Print 3D

Edit:06 févr. 2022, Cre:15 févr. 2016

Lily Delta printer


I designed this enclosed delta printer with OpenSCAD using my own Delta Simulator as engine.

Based on my experiences on the Fisher delta and the D-Box, I designed and built early 2016 a new delta printer, the ‘Lily’ using the Fisher delta components. It was designed to be used with the Fisher mechanical kit, but with RRP closure, I thought it did lost some interest and did not publicised it but I recently discovered (early 2021) that the Fisher is surprisingly still sold by another company! So, reusing Fisher components to build a Lily is still a good option. I bet that many Fishers were abandoned, their components may see a new life.

The really usable space of the Lily version I built is diameter 170mm, 242mm height at center and 223 mm height at 170mm diameter. Base size is 320×370mm, with 500mm between top and bottom plates. That is roughly twice the really usable volume of the Fisher and the same as a Kossel. How did I obtain a much bigger volume with the same columns? With two simple tricks: I installed the columns elevated on a stand, which is ok because the lower part of the column are not used while printing. Also, the columns don’t go up to the top on the motors, there is no need as the structure is the box, not the columns which are only used as guides, the motor is supported by the wood box. Do not forget also if you install a heat bed to rise the column support by the same amount as your bed height to preserve printing height.

However, with the exact same parts (Actual bed plate is 200mm diameter), you could make another variant with 186mm usable diameter and a reduced height of 230mm at center and 208mm at 186mm diameter, but with a base increased to 340×380mm. I never printed at full height while sometimes I print parts a bit too large to fit on the actual Lily. This larger diameter but lowered usable height is what I now recommends, more usable for most people. With a non dismountable enclosure, being a bit larger may also somewhat ease maintenance. It is more accessible than a Fisher and the front door helps, but this is still not very large. Also, for safety, it is better to use fireproof panels (M1). I only found those panels available in 15mm thikness in the local DIY stores (in France). The actual Lily structure was done with 18mm Melaminated parts for structure and 10mm plywood for doors and non structural), so this will be a bit more touchy for assembly (need pre-drilling and good quality screw).

Why designing and building this printer as I already had two working printers?

My aims to replace the Fisher were to:

  • Have a more silent printer
  • Have a part fan controlled by the board
  • Got rid of all acrylic and fragile parts
  • Few mechanical adjustement
  • Being capable to enclose the printer later (was done just a few month later)
  • Enclosed version integrates an odor filtration system (recycling through activated carbon mat - from vacuum cleaners filters)
  • Increase the printable dimensions (nearly twice the Fisher real volume)
  • Install a heated bed
  • Install a power On/Off switch

Threaded rods are used to adjust the triangle and give good dimensional accuracy. After assembly, they do not really participate to the structure which is the wood box. Bottom rods were cut when I installed the heated bed. Top rods support s the wire harness.

Control board is now installed in a closed compartment on the left side of the printer

It is using the Spool on top support

Extruder installed to ease access and minimize bowden tube length (420 mm) The extruder is the Geared extruder which was designed for the Fisher Delta, mirrored

Kinematic effector for safety and probing

I have designed the effector like on the D-Box, with the kinematic positioning system on the effector. hotend is positioned on effector through a very precise kinematic system. This increase globally the effector weight by 15g (including 6mm balls), which is compensated by the lighter arms. I think the fact that I never broke the hotend which is still the original Fisher one is largely worth the extra-weight. It shall be understood that due to their particular geometry Delta hotend crash on the bed much more frequently than with cartesian printers and some ‘safety’ device is required. Now a lot of deltas are using magnetic ball arms which ‘pop-out’ in crashes but his is a heavy solution.The system is also used as a level sensor. You touch the bed with the hotend, that raised the system out of its support, the electric circuit is cut which detect the contact. This is very precise, repetable and reliable. This part is one of the best feature of this printer though as the rubber band stay in place and it is not using magnets as on the D-Box, this is less helpful for maintanance as you cannot remove easily the hotend ensemble. I am surprised that this simple system is not more widely used on 3D printers. I devised the system on the D-Box and published it more than six years ago and it proofed to be an simple and efficient feature.

Simple and light arms with steel(!) threaded rods

For the arms, I have not changed the winning team and they are made with M3 threaded rods, just longer as the printer is larger. This is relatively light and surprisingly strong. I was capable to lift the printer (8kg naked before it was enclosed) by its effector (see photo). Don’t even try that with a Fisher, even if its weight is less than half that, you’ll pulverised the fragile acrylic arms. Please note that you can use this effector and carriages on a Fisher, without diameter loss, the height loss being compensated by the bed removal. The big advantage is that the Fisher bottom plate may be much more flat than the Fisher bed, a warped bed being very troublesome. The rod space was modified to be identical to the more recent Fisher, but that could be modified, as this is a parameter in the OpenSCAD program.

Ball in cup arms with nylon tensioned wires

The arms articulations are made as on the D-Box, with balls in printed cups, maintained with tensioning wires. I was initially thinking that there was no need for very high tension, so I used nylon wire instead of high grade Dyneema, and this is simpler to install and with elasticity of nylon, there is no need for springs. Tension is just made by adding knots on the wires. However, further tests showed that good tension creates friction in the ball cups which dampens vibration and have totally eliminated the banding I had initially on large diameter parts, so I increased the nylon wire size to 16kg (35 lb) and this is now running without failure since five years… It seems that on deltas, some sort of articulation damping is required.

No end stops

There is no end stops for simplicity and reliability (while they could be installed outside the box actuated by a rod screwed on carriage). I stall the motors on the top support. That makes quite a bad noise while stalling and creates rebounds, but precision is not that bad if you use a not too low current. As the bed is probed immediately after, there is no deal. Enclosing the printer significantly reduced the stalling noise. To do so, you need a board with digital potentiometer where the stepper current can be adjusted by a g-code. On recent quality boards, you have stall detection which may give better results than my primitive system, but the Duet 0.6 was designed in 2014 with the traditional A4982 drivers of these times. However then very few boards had digital pot for stepper current adjustment which may have been initiated by the Smoothieboard in 2013.

Simple geometric adjustments during construction

The difficulty of delta printer is accurate geometry positioning, so I use threaded rods to make the top and bottom triangles. On the bottom, the rods shall be cut (with a dremel tool) after assembly (at least one to use the maximum diameter without conflict with the fan part duct).

The assembly is faster than the Fisher, but you have to do some woodwork before (quite easy) and modify and complete the wiring (fairly tedious though not using endstops helps). You shall Buy 2 new motor cables as the original are too short.

Parts printed in PETG

Plastic parts were printed in PETG, and so, all components have been significantly reinforced compared to the Fisher, because the PETG stiffness is just above half of the PLA stiffness. I now recommend to print the structural parts, especially the motor support in ABS, this is better for an enclosed printer wich is used at 35 to 45°C. With the availability of PLA which could be annealed without part shrinking (by example the FormFutura Volcano), building it in annealed PLA will make it more stiff and globally better.


I have used two high quality Sunon 25×25×10 fans (5m3/h),supplied in 5V, which are extremely silent. They are not that powerful, but that seems to work. Hotend cooling does work ok with this low flow, this was the deal compared to the high flowrate of the large radial blower of the Fisher. Hotend plastic parts are more exposed to heat (insulation is done with a plumbing fiber gasket) but yet, they had no problem, while I routinely print at 270–280°C (PETG). Having a separated part cooling help a lot, while it is a bit weak and asymetric. This is not the best for PLA printing, however in an enclosed printer getting warm, being capable to cool parts at very low flow (15 to 25% of this low flow fan) proofed to be very helpful to increase the print quality with ABS and PETG. Yes, it helps (a lot) to cool ABS! but this is with an ambient temperature of 40~45°C.

5V fan installation

This fan setup works, however you shall be very cautious while wiring them. Check and recheck polarity on the controlled fan, a mistake will burn the board (sending 19V on the 5V input cannot be nice). It is safer to use 12V fans, but it needs a DC/DC converter. That DC/DC converter is more or less required for the fumes recycling fan, anyway, so that is now what I recommend, in case of trouble, only the 12V power supply is exposed without risking to burn the board. And 12V is already available if you install a heated bed, which I now considered required.

A tiny hotend in an enclosure?

The Fisher hotend is an ‘all-metal’ hotend with a very small finned heatsink, less than half the area of other all-metal hotend. On the Fisher, this hotend was cooled by a radial fan capable of much higher flow than the commonly used axial fans, so this setup was working well, and the Fisher is an open printer. On the Lily, because I added a part cooling fan and the kinematic system, I had no sufficient room for this bulky, noisy and heavy radial fan. I even had no room for a 30mm fan, so I setup a hotend cooling axial fan of 25mm. In addition, as one of my purpose was to reduce the noise, I chose an industrial grade ‘Sunon’ fan which does have a very limited flow compared to other high speed fan sirens which are used on some hotend (like the one I was using on the Prometheus hotend on the D-Box). In addition, the inlet of the fan is somewhat obstructed by the kinematic system. That was a recipe for failure. My ‘plan B’ was to overvolt the fan if needed. Though, I did not had to rely on this ‘plan B’ (On this printer). How come that a very small heatsink with less than half the surface of other all-metal hotend of the day (E3D V6 was the standard), with very limited cooling flow, running in an enclosure at a temperature between 35 and 45°C can work? There were no Titanium of bimetallic heatbreak in those days. That was a bold design and with my today experience on my other printer (the D-Box), I won’t redo such setup. But on the Lily that works without any problem! How come? The only reasonnable explanation is that the place was so constrained that I ducted properly the flow around the heatsink and particularly that the exhaust flow is well ducted which is extremely rare, most setup having only a duct on the inlet. This is not perfect though because if the hotend ducting is nicely done, the part fan ducting assymetry is far from ideal for part cooling, see above paragraph. Today, I will prefer Titanium or bimetallic heatbreak, which will allow higher chamber temperature (but that need also structure redesign to cope with the higher temperature). Note that the hotend fan failed once and I replaced it by the most powerful Sunon 25mm fan available (0.4W), a bit more than the original one and it is a bit better (and a bit noisier…).

Chamber temperature

As I had a serious incident while testing a chamber heating system on the D-Box (trying to set this wooden printer on fire), I preferred not to install any controlled heater, out of fear. Instead, I use a 60W, 230V Halogen lamp as a primitive heater, along the bed heater, that I start and stop manually with a simple switch. With a relatively small volume, the chamber climb relatively quicly in temperature and you may preheat (also using the bed heater) 20 to 30min before the print if you want to print ABS. For PETG printing, I start the machine then start the preheating/homing procedure which takes maybe 5 min, then I print. The small size of the printer which makes its external surface ratio to power higher than for bigger printer, so natural cooling is sufficient for temperature control. This printer stabilize in temperature and does not tend to build temperature over time like the D-Box. A long PETG print (bed at 78°C) began to stabilze after half an hour and will be stable around 45°C after an hour or so and I shutdown the halogen lamp around 35/40°C. However, the motors being inside the enclosure, I am cautious to not exceed 50°C. For printing ABS, you may be prudent and limit printing in a hot summer if you don’t have an air conditioner. The Fumes recycling is efficient to remove odor but you may not want to open the front door when printing ABS, it is an infection. If the printer began to overheat, opening the fume recycling compartment helps and you are behind the odor filter. It is important to note that as the printer is nearly tight and don’t expose odors, any serious incident will only be noticed late, that is why I didn’t noticed immediately the burn smell in my D-Box during the incident until fumes began to exit the printer under pressure! More recent Duet boards allow chamber temperature control but the old Duet 0.6 don’t have any spare thermistor input, so you rely on manual checking - once again, never left a printer unattended, how much do you value your home or your life?

Back to the basics for the spool support

I simplified the already designed spool on top system, with no bearings neither any mechanical part. Its works better ! However, with experience, I think vertical axis spool holder, while more elegant than spigot supports, are less practical as they tend to create unwinding problems while the filament loops fall by gravity and tend to cross and make knots, even with a rotation brake. So, the simple spigot support, even if less aesthetic, will work better and is faster to print. Swapping sides or extruder compared to my prototype (don’t forget to mirror the extruder) may allow to install spool on the printer left side instead of the shelf bracket on top. As there is no frequent need to access this side (electronic), this is not a problem.

Powerful and reliable extruder

I improved a bit the already designed geared extruder, with stiffer bearing holders. As I already experienced on the Fisher, a good position of the extruder is fairly difficult to establish. I tested 6 positions before settling on the final one. But while the printed volume is higher, this allow to use the original Bowden tube of the Fisher (I have to lengthen mine, which was reduced with my spool on top installation, but this extra-length is the original cutted part…). As the stepper is inside the enclosure, you shall limit the extruder stepper current, typically the highest in a printer, so a geared extruder is imperative. using the standard Fisher steppers, I set the extruder stepper current to 1A (was 1.2A+ when direct drive).

Mods and improvements

There was over time, two revamping, the first was to enclose the initial version, then I added a heat bed in 12V, with a second power supply, both power supplies being now installed in an added socle. Both modifications added a lot of weight and it is now fairly heavy for a printer that size.

Does it work well?

With increased tensioning in wires, I don’t see significant differences either in speed or quality with the Fisher. As for the noise, there is a huge improvement, which was one of my main purpose. Indeed, even in its first open version (just the three panels, see photos), it was a bit less noisy than the enclosed D-Box and way more silent than the Fisher.

Being relatively small, the internal temperature stabilise reasonably quickly without any control system. Now it is equipped with a bed heater and at maximum bed temperature, internal temperature stabilised at around 45°C for PETG printing, which is ok for the steppers (which are inside the enclosure). With steppers inside and parts in PETG, you shall not exceed 50°C as PETG proofed to be a lot less temperature resistant than ABS.

This, added to lower noise, make it my preferred printer for parts which can fit in the volume, even if for some prints it is slower than the D-Box due to the lower flowrate of its small hotend (only for the thicker layers 03~0.32mm). The heated bed and a really flat glass bed have improved a lot the ease of operation as now the first layer thickness is constant and repeatable and there is no longer any first layer adhesion problems with PLA or PETG. I no longer check closely first layer deposit, just computer start and that’s it! Just check after ten/fifteen minutes if everything ok. Though without automatic nozzle brushing as on the D-Box, a check of loose filament deposit is recommended, my first layer is not always perfectly clean because of that. Hotend working well in a very difficult context, I will not modify it (I bought a spare hotend years ago -hotend/heatbreak are an integral part-, still in a box).

What are the drawbacks and compromises ?

  • There is a bit more than twice the weight in plastic compared to the Fisher (total ~350 g), which translate in a bit less than twice the printing time (parts are globally more simple). They are printable on the Fisher, however I printed them on the D-Box, which is faster in PETG because of its hotend and controlled cooling, mostly with 0.3mm layers.

Total printing time ~16h on the D-box, maybe 20h on the Fisher.

  • The initial fixation of the hotend on its support was improved and better cooling flow reduced the bottom hotend temperature to around 52°C with hotend temperature at 280°C. Screw temperature is approximately 10°C below low fin temperature, so this is quite acceptable for PETG support.
  • For weight and dimensions reasons, I have not used magnets to maintain the hotend like on the D-Box, instead I used rubber bands cuts in bike tire tubes. This is less practical to install, but I find a reasonably easy way to do it (installing wires tu pull the rubber bands). So, in case of mechanical trouble, that is not the hotend which is ejected, but the arms which pops out (I tested involuntarily during commissioning…). The rubber tend to age and I had to replace the rubber bands twice. This is the only maintenance, aside rod oiling, that I had to do over time. I never had any nylon wire break (that will ruin the print and can be dangerous).
  • You have to order a few stuff (2 fans, a diode, cables and connectors, some M3 screws, a larger buildtak, 25cm of bowden tube for filament inlet, a DC/DC converter for fumes recycling fan) and buy parts in DIY stores (wood, to be cut by the store for accuracy, M4 bolts and nuts, wood screws, M5 rods, M3 rods, rubber supports, nylon wire, aluminium sheet for board shielding, windows seals, activated carbon mattress), for a total which may not exceed 100 euros for an enclosed printer.
  • Alas, like on the Fisher, you will not escape to do some manual calibration as the offset on the calibration points is not constant, because on the center, you are using 3 belts and near columns, mainly one belt is used, so the offset is larger on the periphery (this is in reverse of the Fisher, where offset is larger at center due to spring loads). But with recent version of DC42 fork, it is a bit easier to do, but a procedure have to be written.
  • The effector plug installation (from the Fisher) is really messy. On the D-box, I wired directly to the board and find it much more simple, with minimal drawback as the route is direct and so wire replacement easy. The best installation may be to use a sleeve instead of zip-ties to group independent cables, so removing/adding a cable is faster than redoing the zip-ties.
  • Ah, and printing on melaminated chip board did not work, even with glue. So I installed a ‘BuildTak’ surface, efficient but quite fragile. Install a heated bed, this is much easier (and participate to heat the chamber).
  • I installed a heated bed in 12V, so I have to add another power supply which is installed (along the power brick) in an added stand. The Bed heater is a kapton heater of 160mm diameter (as used for Kossel deltas), so much smaller than the bed plate (200mm diameter) but with thick aluminium diffuser and bottom insulation with ceramic panel (chimney safety stuff), this proved to be effective and of sufficient power. The maximum temperature you can reach within reasonable time is 130°C. (preheating the chamber with the Halogen lamp will help). At such heat bed temperature, caution shall be taken to avoid chamber overheat (do not exceed 50°C) and the Halogen must be shut down to avoid heat build-up.

What was improved/modified during development and use

I had some experience and already built the D-Box, so I did’nt do much adjustement

  • I reworked a bit the part duct to have it more enveloping, as its assymetry don’t give the best cooling
  • I tested no less than 6 positions for the extruder before finally setting to the actual one, giving the shortest bowden and less tortured during printing
  • Increased wire tension to improve damping
  • I finally abandoned all these fancy vertical axis spool support as they don’t work well, the filament tend to tangle. I installed a very simple spigot support above the printer.
  • I changed the arm spacing to have the effector compatible with the new Fisher version.
  • I added the complete enclosure rapidly with the fumes recycling (use a voltage converter for the recycling fan)
  • I added a heated bed also relatively rapidly with own 12V power supply
  • Installed a foldable steel handle on top, with complete enclosure, this printer is heavy.
  • Added an halogen lamp (60W, 230V) as a primitive chamber heater
  • Intalled a traditional (alcohol) thermometer in the chamber (Was previously measuring with a portable IR thermometer). The old Duet 0.6 board doesn’t have spare input for a third thermistor.
  • Changed Bowden tube for a capricorn one, I amnot really convinced and this stuff is opaque, so less practical.
  • Rubber band used as spring were replace by a printed ‘anchor’, with ‘legs’ locked under the effector and the nozzle maintained by a top screw. This is relatively sensitive to printer temperature but very easy to adjust and this is easy to remove to allow hotend maintenance.

Now in early 2022, how did this printed fared over time.

A quick assessment

I had to do some maintenance recently, but this machine is still alive and well. As it is much more silent than my D-Box and it does heat more quickly, I prefer to use it when possible. Its maximum throughput in PETG is lower than the D-Box because of the shorter hotend heating zone but the geared extruder is quite powerful and this helps. I cannot really estimate how many kilos it did printed but I use it regularly. There was a hiatus of around 1 year and an half in use, but it is now back. What maintenance I had to do ?

  • Thermistor failed once after three years of use but before failing it had became fairly unreliable and the printer was no longer trustable.
  • The 25mm Sunon fan (5V) cooling the hotend failed once. I replaced it with the most powerful 25mm Sunon fan I found (0.4W), a bit noisier but nothing like the sirens you can find on some other hotends as the one I had on the Prometheus in the D-Box.
  • I had to oil the sliding bearings many times
  • I replaced the ‘bowden’ tube by a ‘Capricorn’ one. I am not convinced that there is improvement and the fact the the bowden is no longer transparent is less practical.
  • I had to retensioned the belt once
  • The motor supports did somewhat bent over time. I overtight the bottom screw which somewhat improved the situation but the motors are still a bit reclined, which is not good for the belt and overall friction. A solution will be to install shims (as I had to do on the D-Box for the same reason).
  • The rubber bands used as spring for the kinematic system have been replaced three times (not super easy to do, insert wires to pull them on their anchors). They don’t last in the ambiance. They are done with road bike tire tubes, mountain bike tubes are way too large… This system is now abandoned and replaced by a printed part, see above.
  • There are frequently bad mechanical noises and some extra-friction of an unidentified origin. I grease regularly, so this may be sliding bearings wear or some troubles in the motor bearing.

What is good on this printer ?

  • It is enclosed
  • The very special hotend of the Fisher proved to be remarkable and working properly with a limited airflow within an enclosure going up to 50°C. This is quite impressive and I have more problems with the D-Box hotend which have more finned area and a much larger cooling flow.
  • The odor recycling system is better than on the D-Box because it have more surface (the larger, the better)
  • The geared extruder is powerful and had very low slip, better than other single drive extruder that I know. To have a better extruder, you have to go dual drive (this is what I ultimately did on the D-Box).
  • The glass bed is super-flat, while the Fisher bed was a nigthmare to tune because it was completely warped. I expensed tenth of hours to try to tune the Fisher because of that. Now the firmware can map the bed but this was not the case in 2015 when I bought the Fisher.
  • Enclosure heat-up is relatively fast due to small volume
  • Enclosure temperature stabilize naturally (without any control system)
  • Kinematic system is efficient for bed probing, while its was tedious to tune. You need to check the tuning when changing the ‘springs’ (the rubber bands). It also preserved the hotend heatbreak that I never broke (I bought one in spare, still in a box).
  • The wire tensioned rods with ball in cups is much lighter than magnetic ball rods and it also can ‘pop-out’ in case of serious mechanical trouble (tested - unvoluntarily).
  • The wire tension system provide damping and this is somewhat adjustable (note that on the Fisher, this is the ball pinching which provide this damping)
  • To my surprise the nylon fishing wires used for tensioning never fails and they are much more practical than on the D-box as the wire itself is used as the spring.
  • The nearly tight enclosure (I gasketed the doors with windows foam) will be safer in case of fire because as there is not much possibilities for air inlet, fire will stop by oxygen depletion
  • Duet Board (even this 6 years old 0.6 version) is still one of the best board out there, far above anything else. What is really missing on this old board is silent drivers. Developement of firmware on this board is now stopped but it was still updated till early 2020, nice!

Newly sold Fishers are supplied with Duet Maestro, equipped with silent drivers and capable to have a second extruder.

  • Due to low size steppers of this machine, the board need no forced cooling, natural cooling is sufficient (see photos for the openings).

What could have been better.

  • I used 5V fans with protection with diode but as I installed later a heated bed with a 12V power supply, it would have been better to use 12V instead. But this have been incremental update.
  • PETG proved to be much more sensitive than ABS to overheat. In a chamber, I think it is better to print parts in ABS, especially the motor support and effector.
  • The heating cartridge is way too powerful for the small hotend block (more than 50W). New firmware versions are handling the control very well but initially, that was very delicate to tune. This excessive power is a safety concern, indeed the new Duet autotune told me that it was too powerful and capable to heat up to 900°C, way over aluminium melting temperature. It seems that on the new Fisher, the cartridge power was drastically reduced as now the power only supply 60W (Fisher beta versions power bricks were 120W)
  • Glass is not the best surface to use for adhesion, you need lacq. I prefer special surfaces I use on the D-Box or on the Tronxy X5S. I depolished the surface with a diamond sanding block, which somewhat helped.
  • Part cooling system is asymetric and a bit weak (fan 25mm). Not the best for PLA but I mainly print PETG.
  • Due to the elevated columns, it is not possible for the effector to go outside the printing area and so, I cannot install a cleaning brush aside the bed as I did on the D-box (also because the bed is much larger than the printable area). I miss this automatic cleaning.
  • As the kinematic top part is maintained by rubbers and not by magnets as on the D-Box (to minimize weight), maintenance is less easy, you can’t extract the head, you shall remove the arms for easy access.
  • The heat block being smaller than others and of special shape, there is no adapted silicon socks. In addition, the nozzle protrude of a very low value under the heatblock, so there is not even room for a cotton insulator. For my setup, an insulation of the heating block is required, the only solution being to use kapton tape around the block, but this is a bit messy and doest’t hold very well on the heat block.
  • On the hotend side, the bowden tube is screwed in the heatsink (threaded M4). This is robust but you need to ‘thread’ (with a common nut) the tube end and carefully drill it to 2mm (beware of remaining PTFE chips which may block your nozzle). When tightening the tube in the heatsink, it occurs frequently that this pinch the tube and block the filament. Generally I tighten to the maximum and untighten a half-turn to avoid that.
  • The bowden is attached on the extruder with a brass end with a side locker. Generally, this is more practical that the pneumatic fittings but if the lock is not well install, the bowden displacement may creates artifacts on the print after angles if advance pressure is activated that you may falsely attribute to extrusion problems.
  • As my D-box and a lot of Duet controlled printers, it is headless (no local control), what I miss is pause/resume, filament loading and unloading buttons. This will cover more than 90% of the local operations and this is doable with existing board.
  • Having designed and built two deltas, I now think they are not the best solution for a 3D printer because:
    • They are more delicate to adjust and tune
    • If movement goes wrong (something locked, forgot print on the bed), it may go very wrong and youi can broke arms or something on the hotend side, though nowadays a lot of delta have ‘popping-out’ arms which may save stuff.
    • Due to the inherently faster carriage speed and acceleration linked to delta geometry and all other things equals, Delta printers are noisier than cartesian printer. This is, in the long term, why they will be abandoned. The advantage of well built deltas in high acceleration doesn’t make it. I much prefer to have a longer print if the printer is more silent (this is why I always choose the Lily over the D-box if part fits in, while the D-box is slightly faster for high thickness layers). In the future, I’ll go for CoreXY. The X5S, while verly poorly designed in many critical areas, proofed me that this is a better option and you can build a whole aluminium printer without needing machining. But I will continue to built closed printers with structural wooden boxes, though with fire-resistant panels. These boxes are rigid and stable even when heated and easy to build. I will continue to use Duet boards, even if they are a bit costly. Hey, my Duet 0.6 is nearly 6 years old now and still a very good board, just missing the ‘silent’ drivers…

Maintenance to do

  • Brush the nozzle (with brass brush) every few prints - Nozzle is in stainless steel much harder than brass
  • Don’t forget to oil regularly the sliding bearing (~every three month)
  • Remove plastic chunks and clean the box every three month
  • Check belt tension every one or two years.
  • Have a look at the motor support deflexion every one or two years
  • Change the effector rubber bands every year
  • Check firmware evolution every year and update if interesting
  • Change activated carbon mattress every year

Future mods

  • Adding a filament presence detector as I did on the D-Box, doubling with a pause button (On the D-box, they share the same connexion).

Due to improvement in the firmware, I Recently added multifunctions to the pause button, which will start a preheating macro when printer is not printing, pause being only used during a print. I wrote the multifunctions macro a few years ago but they were not working for unknown reasons, while they are now with more recent firmware.

  • Adding other buttons for resuming print, Unloading and reloading filament. As I don’t use endstops switches (homing by stalling steppers on mechanical stops), I have three connexions easily available and other inputs are available on the main connector but connectors dislodge too easily.

It shall be noted that in principle, end switches are only used when homing so you can parallel button switches with the end switches for other uses, reallocating the purpose depending what is the current task (use start.g and stop.g for usage reallocation).


OpenSCAD program, STL files, assembly manual, configuration files, photos are available at :

Photo Gallery (189 photos)


  • Tim Jacobsen, for the rods in printed cups with wire tensioning system.
  • RepRapPro company, for the idea of kinematic positioning system used as a sensor, but here I set it on the effector instead of the bed.
(c) Pierre ROUZEAU
Privacy - Vie privée - Imprimable - Rechercher
Page mise à jour le 06/02/2022 01:35