AKA47

Ok, again, it has been a while but there has been a lot happening. The Thermal re-calibration worked and the filament no longer  burns which is much better. I have finally settled on a version of the Teacup Firmware that seems to be working very well and have got quite some way into the fine tweaking of Teacup's configurable bits. The net result being that I now have a machine that prints well enough for smaller components. Most of which at the moment have been parts contributed to the Sheffield RepRap group at the Gist Lab.

The configuration tweaking at the moment principally revolves around adjusting how much to retract/advance the filament for each stop/start of the extruder, whilst increasing the printing speed. My Bowden Tube is at the moment a little longer than I would like. The longer the Bowden Tube the greater is it's elastic length and the more I need to advance/retract the filament. These adjustments have a big impact on print quality and they seem to be interacting with the print speeds and influencing the quality adversely as the print speed increases. My parts are printing OK and I have little or no stringing but at the points in the printing where the head pauses there is a small amount of blobbing. I think this essentially means I need to retract more and quicker as the print head comes to a stop. Finalizing the extruder mounting and shortening the Bowden tube to the minimum practical length will reduce it's elastic length.

My Extruder is still not mounted on the Huxley itself but hangs from a hook on the shelf above. This is something of a chew when I take the machine to the RepRap groups build days, which we hold once a month on a Saturday, as I can never seem to find a way of standing it on something without the gears et al catching on something. AJ, a member of the group designed a mounting plate in Openscad but as I have no heated build plate the warping has proved too much to build it as yet. Making this plate then will be a manual hack and cut job, for another day.

I have also attempted to printout 4 Mendel frame vertices to make a filament spool roller, just like the one Nophead carries with him on visits. Again the warping proved too much. I may attempt them again, now I have persuaded the Huxley to print quicker and have found wide blue masking tape for the print bed that the ABS sticks to quite well. We had a brief experiment with a number of types of tape to see which would work best and found the wide blue decorators tape was it. I found it works better if it is applied and left for a day or two before usage. The glue on the tape seems to take a while to get a grip. The Blue narrow tape was feeble, principally because it seems to like to lift along the edges first and being narrow there were a lot of edges.

A colleague gave me an ASUS Eeepc that had a broken screen after he saw me scouring Ebay for them. I wanted something small, cheap and able to run Linux that could spend it's time dedicated to running the printer whilst I got on with other things. Having something that would also network connect for remote printing is great too. It was proving too easy to mangle print sessions part way through by doing other things on my laptop (updates etc) at the same time. A replacement screen came from ebay, and the RAM was upgraded,  finally adding in a larger SDHC card. The Eeepc was rebuilt with no swap and the SDHC card mounted on /home to expand the available SSD space (Only 4Gb). Configuring it without swap is intended to extend the life of what is now an old used SSD, and improve performance. Fortunately it would take enough RAM to let me do this. All in all it works extremely well although the screen is a little small. It is running Ubuntu 11.10, and has the Arduino build environment, Skeinforge and GtkTerm for firmware building, part slicing and printer driving.

Finally on to Heated Build Platforms. Yes I definitely need one of these, having experienced too much warping. It is preventing me progressing beyond small components. Increasing print speed helps reduce the warping but not enough really. Muhammad from the Sheffield RepRap group very kindly gave me a spare bed he had been experimenting with. The heating element measured out at 44Ohms so a little too high to get much wattage out of a 12V supply but probably do able if the voltage is increased. Attempts to make my own PCB heater plate are currently a little miserable as the size at 200mm by 200mm is just too big for my laser printers printable area (bits get chopped off, on printing). My Huxley is just too narrow to take one on of the usual heater plates that can be bought in. So still a work in progress. I did manage to find 4mm Mirror tiles at exactly 200mm by 200mm though to trial as the build surface. As well as self adhesive copper foil to act as a heat spreader and heatproof oven liner to stick the copper foil to and insulate it from the top track on a heater. When I get one made. So still much to be done.

Hot plastic. My extruder was running on the hot side, Nophead suggested this was due to the difference between the internal and external temperatures and he was spot on. Whilst I had originally constructed my Thermistor lookup table using a temperature probe clipped to the outside of the nozzle the inside was running a good 40 to 50 Deg C hotter.

 The problem is fixed now and a new thermistor table is in place all with the help of an Arduino based tool to help the process along.

Progress is happening now although still slowly. Mostly due to available time and resource. The huxley build has now progressed through build and is nearing the end of commissioning. The mini-mug lineup (Or could it be called a mug shot!!!) shows the progresion of increasing print quality.

The mug to the far right is an examplar printed on Sheffield Uni's RapMan this gives me something to compare, aim at and helps me judge how my efforts are going.

The sequence then from left to right as as I have adjusted firmware parameters to improve print quality. The first I aborted as it was laying down far too much plastic. I aborted the print and set about recalculating all the essential firmware parameters. Rather than doing them one at a time I created a spreadsheet and put them all in (even the ones I was fairly sure about) The spreadsheet is attached here huxmechcalc.ods is is an Open Office spreadsheet. The next two mugs resulted as I changes some values before I realized that I had made some mistakes in the spreadsheet which gave wrong values. I corrected these and the fourth mug along resulted.

From the fourth mug onwards the changes in build quality were made by incrementally increasing the steps per mm extrudate and how much the extruder retracts the filament between sections.

As I have used a Wade's Extruder and Bowden Tube feed to my print-head I need to use the retract feature in the firmware to reduce the blobbing and stringing that results from the flexibility of the feed tube. I chose to use the firmware's retract feature over the slicing software's equivalent because I believe that the slicing software should have to know as little as possible about the machine specifics. I can then use the same slicing configuration setup across a number of machines. I guess also in trouble shooting it helps to be able to partition a problem st so as to work through the issues logically.

The extruder hot end is my own design which I will publish here very soon as I feel reasonably confident that it is working OK.

A major item left to fix is the discoloration. The exemplar mug was printed with a segment of exactly the same feed stock that the test mugs are printing with and I am seeing noticeable discoloration. I wasn't sure if it was running the print too hot despite having calibrated the thermistor table using a known good high temperature thermometer. Perhaps the bit I clipped the thermometer probe to was cooler than the rest or maybe the PID parameters which are as yet untuned are causing too much overshoot and instability.

Throughout I have focused on turning up the Firmware, to make sure I was only seeing the results of firmware tuning I have used the same G-code file for all the tests. It took some doing to resist the urge to re-slice but was worth it.

Dig out your Pocket Tech (or Diary) and make a note of the 28th of March.

Adrian Bowyer The man, Maker and Reprap visionary is coming to Sheffield to visit the Sheffield Reprap Users Group.

Everyone is welcome, if you would like to drop in and meet Adrian, see some Reprap machines, hear Adrian's presentation or get in touch with the User Group, sign up for a ticket (currently free) and come visit us.

eTickets are available online here: http://reprap1103.eventbrite.com/

Please do take the time to register for a ticket as this will help us towards making sure we have enough space to get everyone in.

I have been chewing on the idea of transferring a bunch of the motion control parts of Reprap style machinery from microprocesor firmware and into dedicated hardware for quite a while now.

Uptake etc is unlikely to be large enough to warrant dedicated silicon and I often wondered how big a programmable logic device I would need to put a design into. How much of the motion control would be good to transfer ??  Which device to use ?? All those sorts of questions are very interlinked, the answer to one will invariably impact on the others. Perhaps so much so that it can drive a person mad just carrying it around in their head. A bit like one of those terrible jingles or tunes that you sometimes hear and can't get rid of.

That doing so will give performance benefits is without question. The doing of it though is full of them and the only way really to answer any of them is to have a go and see.

It may push up the end cost of a machine and will also increase the technical complexity. Arguably beyond the skills set of the average tinkerer. But the questions and thoughts won't leave me alone, so it is time to start progressing something that will answer them.

I have experience of programmable logic design having done some professionally in the late 80's and early 90's but was out of touch with the current state of the art. Rummaging around on t'internet It is clear that rather than becoming more capable, generic and easier to work with perhaps the reverse is true. What I did stumble across though was the web site for Eric Rogers and the AMANI GT given the obvious synergy between this project and the arduino it looked like a great way to start and a useful test bench to work out some ideas.

Eric was kind enough to send me a board so I can build one up and see where it goes.Thanks Eric much appreciated.

The major downside to programmable logic across the board is that there is very little in the way of an integrated open source tool chain. Work for now will have to be done using the free development tools made available by the device manufacturers. I have created MoCo5D a public project over on GistHub so as to publish the work in progress and open it up to anyone crazy enough to want to join in.

The Mechanical and Electrical build of the Huxley machine is about done. It is now time to load up some firmware before working through commissioning and testing. I haven't implemented a heated build plate yet, but this is definitely on the cards.

The eagle eyed will spot that I have used a RAMPS shield with a spare Mega I had available. Even though I am still working through using the min644p and rugged stepper driver work. I am a touch easily distracted and after 3 years of getting sidelined into new avenues of investigation I finally decided I should focus on getting something completed.

Even if only to use it as a test bed and extension of my tooling for the other work I am doing. It is arguably a bad move to design things for folk to create using Reprap style tooling without being able to trial the design and iron out the wrinkles before publishing the work. A chicken and egg thing I guess.

So as cash has been a little less tight this last month I splurged out on some parts to accelerate progress. Having a machine that works should double my productivity because it can be printing whilst I get on with the other tasks I want to focus on.

The controller electronics are mounted on the side bars to the bottom left of the picture so that they don't get in the way of the print carriage and I can get a little more vertical build space.

You can probably just make out the squash ball shock absorbing feet tie wrapped onto the bottom bar of the frame. These lift the machine up enough to stop the desk catching on the Z axis belt and absorb some of the shock created by the rapid moment of the bed and carriage. I made these by sawing off the bottom third of some squash balls and punching out two holes to cable-tie them to the cross bars. The idea wasn't mine it was posted up here as comments to some earlier articles. Many thanks to the contributor, you are a star.

I am planing on using the Teacup Firmware, the guys working on this have been doing a great job and I feel this needs to be rewarded by greater uptake. The build process is fairly straight forward, it builds and installs most times with little or no trouble. Which host software best matches the Teacup Firmware is at the moment a little uncertain, Partly because I have only just got to the point where I am far enough on to put aside the mechanical and electrical build and focus on the warez.

I desperately want to get the machine printing now as the urge to create ball and claw shock absorbing feet using squash balls and a printed claw holder is driving me mad.

Finaly got around to writing some OpenSCAD parts and published them via Thingiverse.

• Printable Bowden Wade Adapter

• Printable Spring Seat

• Printable Spacer

 The adapter and spring seats are printable versions of the ones I made on the lathe for my Huxley build. The spacer designs are simply to ease the frustration of never having the right size spacers to hand. They can also be printed thin and used as washers. I guess this reduces some of the reliance on Vitamin parts for RepRap builds.

First part of my two part project to create a decent rugged stepper driver board is complete.

The project is written up as far as it goes here.

This is a simple, manualy configured driver board that can do micro stepping and can be made by a kitchen table constructor. It is not meant to be small and there are no surface mount components used in it's design.

Based on the Toshiba TB6560AHQ stepper driver IC it features 2.5A continuous and 3A peak drive with advanced current control features.

Looking at the space taken from the underside of the print bed down to the surface of the desk on which the machine is sat it is clear that there is a lot of wasted space that could be re-designed back into the build space.

Indeed I initialy tried to assemble the machine in a way that used more of this space. Primarily by mounting the rails under the cross bars rather than sitting them on top. Unfortunatley as you can see in the photo the fact that the Z axis cross bars are in this space meant that one got in the way of the other.

Prusjar in his designs published on Thingiverse has I think recognized a bunch of these things and is attempting to re-design some of them out.

In an extended format like this build the wasted space is a minor nuisance, something trivial in comparison to the work that has gone onto the whole thing. In it originally envisaged size though it could be a bit less trivial.

In summary, despite my zen urges, the design and modifications so far that have gone into making it possible for me to build and critique this machine are awesome. The machine as built is awesome and I can't wait to print with it. It is very much fit for purpose and I could recommend it to anyone wanting to do the same. There is still room for further improvement and more parts to design as both part of that improvement and also to reduce the number of bought in parts that are necessary.

The Huxley variant as I have built it uses the standard or improved huxley parts throughout but has longer framing and rails. This makes for a huxley with a larger print volume than the standard smaller assembly.

I especially kept the squashed frog to the initial dimensions. This has the effect of increasing the amount of travel that is available to the bed carriage as you can see from the photograph the bed is now free to project out over the frame ends at the extremes of its travel. Whilst it places the mounting screw heads firmly within the build space I have countersunk these to compensate.

Having assembled everything I found that the bed could have actually been a little larger along the Y Axis as the print head is clear of the bed at both extremes. I estimate that the print space on this build to be 190mm * 200mm.

It should be theoretically possible to expand the framing and rails further there is however a very real question as to whether the rigidity of the M6 frame and rails would be enough to produce a machine without wobble. The current sizing is I feel at the upper bound of what can be got away with. This is in part made possible by keeping the mass of the print head and carriage quite low. Clearly if you are going to build a larger machine it would be better to move up to M8 framing and a full Mendel build.

Here is a snapshot of the underside of the X Axis carriage with the print head or hot end.

Looking up into the heat shield you can just see the aluminum block of the heater assembly. This uses Nopheads arrangement of vitreous power resistors mounted into an aluminum block along with a thermistor and the brass barrel of the nozzle.

In this case I have used two 3R3 10mm by 5mm resistors wired in series to give 6R6 in total and also gives better coupling and dissipation into the heater assembly than a single 6 Ohm resistor.

The nozzle is a standard brass acorn nut in which I bored a 0.5mm hole. The heat shield is turned from Peek on the Lathe as is the stem. I used 10mm Peek for the stem drilled out to 5mm to accept the Bowden Tube and tapped to M6 to accept the brass heater barrel at the hot end. The Bowden tube feeds right down the stem and buts up to the end of the heater barrel. No idea yet how this will perform until the printing starts.

Here is a bottom view of the Y Axis carriage or Bed.

The view shows clearly the plywood squashed frog and how the trucks and clamps mount to it. Fully assembled there are some holes left unused. Two of these are for the purge plate. I left this out as I really can't think of a use for it. The third hole i have no idea at all about.

When trying to initially put all this assembly together it took a little while to work out as the available pictures didn't quite show how it went together.

Having got this far with the assembly I feel that the design has done bearings to death. Although the action is pretty smooth as a consequence.

The Y Axis suffers from a little bit of the same problem as the Z Axis in that the teeth of the drive belt are running over at least two of the idlers. With this axis though you can definitely feel the effect this has as you can run the bed back and forth by hand and feel the juder that this causes. Again it will be interesting to see if this makes any difference to the prints or remains as a slight nuisance. The same fix for the Z axis would work here too.

Here is a close up of the adjustable sprung bed.

The mounting screws are counter sunk to avoid the likelihood of the nozzle getting caught on them and breaking something.

I used spring seats or retainers to keep the springs centered and to make sure that they sat flat against the bed and carriage. These were turned on my lathe but are a part that would be very easy to print. Again a lathe is what I had to hand and a working printer wasn't. Another part then for Thingiverse.

Springing the bed gives a small degree of protection to the nozzle in case of touch down. But more importantly makes it very easy to adjust the bed to be level with respect to the nozzle by turning the screws. I did toy with the idea of using neoprene or something similar bu in the end went for what i had to hand.

Through the bed you can also see the top of the plywood squashed frog that I cut using the published templates. Also the drive belt clamping to the carriage assembly. The cable tis is to fasten the spare bit on the end out of the way.

Here is the right hand side of the bottom of the Z Axis. Just like the other side this part projects below the level of the standard frame corners. The side we can see is the driven end as there is the motor. The sprocket on the motor is another one of Nopheads wonderful creations.

Whilst none of the toothed side of the drive belt runs around any of the idlers on this end the belt rubbing against the stationary mud guard washers could be improved by replacing the idler with a pulley

The left hand side of the bottom of the Z Axis. This part of the assembly projects below the level of the standard frame corners. The side we can see is not the driven end as their is no motor. 

Looking at the top idler we can see that the toothed part of the drive belt is running around a small diameter bearing as the idler between stationary mud guard washers. this produces an interesting feature in that the assembly is noisy when compared to the carriage assembly where this does'nt happen.

Put simply the gaps between the teeth on the drive belt are preventing the drive belt from running around the idler smoothly. Depending on how many teeth are in contact with the idler the length of the belt path is ever so slightly longer or shorter. Whether or not this is anything more than a slight nuisance remains to be seen. An inconstant belt path length may result in very minor artefact's in the print.

This idler could be improved by replacing them with pulleys with faces that are matched to the teeth of the drive belt.

The same idler just discussed is also the adjustable component that is used to tension the belt by removing slack.