Huxley RepRap Rebuild

posted 21 Nov 2015, 04:56 by Colin Bell   [ updated 2 Dec 2015, 22:56 ]

My first experience with 3D Printing came with the Huxley pictured to the left, way back in 2011, in which the build experience and learning was documented on several blog posts here. This little printer will always be dear to me, as apart from being my introduction to 3D printing in general, it also got me through several months of illness and taught me a great deal about engineering machines with moving parts, designing and manufacturing complex and functional parts, how to interpret some very experimental guides and use cases. It also opened my eyes to the fact that machines in general are creations that can be modified to do exactly what you need them to do, as opposed to what a manufacturer wants you to use it for. Long live open source hardware!

As with most first attempts though, although the building and seemingly endless calibration gave some reasonable printed results, it never produced anything close to the quality I had in my mind's eye. Sure, parts were functional, and some of them are still in use today, but for the most part they were not even a little bit on the pretty side. 

There were a number of factors which caused my aesthetically not pleasing parts - my own mechanical mistakes and dodgy engineering skills of course, of which there were many, old school slicing software (does anybody still use SkeinForge?) inaccurate flow rates and speeds, bed not being level, poor quality filaments...the list goes on. Combine those things with the fact that the design was based on stuff laying around in a machinists shop and the legacy was based on mature but not perfectly suited CNC tool-path software and methodology, and your 1st Gen' Frankenstein Cartesian plastic spitter was at best a great concept, but not the speedy, highly detailed model producer you imagined it would be. 

But, it inspired me to follow developments and also make my own improvements. It also introduced me to a large community of like minded tinkerers, creators and engineers on forums, and of course has quickly developed into the medium-large(ish) industry that it is today. My current production machines are faster, more accurate and repeatedly produce the kind quality that I only dreamed of back then - with about a 10th of the hassle to boot. But without the old Huxley, I'd never have gotten the bug, and it's for this reason that I decided to pull it off the shelf, blow off the dust and recondition it with a few modern additions to make it a useful machine again.

Visually, you can tell this device was a constant work in progress. The wiring is a disaster from several quick repairs, the electronics just hanging off the side, Molex interconnects replaced with quick and dirty solder and tape, and there's more cable ties than you can shake a stick at. Clearly all of this needed to be addressed. Obviously, wiring tidy up is one of the last jobs on any build, but I committed to at least make it look safe, if not tidy. The open frame design makes it impossible to completely hide the wiring, but at least it could be rationalised.

The most annoying thing visually by far was the frame mount I'd hung the Sanguinololu electronics in, and the cable tied power switch. I decided that I wanted the electronics to be mounted within a separate enclosure, along with the power switch and inbuilt cooling for the stepper drivers.

I started by designing a basic chassis to house the Sanguinololu, with two separate side boxes - one for the power switch and the other to house the power input and hide the wiring. The design was made as usual with 123D Design, and printed in ABS so that I could Acetone weld the parts together later (this works very well by the way and produces a much stronger bond between parts than most glues). I also created a top for the boxes, which has a hole for a fan, and also a simple slide mount for the display to be added later. The fan in use on was a cheap 5v axial, and the incoming voltage to the Sanguinololu was 18v, so I put a simple 5v regulator (an LM7805) mounted on a bit of strip board to supply it. 

I also created a custom bracket which fits into the boxes, and then can be bolted onto the existing support rails of the Huxley. This arrangement along with the addition of modular Molex plugs on all the cables means that the electronics can be removed when the machine is being moved around, or left in place permanently.

As you can see from the various pictures, I went through several iterations of the design, however the basic design stayed pretty much the same.

Unfortunately, I don't have a picture of the finished case fitted to the printer, and tidied up properly, but these shots should give you the basic idea. 

It did all seem to work OK anyway, so I moved onto the next thing. If anybody needs a modular box for their electronics, it can be found here.

Next up was the mechanical side and I could see a couple of minor, but nonetheless impacting issues with the machine as-was. In terms of frame, rods and bearings, it all looked pretty much solid, however when I built it, I only had the printed parts which were supplied in the eMaker kit, and I had no CAD design knowledge to easily replace them if they broke. Given that the parts used on this were from the 2010 Indiegogo campaign, they were quite rough in terms of quality, and also some parts were not particularly strong due to design constraints of the time. Over time, use and abuse, some of the parts had required repair with glue or heat welding, and also due to the fact that I only had a limited amount of filament on hand and no easy way to get more, I didn't print spares. These days, my store carries around 100kg of materials of all descriptions at one time or another, so I carefully inspected all printed parts, and created new versions as required. In particular, the X-ends had very weak slides where the bushings sat to hold onto the smooth Z rods, and also never had a very smooth belt path, so I reprinted those using an updated version. 

The Z couplings also required replacing, along with the PTFE friction fit tubes used to hold the Z thread rods to the motor shafts - those things always managed to work themselves loose, usually leaving the Z axis only lifting on one side during a print - very annoying. A touch of two part epoxy and the couplings printed at 99% normal size fixed that issue permanently.

The extruder that was fitted, had proven to be very reliable, however it was noisy, using PLA printed gears and additionally the bearings inside were heavily worn. The design of the original wasn't the actual 'original' though, it was a design from the times when the choice of filament colours available were white or eh...white. This extruder was special in that it had holes within the filament path allowing you to stuff coloured pens into the filament as it passed through - instant coloured prints! Well sort of - it worked after a fashion. I remember I ironically printed this colour changer using a sample of blue & silver speckled filament which actually looked great. For the rebuild though, decided to go back to the original design, but slightly enhanced with ABS Herringbone gears and shiny new bearings, which produced a virtually silent extruder that made a light whizzing noise rather than the bone rattling racket it previously made.

One of the most annoying parts of any 3D printer is the dreaded hot-end block, which often requires a complete strip down of the hot-end, and liberal use of Napalm and Atomic weapons to clear. Well, maybe not quite that bad, but trust me, the original hot-end on the Huxley definitely fell into the clogged more than not category. In retrospect, this was down to the Bowden tube setup and software controlled retraction being an unknown black magic art at the time. Whilst I suspect that this could be largely fine-tuned these days with better knowledge of what's actually causing the problem (after all, I have other Bowden set-ups with no oozing or clogging issues these days), I decided that I wanted to move from the original hot-end and replace it with a modern version. 

I decided on a stock 0.4mm Ubis hot-end from PrintrBot, largely because I had it lying around, but also because I've never had a single problem with them.
Of course, this required a complete redesign of the original carriage, again which I created in 123d Design. I used a couple of existing designs available, and added my own modifications to mash up a new carriage which sports a filament fan, and improved belt guides and grips. One of the challenges was designing the mount so that the taller Ubis would fit through the gap at the top of the frame, so as to not lose any Z height. Overall, I think the design of this part, building on other people's excellent designs was one of my favourite parts of the entire rebuild. The addition of three new high quality linear bearings completed this part of the project.

The X carriage/Ubis Mount models, and further details are available here for anybody that needs them.

The next thing that required some TLC, was the original Nichrome wire based heated bed. I've already blogged about this particular upgrade in a fair bit of detail here, so won't cover it again. I will just say though, that replacing the Nichrome with a modern, safe and efficient silicon pad, was the smartest move I made on this project - I'm pretty sure the Nichrome bed had me on its hit list for one day...

(Yep, thats how the wiring looked on the Nichrome bed after a couple of years use...I'm pretty sure it would have killed me somehow at some point!)

Of course, while I was working on the heated bed, I cleaned off all the old Kapton tape and prepped the surface with a nice new PEI sheet. Well, it would have been rude not to, right? 
Model to bed adhesion during printing was always an issue on this machine, partly due to bed leveling, Z height and uneven heating - the PEI all on it's own takes care of around 60% of the issues, with another 30% down to even and reliable heating - the other 10% is bed leveling and consistent Z height before printing. As mentioned before, the Z couplers help a lot with the consistent Z height issues, and by putting smaller, stiffer springs under the bed I was able to virtually eliminate that last 10% as well.

Readers of this blog will be well versed with my love affair with PEI, and I have it installed on all of my production machines now. If you want to read some of my reasons and impressions, have a look at this post.

Mechanically, the printer was ready for action again at this stage, however I felt that to truly modernise it, it needed a display and rotary encoder - well it is 2015 for goodness sake! No self respecting machine is complete without a display, and the rotary encoder means it can be used without a PC, which is surprisingly handy on a small portable machine like this. I've made this upgrade to all of my printers now, and can't imagine not having it there anymore. Looking around in my spares bin, I found one of the original Panelolu displays, which I'd actually bought for this device over three years ago from Think3dPrint3d but never gotten around to installing.

The kit comes in component form, and consists of a PCB for the buttons, LED and rotary selection switch, MicroSD card reader and of course the display module itself.

Wiring was achieved with a ribbon cable, and the schematic to get this right is here, however I must admit I struggled to read the schematic correctly and ended up following simplified instructions from the Soliforum unfortunately although the post is still available, the pictures are all missing, so I've contacted the original author for permission to repost the offline copy I have of it here.

The only thing to look out for if you make one of these, is the length of the ribbon cable which goes from the display to the controlling electronics - any longer than 15cm on this design and the display doesn't work. I've had similar issues with these panels in the past, so my guess is there is a signal loss that cables of longer than 15cm generate and the displays cannot compensate for it.

Additionally, I printed a case for the display hardware, and a simple stand which would slide into the box I made for the electronics, mounting the display on top of it. The case itself is based on the Think3dPrint3d design, however I designed in a spacer in order to give the wiring a bit more space (yeah, I added some additional electronics to play with).

Of course the printer's firmware will need updating in order to use the display and encoder - as the Huxley had a three year old firmware on it, I upgraded to the latest Marlin code and modified it for the Huxley's specification including the added display and encoder. Because I'm writing this article some months after the event, I won't go into the details here as it's all a bit hazy now (I'm getting old...) but if anybody needs help with this aspect, just reach out to me and I'll be happy to try and help out.

Apart from a bit of belt tensioning, and getting an initial level on the Z axis and bed, the hardware side was complete now and I headed into the calibration stage - previously this had been a never ending task on this machine, but I'm pleased to say that all the upgrades actually made it very straightforward. The only real challenge was getting the e-steps/mm right for all the axis' as initially every print was too large by around 4mm - this wasn't too bad to sort out, and I had a nice pile of calibration squares once complete!

The new heated bed worked like a champ' as expected, and heated up in no time. The Ubis on it's new three bearing carriage also performed flawlessly and was producing excellent prints in no time - there was a big of stringing going on due to the Bowden setup, but this was easily fine tuned in the retraction settings. Additionally, the new stiffer springs in the bed made levelling it much easier, and the new Z couplers meant that once it was set level with the bed, it stayed there.

One of the objects I printed with the Huxley back in the day, was the requistite Yoda bust, so I re-printed it with stock settings to see how they compared - the original is on the right and the new version on the left - not bad I thought!

So the next step was to get the machine to it's new owner, my son. We currently live in different countries, so I carefully packed up the printer, with a bunch of handy spares and some filament, and lovingly loaded it into the taxi - of course when getting out of the taxi at the airport, the bag got dropped, smashing a couple of plastic parts on the X axis...but fortunately, part of the 'My First 3d Printer' kit I'd put together for my son included a full set of spares. This meant he got a live masterclass in replacing parts on his first view of the printer, which was kinda fun although I swore quite a bit as you can imagine! The lesson here is that there is a reason why the pro's package things so carefully and don't just chuck them into a duffle bag!