The true story of the Smart Orbiter v3

Orbiter 3.22 Smart LDO Toolboard v1 v172

This is the true story of the Smart Orbiter v3 design.

I admit its a little too technical but face it people! A good design involves deep technic gizmos only a genius can understand and you!

A big thanks for Jason form LDO who made this project possible. And all the LDO people who worked hard on building samples and helped making this project a reality.

Last but not least my full gratitude to the girls & guys over E3D for their great innovative spirit and motivation inspiring the development of the SO3 Revo version (coming soon).

Font heatsink

Started the design by defining a cool front aluminum heatsink. Initially I wanted an add-on solution for the Orbiter v2 but I did not really liked the initial outcome...so to be clear it was ugly looking this is also why I do not show it here.

I realized that trough a heat-break there is not much of heat escaping and most of the issues we have with current hot-end clogging is not because of the fan or hot-end cooling but because of bad cooling airflow design.

So, to make the extruder as compact as possible I adapted a frameless fan design which is integrated inside the heatsink.

I know I know, a blower type fan would do a better cooling job, but I sticked to what I could purchase as ready-made fan. We have not designed here a custom fan. In the end as mentioned before it does not need much cooling. Thermal simulations and real measurement showed that this fan is more than enough, I actually run it on 70% only to reduce the printer noise.

Thermal simulations….yes I optimized the heatsink shape using finite element thermal simulation, special thanks to my colleague Gheorghe Caragea who is an expert in these gizmos by his actual job. He taught me all the tips and tricks how to setup the simulation correctly. I still owe him a few bears for that...

Result speaks for itself, heatsink temperature is about 48deg C having the hot-end at 250 with fan actually turned OFF.

Bimetallic heat-break design

The heat-break consist of a steel tube with 0.25mm wall thickness and extremely smooth finish.

The bottom side is standard V6 compatible piece of copper. Copper to have the best heat transfer performance form the hot-end heating core.

On the other end we have a copper ring. Now why that copper ring?

The first version of the SO3 I designed without that copper ring. Just a 2.5mm hole, the steel tube was inserted directly in the heatsink.

I basically I used a heat-break from TL Matrix extruder, shortened to the length needed for SO3.

You need to understand a bit of thermodynamics here and how heat is transferred from one material to other.

So, heat travels by atoms of the different materials “smashing” into each other. As the contact surface of two parts cannot be made perfect there will be always tiny gaps of air trapped in between the microscopic spaces creating a thermal insulation, thermal contact resistance. Surely this can be improved by assembling with a good thermal conducting paste.

Thermal resistance between two materials depends on the thermal contact surface area. If you increase the surface area even with bad thermal contact the overall thermal resistance decreases.

For the heat-break design variants, the surface area contacts between:

- Steel tube directly inserted in the aluminum heatsink -> 65mm²

- Steel tube with copper ring in the aluminum heatsink - > 122mm²

The contact surface area is doubled reducing thermal resistance to half. The copper ring is press fit onto the steel tube leaving no airgaps making almost perfect thermal contact with negligible resistance.

Another positive effect we get from the copper ring is conduction of the heat to a larger surface area more quickly, from which is taken over by the aluminum heatsink. This idea is actually not new, it is used since more than twenty years ago in CPU coolers.

The result is awesome heat-break performance! You can see in the next pictures the extreme thermal gradient along the heat-break length. You can see in the plot that the heat goes from 350°C down to about 46°C within 2.6mm distance which is just slightly longer than the heat-break length of 1.7mm.

You might think now yeah, yeah we've seen this in other hot-ends as well, fair enough, its not my invention...but did they shown you how it really works and what performance they achieve?

Hot-end design

As first trial I designed a hot-end using a standard 6mm heater cartridge and volcano nozzle. Design optimized of course with thermal simulations.

The design was pretty simple, three parallel holes, inside a machined copper core. Middle hole was taped with M6 thread. Two side holes one for heater cartridge and second for temperature sensor on the opposite side. Overall, the design was working to well. Heat was transferred so efficient that melted filament was pulled out by gravity causing some over extrusion artefacts on slow printed fine detailed areas, which I did not liked much.

In the meantime, Bondtech launched their high flow nozzle, so I changed gizmo tactics to ceramic ring heater with shorter melting path and Bondtech CHT nozzle. First concept sample with CHC heater core from TL. Concept being proven I designed together with Phaetus a new 72W heater core and heat-break which is now mounted on the SO3.

The result is higher flow compared to the volcano nozzle version, less drippy behavior, and half the weight.

Volumetric flow for different hot-end types.

Volumetric flow for different nozzle size and type.

Note: flow depends on the filament property as well. Some leading higher or lower volumetric flow for the same conditions.

This results led to the obvious choice for best weight vs flow performance to ceramic heater ring heater plus Bondtech high flow CHT nozzles.

Now I hear all the Facebook engineers...juuust 72W heater? Well yes, just 72W. It is more than enough and here is why:

Aspect 1: heating time is directly proportional to the heating power and inverse proportional with heated mass. Since the SO3 heat-block is a small ring of copper its mass is pretty low compared to volcano or supervolcano heat-blocks, therefore it requires way less energy to heat it up in the same amount of time.

Aspect 2: Heat required to heat up ABS from 25°C to 250°C needed to melt ABS material with 20mm extrusion speed meaning 48mmc meaning 480mm/s with 0.25 x 0.4mm layer dimension is about 22W

Same calculation with PLA heat up from 25°C to 200°C needed melting with 20mm extrusion speed meaning 48mmc meaning 480mm/s with 0.25 x 0.4mm layer dimension is about 14W.

We lose about 5W + about 5W radiated heat we actually require about 24-32W of power at insane print speeds. Therefore in this application the 72W is more than enough.

Wire strain release for the heat-block wires is essential. Obviously one concern I had over the straight strain release stainless steel piece attached to the hot-end (similar to CHC) was its heat conductivity. Indeed, thermal simulation showed a straight piece of metal will have its end where the wires are attached about 100°C.

This led me to this optimized shape for worse thermal conduction. The cold end to this strain release is at about ~57°C (heat-block @ 350°C) and less heat loss from the heat-block.

Quick change nozzle

You may observed I have not included any one hand change nozzle features, I personally do not like those kinds of designs, because of too much heat escaping trough those fixing structures which transfers more heat into the heatsink worsening its high temperature gradient performance. Remember one heat-break transfers about 5W of power...four of them would transfer a little more...sure not four times more because the worse thermal contact but still is not negligible plus its someone's "my precioussss" idea and not for free.

For fast nozzle change I recommend to change the complete hot-end assembly which is held in place by one screw accessible from the extruder front side. The hot-end wires have a Molex connector for quick change. I know its not the cheapest solution but performance wise is the best.

E3D Revo ecosystem

In the summer of 2023 E3D contacted me via LDO motors. Really exciting times, they liked my SO3 design and proposed to design a version using their Revo ecosystem. Since the heat-break design was already finished and could not be made compatible to the Revo system, the simplest solution was to slightly change the heatsink part to accept quick change Revo nozzles. This means a second variant with different front heatsink.

I think this its simply awesome!

Tensioning mechanism

Depside looking like a pretty simple solution, this was for me the hardest part to design. Took me over one year to get a working design and about six months just to come up with the concept. I had lots of ideas but none of them were good enough. Yes, I did asked mum, wife lovely kids they had no clue what I'm talking about!

Many people find it hard to get the right tensioning amount for extruders based on a screw and spring tensioning like the Orbiter v2. I wanted a tensioning mechanism with fixed tension levels like Bondtech LGX series, adjustable from the top side of the extruder but with “springy”, elastic behavior not hard levels like LGX extruders.

My first idea was to use a “rosette” knob with six tension levels which pushes an elastic lever connected with the idler gear.

The idea was looking awesome, with six fixed tension levels, I taught it will be the solution, but the devil had other plans.

During the testing phase I find it extremely hard to engage level five and six, which should be the level for most hard materials.

When I tested the Bondtech LGX, I reached the conclusion that there is no practical use for six or more tension levels , I always used just two of them. Low tension for soft and max tension for hard filament. (ups...yes...looked into competitor designs for stealing ideas...what? everybody does it, it's called between professionals benchmarking!...).

Therefore I changed it to a lever design with just three levels OPEN-L1-L2 with a longer arm which makes tension level change much easier and 100% repeatable.

Actually this part was the reason of the Orbiter v3 samples delivery delay just after we announced it.

LDO has manufactured several hundreds of SO3 parts and in the last moment just before shipping the first batch one of our friends who got an earl sample for testing showed us a picture with a bent tensioning arm.

Surely enough this was a major issue, and we stopped the shipment of the SO3's until we figure out what the issue was.

It turned out the part was manufactured out of some regular steel not springy enough for this application and not made out of the material I used during the design validation.

393862385_1306764379974987_707271977227379719_n

LDO ordered some new parts manufactured out of spring steel (65MN) and sent me 3 parts for testing. I assembled than on my extruder samples and I hit another setback. One of the tension blades broke in two.

One of the tension blades samples broke in two.

I was not sure what was the reason, do I overstress the material and is not fitting to this application and we need to find another type of material or something is wrong with the design.

I asked one of my colleagues who happens to be an expert in structural simulations if he can help me out with some simulations..

Below you can see the simulation results for a displacement of 2.5mm. The stress in the material does not exceed the material specification (920Mpa).

rn_image_picker_lib_temp_0554d442-9d70-49b2-bc2a-7793dbffb0f2

Since one one blade broke out of all the samples we made and simulation proved we are not out of spek even with displacement of 2.5mm, (in reality we have less) I personally concluded that there was something wrong with the material of that blade not fulfilling its spek.

Nevertheless we decided to try and simulate with a second material as well stainless steel 301. This material has double the elongation brake capability (over 20%) compared to 65Mn and corrosion resistant. In the same time I changed the shape of the blade to ease its removal in case of filament blockage inside the extruder.

Material 65Mn, bending force = 58.2 N

Material Stainless steel 301, bending force = 60N

These simulations proved that the 301 stainless steel material performs even better compared to spring steel 65Mn with slightly higher force (better filament grip). The slight change in the tension blade shape had no significant impact on the force required to bend 2.5mm, meaning we have the same grip performance of the extruder.

Smart Features

Myself, being electronics designer, I had a strong motivation to integrate some smart features as well. I just couldn't live without incorporating some electronics inside the extruder, electronics is in my DNA!

After the tensioning mechanism this was the second hardest thing to design, The main question was requirements, what features to implement, how many features are enough…which firmware to support etc.

My first design version was including a small ATtiny404 microcontroller for signal conditioning and some basic functions like driving the RGB LED based on filament presence and unload status.

This design was compatible with RRF, Marlin and Klipper, but ATtiny firmware could be loaded only in the factory. Was not really what I imagined, it wasn’t good enough for me. Also the wires over the top side was not very convenient even though I tried a version with side connection using FPC cable. This second was more user friendly but required special cable and adapter from FPC to normal wires.

Since Klipper gained so much popularity it became clear for me that most users of this extruder including myself will use it with Klipper. So I skipped supporting other firmware. Of course this does not mean it cannot be used just is not tested by me and I have not defined any macros for other than Klipper.

Next, I went with the design to the other extreme, integrating a full tool-board inside the SO3. It had sensors, microcontroller, stepper driver cooled by the front heatsink. Due to leak of space for connectors I planned a second board attached to the bottom of the SO3 which included several connectors for the stepper, heater, temp sensor, USB etc.

This PCB design was one of my most challenging I ever made because of the so tiny available space. Nevertheless, I managed to make it working but in the end this solution proved to be very sensitive to manufacturing tolerances and assembly. Overall, it was an overcomplicated design and still needed to have a second PCB board over the bottom with the rest of the connections which also made it unpractical for many 3D printer designs.

Considering all these I decided to go for a simpler and more robust solution and integrate the minimum needed just for Klipper. Also moved the signals connector to the bottom side of the extruder.

I know some may have concerns that the wires are too close to the hot-end and may suffer from it in the long term. In fact, the connector is recessed in the aluminum body of the extruder, where thermal simulation shows that the temperature does not go over 40°C degrees. In enclosed chamber (60°C) the temperature is about 70-73°C which is well in the safe zone.

I designed also tool-board which can be attached to the stepper backside, but this is for another story.

Aluminum Housing

Initially only the heatsink was planned to be machined out of aluminum, the rest of the components where designed for MJF printed glass filled PA12, beside the molded planetary gears. Similar to the LGX extruders printed plastic parts are way cheaper compared to machined aluminum.

For mounting the extruder to the printer and adding accessories, I used slots with square nut inserts.

I tested several versions of the SO3 using these plastic parts, and just before announcing it I received some bad printed samples. Similar issue like we have with some of the sensor housing. Then It became clear to me I have no control over the quality of the external printing service supplier. Some parts are good some not so much. For the latest printed parts I received it looked like they shrunk after printing. Shrunk in a way that reduced the thickness of all the part walls. Overall outside dimension where kind of at the minimum tolerance, but all the holes where like 0.2mm larger. This was a major setback and unacceptable.

Together with Jason from LDO we decided to change the complete housing design to aluminum machined parts. We focused on quality over manufacturing price.

To my surprise it came out really good. Saved lot of space inside, since I could remove the slots and use taped holes the overall weight of the extruder was not affected at all.

In addition, since I’ve done all the thermal simulation considering plastic housing now the additional aluminum parts acts as more cooling element for the heatsink and stepper motor improving the extruder thermal behavior.

Drive Gears

There are lots of debates out there which drive gear design is the best. Some uses small gears but angled tooths, others large like the LGX. The Orbiter is somewhat in the middle, you may read more about that in the story of the Orbiter v2.

Another debate is gear eccentricity, new drive-gears with integrated shaft reduces eccentricity improving surface print quality like the Bondtech IGDA or the gears of VZ-Hextrudort. Let me say this is not a new topic, just a reinvented one and of course well documented and explained by VEZ and HevOrt, a big thumbs up for their great work! All E3D extruders from the titan to date already have one piece gear & shaft. I have a feeling they also know why, not just a lucky guess.

Drive gear eccentricity

Eccentricity error is always present there when two round parts are fitted together unless they are press fitted together, considering good machining tolerances.

For the Orbiter v2 and SO3 a shaft integration together with the main gear was a no-go, for two reasons:

1) There is a need for a bearing between the drive gear and the planetary gear which would have been impossible to insert. Of course, one solution would be to use a bearing with a much bigger diameter, but this would still not really solve the issue because you could not cut the tooths of the drive gear. I was searching for a solution which could be adapted to the Orbiter v2 as well as upgrade kit for existing owners.

2) Second the weight of the gear would have increased to much. Steel is much heavier than aluminum.

A less complicated solution is using a slightly deformable spider shaft design, an octagon shaped instead of round.

The outside diameter of the octagon spider shaft is slightly bigger than the inside diameter of the bearing and the drive gear by 0.05-0.1mm. This means the bearing can only be installed by press-fitting. During the bearing insertion the octagon shaped spider shaft will slightly deform, taking over the inner diameter of the bearing. Kind of a ramming process.

After this the shaft is preformed and the drive gear will slide on relatively easy without much play. A grab screw is still needed to keep it from rotating but the gear will still be well centered. This method reduced the eccentricity of the gear by factor over five compared to the first Orbiter v2 version.

The picture you can see the improvement between Orbiter V2 with 7.9mm anodized shaft and the Orbiter v3 with octagon precision shaft.

Please note here I used the worse shaft of the Orbiter v2.0. The current Orbiter v2 have 8.0mm shaft and its performance is closer to the SO3. In the near future we will equip the Orbiter v2.0 with precision shaft as well.

Surely the difference is not tremendous but there is an improvement. Hard to make a picture to really show the difference.

When I first saw it I was very puzzled by the twirled tooth design. I was not able to judge if its better or worse.

The idea is that the twirled design improves filament centering resulting in improved extrusion consistency. One gear pushing the filament to the left, the second to the right with equal force, so it’s kept centered by the “force”.

Here are my concerns with that idea:

1) If the tooths are angled the extrusion force will be split in two forces, one pushes the filament down the nozzle and second, pushing against the other drive gear, basically wasted. This means reduced extrusion force. The loss of extrusion force maybe not big but is not zero either.

2) If we have a lateral force, every force has an equal and opposite force. This means the gears are pushed towards the housing / latch arm sides which shall push it back. In the VZ/Hextrudort there is a big gap between the idler gear and the latch side.

There are no trust bearings to take the side forces, when extruding the drive gears are not touching the sides...so in my mind I’m missing the equal and opposite force, so maybe there is no force in the first place?

I had to do my own research to better understand what’s happening there, therefore I bought a VZ/Hextrudort with twirled gears "for benchmarking reasons only" and tested it against the Orbiter v2 with standard Bondtech gears and 8.0mm driveshaft.

The result was surprisingly good, the VZ/Hextrudort was performing very well, in the right-angle surface seem to be slightly smoother, but wooden pattern test was slightly more pronounced on the VZ extruder (compared with orbiter v2 8.0mm shaft - latest version not anodized which is more precise).

I concluded that maybe my understanding is wrong and twirled gears do improve. I designed twirled gears and with the help of Jason from LDO we ordered samples from Phaetus.

Surprise again, the angled tooths on the Orbiter v2 and on SO3 performed worse than the straight ones. More severe under-extrusion in high nozzle pressure scenarios. Surface of the prints show no visible improvement. So what is happening here? Again asked my wife mum, mother in law lovely kids...again, they had no clue what I'm talking about.

Investigation....

When the nozzle pressure increases the force pushing the filament upward increases as well, the tooth being angled it forces the filament to twist clockwise, like an unscrewing effect resulting in under extrusion behavior. You can see the increased extrusion error over 30mm³ (blue colored line).

I did tested with several filaments and the result was similar, here I posted the results with ABS and PLA.

First I have not observed this effect with the VZ/Hextrudort. Took me a while to realize something is fishy here...after re-testing I have observed the similar behavior slightly less probably due to different gear size or less grip or different tooth shape...Using the Goliath hot-end most probably this effect is hard to observe due to its very high flowrate. This effect requires high backpressure which is reached near to the max flowrate of the hot-end.

Similarly, the surface artefacts where not improved due to forced centering of the filament compared to straight gears. Results where exactly the same without any notable surface quality improvement.

I honestly concluded that my initial concerns where correct, "the force is missing" the improvement in the VZ/Hextrudort is mainly due to the better concentricity of the parts. The twirled toot gear design does not really improve in my opinion. According to my evaluation it actually worsens extrusion performance in high nozzle pressure scenarios. This can be observed over print surfaces where fast printed layers are near slow printed ones. The slower printed layers appears over-extruded.

I know some readers will not agree with me on this but I’m OK with that.

May the extrusion force be with you my friends!