Sunday, 15 November 2015

Long time coming but here's an update.

Being so busy at work and having another 3D printer at my disposal the delta has taken a bit of a back seat but with the holidays coming around I think it is time to jump back in.

I have done lots of testing and have found the following.

Heater Bed:

The bed on my delta is made from 6mm aluminium plate with a 280mm diameter piece of 6mm float glass on top.  The plate is secured to the printer floor in 3 places corresponding to the 3 towers using 6mm  bolts.  Between the floor and the bed plate are 3 heavy springs.  This means I can adjust the plate up or down easily using the bolts whilst keeping a constant upward force on the bed plate.  The advantage of being able to adjust the bed manually is it means there is less tweaking with the end stops during calibration.  I much prefer each end stop to need the same amount of adjustment in the firmware.  

I started with off with a MK2B 24Vdc heater for my bed but unfortunately the time taken to get to 110ºc was nearly 50 minutes, and that is even with insulation on top of and under the bed.  This pre-heat time was just too much for my liking so I have changed over to a 220vac 500w silicone heater pad form Keenovo.  It will get the bed to 110ºc in just over 4 minutes which is much better.  Yes you have to be more careful with mains voltage but with the correct wiring and precautions there is no problems.

Testing the bed for movement whilst heating. 


To my surprise even with the full aluminium construction, 30x30mm extrusions, 12mm linear rods and additional supports the frame still flexes when printing as confirmed buy putting a dial gauge on the bed and pushing the frame.  It's not much but enough.  To fix this I am going to fold a 0.9mm sheet steel skin that will be fastened to the frame.  This skin will tie the whole printer together and should stop all flex.  to my understanding the flex was causing small ripples in the prints as each ripple always followed the same plane.

Bowden Tube Extruder

After watching the bowden tube in action I am going to suspend the extruder above the hotend to give as shorter tube as possible.  I have found that the retracts are very hard to get precise and there is far too much fiddling involved to get this right, or should I say nearly right.


Once the enclosure has been complete I am going to rig up a fume and particle extraction system.  the idea is to recirculate the air within the chamber so as to not pull the heat from it which will affect the ABS prints.  There will be a turbo mode that can be turned on just before the door is opened to scrub the last of those pesky smells that we love to hate.


this is where I am still on the fence.  I have had great results with Cura and Kisslicer but haven't fallen in love with either.  Each have there pro's and con's but Simplify 3D has caught my eye but not being able to download a test version it is a concern given the outlay if I don't like it or it doesn't suit my printer.  I have seen it in action at my CAD software vendor and heard good things but I am still on the fence.

                                    Lots of dragons printed testing the various slicer settings

Dual printing

I have been working on a dual extruder end effector in CAD but it is still very much  in the thought stage but something I want to get going.  Even if it just to print soluble support it would be great.

Anyways, just a quick update but will try to add more post sooner rather than later.

Wednesday, 14 January 2015

Bolt hobbing and finishing the Extruder

The Infill 3D extruder needed a hobbed bolt so a good opportunity to make one up.  Before starting I made a simple jig to hold the bolt to be hobbed in the vice of my mill.  It has a bearing at each end so the bolt can freely rotate to ensure nice clean cutting and can be downloaded here if interested

So, the bolt was fitted to the jig and the assembly held in the mill vice (I have removed the surface rust from the vice since the photo's were taken).  I opted to use an M5 3 flute steam hardened gun nose tap as these are the ones I use mainly for stainless.  The tap was held in the mill spindle with a collet holder to keep everything nice and rigid.  Then the tap was brought up to the bolt until the it just made contact then the X axis was zero'd on the DRO.

I chose a slow spindle speed and used Rocol cutting fluid, clearing the swarf from the tap every few revolutions.  After some trial and error I found that driving the tap 1mm into the bolt gave the right balance between depth of the teeth and depth of cut to give a nice radius to grip the filament.

All in all a great result was achieved.  The bolt cuts nicely into the filament without causing distortion and I have been able to run the extruder up to a speed of 600mm/min which is more than enough.  The limiting factor before slipping was the hot end not being able to melt the filament fast enough, not the extruder mechanism.

It is also interesting that the Verbatim ABS filament seems far less brittle than a cheaper 3rd party filament.  Given how long a print can take I think the few extra $$ on a quality filament is worth it

Hole through the jig to to aid in seeing the hobbing depth if a DRO is not fitted.

Now the bolt was machined I was able to finish the assembly of the extruder and tune the feed rate. Proterface was used to send the signals to the extruder which is powered by the #smoothieboard.  Again I cannot speak highly enough of just how easy the Smoothieboard is to use.

Tuning goes like this:

First do some test runs to ensure the tension on the forcing springs are enough, that everything moves as expected and things are tight.  Also check and adjust if required the stepper motor amps and in the firmware.

1. Mark the filament 120mm down from the extruder inlet using some vernier calipers.  The spot you use as a reference on the extruder needs to be repeatable.

2. Send 100mm of filament to the extruder using Pronterface.  I chose a feed rate to 300mm/min as this seemed a good staring point similar to what will be used when printing.

3. Measure the actual amount of filament extruded using the same reference as above and write the number down. Mine came up short only extruding 89mm.   My firmware by default had a value of 140 for extruder_steps_per_mm and it is this that is altered to correct the feed length.

4. Using the calculation (length_sent/actutual_length) x actual _steps_mm gives me (100/89) x 140 = 157.3 steps/mm.

5. Open the config file on the mounted usb using notepad, replace the steps/mm with the new value, save and close notepad and reboot the smoothieboard.

6. repeat steps 1-5 and adjust if still not correct.


An update on the extruder and how well it works.  Since there are still parts to be made for the heated bed and really wanting to print something,  I zero'd the X, Y & Z by send the nozzle down to the floor plate of the printer and adjusting it's height until a piece of paper just slipped between the nozzle and base.  Then the code G92 X0 Y0 Z0 was sent to the #smoothieboard via Pronterface followed my M500 to save the new data.

Next, some green masking tape was stuck to the bed and cleaned with an alcohol wipe.  Using Cura I sliced a smaller version of one of carriages I used in the build and exported the g-code.  Using a print speed of 100mm/sec and 0.4mm layer height, the parts G code was loaded into proterface and pressed print.

Carriages homed, the hot end heated to 240°c and down she went.  To my surprise since I have not done any calibrating or tuning yet it started printing and printing well.  The video is a short slice of the print.  I later printed 2 of the Aria Dragons at a print speed of 150mm/s and had equally good results.  Can't wait to get the printer finished and tuned as I have a feeling it is going to give excellent results.

Should also say that the Infill 3D extruder works brilliantly so thanks guys for providing the files for us to make our own.

Sunday, 4 January 2015

Bowden tube Extruder fabrication

I was thinking of designing my own extruder but rather than reinventing the wheel I opted for one designed by the guys at Infill 3D.  I used this one on thingiverse.  It has a 5:1 gear ratio using herringbone style gears and suits the nema 17 motors I am using for this build.

I printed it out on my UP printing using fine quality and a layer resolution of 0.2mm.  It printed very nicely and I vapor smoothed it to add to the bling factor since it will be mounted on the outside of the build chamber.  I also thought that the red colour look squite nice also

The extruder will be mounted on some 30x30mm aluminium extrusion but the mount that Infill 3D designed did not allow for this so I knocked one up in Inventor and printed it out.  This was also given the vapor smoothing treatment to match the extruder.  It is secured with M6 T-bolts and can be found on thingiverse here

I still need to pick up the bearing to suit the extruder and make the M8 hobbed bolt, then I can calibrate the extruder and give it a whirl.

3D printed extruder bracket front view

3d printed extruder bracket rear view

Infill 3D bowden extruder mounted to the bracket.

5:1 gear ratio Herringbone gear train

Since the extruder is going on the outside I needed a spool holder spindle so I designed one of those also to allow mounting to the 30x30mm extrusion with T bolts.  It was designed to suit the Verbatim filament spools I currently use and also 3rd party spools that are a completely different size.  Since the Verbatim spools are narrower I needed to print some spacers to keep it from sliding back and forth on the spindle.  To stop the spool falling I added a left hand thread and nut to the end of the spindle.  Since the spool unwinds in an anti-clockwise direction the LH thread means the nut cannot come undone accidentally.  The thread is a 3mm deep 60° angle with a 4.5mm pitch.  This spool holder can be found here on Thingiverse

Mounted to the extrusion

left hand thread and nut to suit

spacers fitted to narow filament spools
all ready to go

Thursday, 1 January 2015

Putting it all together

With some time off work it would be a good time to fabricate the frame parts and put it all together.

The BOM so far is as follows:

- 3 lengths of 30x30 aluminium extrusion for the uprights.
- 6 x 12mm linears rod for the carriages to run on
- 12 x 3D printed 12mm rod end supports
- 6 x LM12LUU linear carriage bearings
- 12 x 21mm external circlips to lock the bearings
- 12 x 10mm diameter N42 magnets
- 6 x 10mm diameter x 8mm bore carbon fibre rods
- 16 x 30mm right angle brackets
- 4 x 47mm Nema 17 steppers, 1.8° step, 47
- 3 x  right angle brackets for mounting the steppers
- 6 x T2.5 16T sprockets.
- 6 meters T2.5 polyurethane timing belt with metal reinforcement
- 9 x 625zz bearings for the stepper shaft supports and belt idlers
- 3 x laser cut 6mm plates for the bas, bed and roof
- 1 x custom all metal hot end
- 1 x 3D printed end effector
- 3 x 3D printed linear carriages
- 3 x 3D printed stepper shaft supports
- 3 x 3D printed adjustable belt tensioners
- 1 x 40w heater cartridge
- 1 x ptfe bowden tubs, 2mm ID
- 1 x 5XC Smoothieboard
- 1 x LCD panel for the smoothieboard
- 1 x 5v voltage regulator for the smoothieboard
- 1 x 24vdc power supply
- M4, M5, M6 and M8 bolts, caps screws and nuts
- blue loctite to keep things from coming loose

Still to purchase are the end stops, themistors, extruder drive wheel and extruder nozzle and I still need to print the extruder parts.  I like the 5:1 extruders so will go this way I think.  the extruder that RichRap uses has been proven to work well so a good place to start.

I am still undecided on what type of heater bed I am going to use.  A 240vac version will be much easier and draw less power and I am steering towards a silicone heat pad at present but we will see.

Frame Uprights:

The three 30x30 frame uprights are just under 1,100mm in length and were cut using a mitre saw.

Then, since each end of the extrusion will be bolted to the floor and roof plates I drilled a 7mm hole then tappped an M8 thread into the extrusion.  The important thing here is to ensure there are no burrs as this will cause the extrusion to not sit flat when bolted to the plate  and therefore introduce the  wobble in the frame.  

Floor, Bed and Roof plates:

I opted for 6mm thick aluminium plate for these parts.  The original plan was to drill all the holes using the mill and DRO but sinve my mill only had 250mm in the Y axis it would be inpossible so  purchased the plate and sent it, together with the .step files over to my local laser cutter to cut.  the results were outstanding and given that Delta printers rely on acuracy, this was the most important thing for me.  Cost wise the 3 plates cost $110 to cut but well worth the money.  

Assemebly and drive parts:

Having the base, bed and roof plates laser cut was one of best things I could have done as the accuracacy is second to none.  Once the smooth rods and vertical extrusions were cut to the correct size it was jsut a cose of bolting the peices.

I started by bolting the vertical extrusions to the base plate first, fitting the stepper motor shaft end supports into the extrusion then fitting the 3 stepper motors.  This was much easier to do without the bed plate fitted.  The stepper motor shaft supports have a 625zz pressed into them and a 5mm nut and cap screw which goes through the bed plate.  This allows for adjustment of the bearing centre height to suit teh stepper shaft.  The idea behind the supports is to stop deflection of the stepper motor shaft when tensioning the belts which will stop premature failure of the shafts.

supports with 625zz bearings 

fittind to the alloy extrusion

motor removed after locating the bracket so the cap screws can be tightened

all mounted

stepper shaft captured by the support
Next to be installed was the bad and roof plates, linear shafts and carriages.  The bed plate is supported by six 90° angle brackets which allow for adjustment of the plate.  The roof plate is bolted to the vertical extrusion and has three angle brackets.  The 12mm linear rod passes thought 12mm clearance holes in the plates and are secured with 3D printed mounts that are bolted to the plates.  All in all it make for a very strong and rigid structure.
90° angle brackets

12mm linear rail mounts

bed plate fitted
fitted linear carriage

12mm linear rod before mounts were fitted

  Belts and Tensioners

he belts used as mentioned earlier are T2.5 polyurethane steel reinforced  running on 16T sprockets.  I opted to use sprockets also for the idlers to ensure the belts are not stressed when going around the diameter.  The tensioners were 3D printed with 2 x M5 bolts and nuts fitted for tensioning and an angles slot to allow quick removal of the idle pulley.  Like the stepper shaft supports they are also made to slot into the alluminiun extrusion profile for added rigidity.  The pulley is supported each end with a 625zz bearing so runs very smoothly with little friction. As seen in the picture below there are also 2 M4 cap screws going throught the body of the tensioner to stop the printed part ever delaminating under load.

close up showing the idler pocket

fitted to the roof plate 

2 longer bolts are the belt tensioners.  

The belts were then fitted to the carriages and cut to length.  Since this is a custom build this involved removing the alluminum upright to gain access to the rear of the carriage.  Once done the upright was re-installed and the belt tensioned.  This was repeated for the remaining two towers

Diagonal rods and rod ends

The diagonal rods are made from 10mm OD x 8mm ID carbon fibre tube that was cut to length using a jig to enable all the rods to be the same length.  

The rod ends were made from M8 mild steel bolts.  In the lathe I turned the flats off the bolt head, then using a 10mm ball nose end mill cut the pocket into the bolt.  By using rocol cutting fluid it is possibel to get a near mirror finish on the parts.  To ensure all the pockets were the same depth the bolts were pucher hard against the lathe chuch, and a stop was fitted to the tool pust for teh drill chuck to hit when home.

The treaded portion of the bolt was then cut off leaving a 27mm shaft plus the head.  This provides plenty of metal for the magnets on the effector and carriages to hold onto.  The metal rod ends were then cecured into the carbon fibre rods  with epoxy glue.  To ensure they are all the same length a jig was used consisting of 10mm balls and my mills x axis table and DRO. 

pocket after drilling

end stop for the drill chuck

finished rod end and one inserted in the carbon rod
(not the rod is not cracked, it is just the light reflection)
These were then fitted to the linear carriages and end effecto and finally it is starting to look like a 3D printer.  


For the endstop  am using small mechanical switches bolted so custom mounts with M2.5 bolts.  The mounts are fully adjustable via a bolt that goes throught the roof plate and again slide in the alluminium extrusion.  There is also a bolt to lock the mount to the extrusion when happy with the position..   The limit switch arm is activated by the top of the carriage.  The wring for the swithc is run inside the extrusion keeping things neat

M2.5 cap screws securing the switch
back side showing the switch and extrusion locking bold

mounted to the upright
switch location when fitted

All assembled

All assembled ready for the electroinics

The printer stands 1100m high and is 540mm wide.  Weight wise it feels around 25kg.

Next to my Up Mini printer


After much research I bought a smoothieboard 5XC controller from Robotseed and couldn't be happier.  Admittedly  this was the part of the build I was most worried about especailly after reading some have spent weeks trying to get their printers to work.  Well, my worries were unfounded and the Smoothieboard has surpassed all my expectiaions in build quality and easy of uses.  

The guys at Smoothie have done a great job with the instructions and almost every question I had could be found online.  Also the SD card that is pre-installed in the board has all the files, software like Slic3r, Cura and Pronterface, plus masses of documentaion to cover everything one could imagine.  The user forum located here was also great for getting questions answered fast by the developers. 

The kit I bought had all the conections pre-soldered but the user still needs to install the crimp connectors and fit them to the plugs for your steppers, end stops and other parts.  To be doubly sure there was a good connection I also soldered the wires to them

The installation guide located here covered pretty much everything I needed to know and within a few hours I had the printer up and running.  There is no firmware to muck around with and if anything needs to be changed it is done by opening the config file in a text editor, changing the value, saving the file and rebooting the board.  

For a delta printer a few extra lines are codes need to be pasted into the text file.  This allows the input of the ARM_LENGTH and ARM_RADIUS.  I also altered the max stepper current which I set at 1.3A, although my stepper have a maximum current of 2.5A.  This has not affected their performance and my initial movement tests at 150mm/s proved that. Other settings that need to be changed are steps_mm, speeds for homing, and possibly the motor direction.  

You also have the option of downloading other peoples config files for delta printers and use their config but this would be more useful on generic printers, not custom ones.  Most of these can be found on GitHud.  

Anyway, I wired the motors and enstops and loaded the supplied Windows drivers for the Smoothieboard.  The baord was then connected the board to an external 24vdc psu and the board plugged into my computer USB port.  The best thing to do here is to eject the printer via the "Safely Remove hardware" icon to avoid any hangups.  I found if left mounted Proterface would freeze occasionally.  Then go into device manager and set the USB port speed to 115200 to match the smoothieboard.  

I opened Pronterface and connected to the smoothieboard via com6.  A few settings in Proterface needed to changed like build area etc and this will be different for every printer.  Pressing the console home icon casued the carriages to head upwards until they hit the end stop switches, then they backed away and moved up slowly for a second more accurate position.  I was not happy with the fast homing speed so I redued the fast_homing value in the config file,  saved and rebooted the board.  A second test was much more to my satisfaction.  

Bed height was set by homing the printer with a G28 command,  and loweing the carriage down manually using the pronterface console until it was a sheet of paper thickness off the bed. Then send the command M306 Z0 followed by M500 to save it.   This is not a complete calibration but enough to move the head up and down without crashing and more than sufficient to allow air printing for motion testing etc.  A more complete configuration guide can be found here 

The video below is a motion test at 150mm/s . I printed a pen mount that fits the end effector so I could see what is was doing when printing.  The result was fantastic.


I also have the full function LCD for the smoothieboard that needs to be wired in but I first need to work out which pins need to soldered to the board for it to run.  I have the V1 shield but there is a V2 shield ariving soon.

I will try and print an extruder thisafternoon if all goes well but will need to wait until some thermistors arrive to test the extruder properly.  

Thanks for looking