ve got some more detective work to do with the main box.  But at least we will have the problem isolated by 50%!

UPDATE #2: Sent the head back by air mail to the UK and I’m now waiting for the head to be returned – they quickly checked it on one of their machines and it worked. So the culprit in this case was not the head, but my worry now is that the main board is at fault.  They are sending me some replacement head cables in the hopes that this is the problem.  Despite making some rather close inspection of that part of the system, including the connectors at the head shell itself, still this indicates that the problem is with the machine.   While I’m pleased with the level of response from their support team, I’m only a little closer to getting this to work.  And since I never got any error messages or clues from the system, I’m probably going to go to low-level continuity checks of the wiring to the head to get a resolution.  I have already invested 20 hours of debugging time with more to come.  I sure would like to get this thing to print!

UPDATE #3:  The head is back and they sent spare head cables in case something was wrong with the original parts.  So I refitted the head and powered it up – connecting the box to my computer again.  Of course the first thing I wanted to do was test the head nozzle shift – and I hit the button for a nozzle change and AMAZING — it worked!  Thing is, I had never seen this movement before in response to the nozzle change commands.  And I didn’t do anything to the electronics before this was returned, other than a firmware update that came while the head was in transit.  OK, so it is now on to giving this a try for building a part!

UPDATE #4: One more problem – I calibrated the nozzle height according to the instructions, but the head is still too low when it starts a job.  The result is a smeared first layer and occasional grinding noises (from filament getting jammed).  I’ve opened a support ticket.  I’ve not been able to make a single successful print on this machine yet, so despite what appears to be a well-engineered machine, I’m still unable to use it.  Sadly, this failure came during a somewhat public demonstration at my college, so it wasn’t a good sell for the product.