Introduction
The true story of the Orbiter v2.0
In this story you can read about how I designed the Orbiter v2.0, full
with detailed technical description, principle theories and some intrigues we faced
along the way.
There are many misconceptions about what makes an
extruder design great, I'm positive about that reading this story will help everyone get a better feeling on what to really focus on, and see behind advertising data like a pro!
I started the design of the Orbiter v2.0 is the summer of
2020. The Orbiter v1.0 and v1.5 where already pretty mature designs at that
point but both had a major drawback in my eyes. Both uses standard on the shelf available
components. On one hand this is great, makes them a very nice DIY project but
in the same times comes with the compromises and drawbacks of the standard components which weren’t specifically designed for the Orbiter extruder. For me
the Orbiter v1.0 was mainly test vehicles to prove the concept itself.
To the disappointment of many, the Orbiter v2.0 is not a
DIY project anymore. It was redesigned completely from scratch, with custom
made parts to leave behind the drawbacks of the on the shelf components.
The main re-design targets for the Orbiter v2.0 where:
To reach these goals almost every aspect of the design was re-imagined and redesigned from scratch, you can read about each step in the next sections.
New filament drive gears
First step to achieve the compactness was to redesign the
drive gears.
I decided to keep the same diameter but
reduce as much as possible its length. I designed a new gear with 10mm
length instead of 15mm.
Weight and length reduction of 33% in the same time great!
Well not exactly ☹. Gave it to manufacturing
and spur gear tooth cutting tool was cutting into the filament tooths as well, “great design” but no tool to manufacture it. Therefore, I increased the
distance in between by 1mm ending up with 11mm long drive gearset.
Jason ordered first samples from Phaetus, they were working just great.
Living by the rule of Jason to support the original
creators he insisted to have the gears manufactured by Bondtech. I was not so
sure about ordering components from “competitors”, but hey let’s give them a
chance, LDO also supplies them motors for the LGX and BMG so why not.
Received the first samples from Bondtech…oh boy
disappointed I was…way worse compared to Phaetus gears. Again, Jason insisted
to give them a chance to correct the mistakes, in the end we are in the
development phase not everything goes smooth from the beginning. Since Bondtech
has a long history to make very good quality gears I assumed they will nail it
by their first shoot. It wasn’t the case.
We scheduled a meeting with Bondtech to discuss the problems we see and solutions to fix them.
I have to admit I was positively
impressed by their willingness to collaborate and the openness they have shown. They
were really hands on and immediately made corrections / adjustment of their machines and sent a new set of samples. And yeah, second time Bondtech nailed
it perfectly.
For the best grip the tooths are extremely sharp, about 40 microns width at the top.
You might remember Bondtech claims their gears surface is hardened and they last longer.
I can confirm this. Overall, the
extrusion and filament grip performance of the samples received from Phaetus
and Bondtech were similar. One plus to Bondtech for the harder surface
which increases lifetime.
You can see the scratch test result, left Bondtech and right the Phaetus gears.
I have seen new “RNC Nano Coated Gear” advertised on
Aliexpress, even if some say its authorized Orbiter v1.5 version, I was not
involved in that I have not received nor tested any samples I do not know if
it’s any better or worse.
From I read in the literature both hardening processes reaches similar surface hardening of over HRC60.
I have to admit they are looking cool :).
Gears Diameter
When I started the design of the first Orbiter extruder,
I chose the bigger 12mm instead of the most common 8mm drive gears. Even if this
ad a bit more weight I considered the advantages given by the better grip
outweighs the disadvantage of 10-20 grams additional weight.
Let’s look into the details!
Next picture shows the
contact area VS pinching depth of the driver gears for 8mm diameter used in the standard
BMG style extruders, 12mm diameter used in the Orbiter (also Bondtech QR) and the 18mm diameter used in the Bondtech
LGX.
Note the effective diameter of the filament path is smaller by ~0.6mm.
The data is summarized in the next table, out of the calculations the contact area increase is roughly 25% more for each.
Now why is that important?
When the extruder pushes
with high force the filament plastically deforms under the high pressure. This
leads to under extrusion at higher print speeds. Eventually if the backpressure
is high enough the filament is grinded or stepper skips. With more tooth coming into contact
with filament, it can be pushed with a higher force before it gets plastically
deformed and grinded.
This effect is worse for softer materials like HIPS, ABS, ASA or Polypropylene.
This picture shows extrusion performance comparison between 8mm and 12mm drive gears using ASA, tested using Dragon HF hotend with standard plated copper 0.4mm nozzle @250°C.
At high nozzle pressure the
filament is either grinded or the stepper motor skips depends which one gives
up first.
The Orbiter extruder uses
12mm gears, in all my tests the stepper started skipping first, which is a good
sign because it means the gears have the exact amount of filament grip, we
need. Of course, even bigger drive gears would slightly lower filament
deformation and more precise extrusion in the expense of increased weight,
drive gears being made of heavy steel.
Overall, I think the short
gears of Orbiter 2.0 is the best weight (10g) / performance ratio. Bigger gear diameter will not improve much since the limit with 12mm gears is the stepper power itself.
Full test comparison results between 8mm and 12mm drive gears you can see below. In all cases the 12mm drive gears gave better extrusion consistency and lower extrusion errors.
The tests were performed using Phaetus Dragon HF hotend with 0.4mm standard plated copper nozzle.
Housing and latch
I had four main goals with
the new housing design.
1. Flip the drive gears, the filament path is moved closer to the stepper, this makes the Orbiter v2.0
to perfectly fit for delta printers.
2. Align the fixing screw
positions with the filament exit path. This is a must to have no twisting & tilting forces acting over the extruder and extruder mount by hotend pressure change during printing.
4. Adjust assembly screws positions for easier access.
3. The design should look very cool.
To reach my goals I had to start from scratch. Flipped the gears and rotated the stepper motor for additional 10° for a better clearance towards fixing screws.
After a few weeks and tests I finalized the first design version and sent it for manufacturing.
Surprise 😊 the
design I’ve made was not manufacturable.
There was no possibility to make a mold for the new housing. Only 3D printing option was possible. Jason was on the opinion to better change the design and make it moldable, this offers better quality and higher temperature rating.
Luckily the Orbiter user group has many experts in
different fields including molding technologies.
With the help or Yves
Fractal from Fractal Engineering we identified the issues and made a plan for a
design correction. I’ve implemented all the changes and via Jason from LDO I sent the design back
for manufacturing (same manufacturer who made the molds for the Orbiter v1.5).
Surprise again – “the
design is to complex and we cannot make a mold for it”
Jason sent the design to more experienced mold manufacturer, this
time they advised for some further minor changes and with that they have
manufactured the mold for the Orbiter v2.0 housing and latch.
As usually surprises do not
end here as I like to say “some can clone faster than I can
design”
Shortly before the release
of the Orbiter v2.0 release I have received a message from a well known
reseller an manufacturer of 3D printer parts:
“We have also made your 2.0
design, and x% of future sales will be sponsored to develop new products for
you.”
Cannot say I was not
surprised, asked them if they have collaborated with LDO upon this design. They
said yes this are just pictures of LDO samples they received…pictures my as..
In reality the first company
(Formbot Vivedino) who declined to make the Orbiter v2.0 housing actually have
made the molds based on early version of the design (as far as I can tell). If
you take a close look, you can see the mold contact points are in different
position compared to the mold we made together with LDO. Plus you can clearly see some warping effects on the surface of the extruder.
This is not the
way I collaborate! I’m not selling my approval for a design for x% just so they
can mark it genuine and boost their selling. Especially since I have no clear
picture how they got the design files and have zero clue what and how they
implemented the design. Never tested and never seen their design in real life.
The stepper
This section is a bit more
technical I’ll try to be easy for everyone’s understanding.
The first motor I used to test the Orbiter concept was a Wantai 36HS2418
stepper, actually that motor with a T10 spur gear inspired me the possibility
of using planetary gear reducer in an extruder.
After I published the Orbiter v1.0 files, the Voron design team asked Jason from LDO
to make a better alternative. They have made the LDO-36STH17-1004AHG high
temperature stepper which had similar performance but less weight.
LDO-36STH17-1004AHG - optimized
for low speed => high inductance (6mH) and for fast current decay => high
phase resistance (10Ω).
At that time everybody was
on the opinion that an extruder stepper must have high inductance and
resistance. Higher inductance to improve torque at low-speed range the extruder
operates and higher resistance for a faster current decay time which helps in
faster extruder reaction.
Well, partially true 😊.
The high phase resistance is only important for slow decay operation
modes of the stepper driver. Basically, it uses the phase resistance to consume
the energy stored in the windings inductance.
All new driver chips includes fast and mixed decay modes.
Slow decay mode means when
the driver wants to reduce motor current it short circuits the phases. The
current decays with a time constant defined by:
tdecay=Lphase/Rphase
Fast decay mode basically means that the driver reverses the voltage applied to the motor phase during the current decay time. This way the current changes much faster and the value of the phase resistance does not matter anymore.
A drawback of the fast
decay mode is higher current ripple in the motor.
To overcome this negative effect, the best choice is to use mixed decay mode. The driver applies slow of fast decay mode depending how fast the current needs to change in order to construct a sinusoidal shaped phase current. In short words slow decay on the rising edge, top and bottom areas of the sinewave and fast decay on the falling edge.
A huge disadvantage of the
high phase resistance is higher power loss.
The electrical power loss of
a stepper motor is:
As many users have experienced on SLS printed Orbiter v1.0, a slight
misconfiguration of the LDO-36STH17-1004AHG current, lead to a full meltdown
due to high stepper temperature and low heat deflection temperature of the Nylon
PA12 SLS printed parts. Same issue using LDO-36STH20-0504AHG (8.5mH &
13.5Ω).
Since we learned phase
resistance is not really needed due to mixed decay modes all new drivers have, lead
to a couple of questions:
- Our current motors (36STH17-1004AHG and 36STH20-504AHG) are rated to 1A but we can use them only up to 0.35A without complete meltdown – what if we reduce resistance and increase current, aiming for the same motor temperature (about 70°C) do we get improved motor torque?
-
Reducing motor resistance, we also reduce the
number of turns in the winding and its inductance which leads to reduced torque – so what is the best compromise of resistance VS number of turns and resulted torque in real printing scenarios?
-
Could we get better torque curve at low
speeds using 0.9° steppers?
In the beginning of 2021
with Jason from LDO we started to investigate. They made several winding
combinations and step angle versions. They did some torque curve performance
analyses as well.
Maybe you have seen such graphs already.
Then I realized those curves say nothing about reality.
Here is why:
Let’s consider a dragon HF
hotend which tops out at about 25-45mm3, depending what filament and
temperature you use. OK let’s consider 45 mm3. We know the diameter
of the drive gears 12mm and the gear ratio of the planetary gearbox is 7.5:1.
Out of this we can calculate stepper RPM VS flow. And stepper RPM VS print speed if we define layer width and height. Next chart shows print speed vs stepper RPM considering layer width of 0.48mm and height of H0.25mm
Did you notice? The first stepper torque chart RPM range is between 400-2800PPS which means RPM range 120-840? And we top out the dragon HF considering ridiculously high flow of 45mm3 at about ~260RPM which means ~400mm/s print speed (W0.48/H0.25)? In reality Dragon HF is more like 30 mm3 which means max RPM of 175 (250mm/s W0.48/H0.25). So, we talk about the first two dots on the regular stepper torque performance charts.
In reality mostly we do not
print so fast and torque data for print speeds below 175mm/s is not even
on the chart.
Based on this I asked LDO to
repeat the test at lower speeds as well, here is the result, nevertheless I
decided to evaluate the steppers torque performance in a different way using
real printing scenarios.
The test I came up with is very simple:
- Make a mark on the
filament @ 305mm before the extruder. Command extrusion of 300mm and measure the
remaining distance;
- Repeat the test with
several extrusion speeds;
- Use same setup filament and temperature for each stepper evaluation;
- For a
meaningful comparison between the steppers I adjusted the current for each
stepper motor under test to have about ~3.5W electrical power loss, reaching ~70°C.
Next chart presents the performances of the motors and variants we have evaluated.
The test was performed using Orbiter v2.0 + Dragon HF with 0.4mm standard plated copper nozzle, PLA @ 220°C.
Apparently, the winner
configuration would be a stepper motor with very low resistance and inductance
1.6Ω & 1mH. Well things are not that simple as it looks like.
Because our stepper motor
operates at low speeds, we have to consider quantization error of stepper
drivers. I know it sounds sci-fi but is not that hard to understand.
Stepper driver regulates the
motor correct via controlling the voltage it applies to the motor phases. As
the stepper speed increases the back EMF voltage increases as well, so to keep the motor current the stepper driver increases the phase
voltage amplitude. This is the reason beyond a certain speed the stepper loose
torque. Increasing motor supply voltage expands the stepper speed range.
Now the applied phase voltage amplitude has a finite granularity we call
quantum q or resolution. This is the lowest output voltage change the
stepper driver can produce.
Stepper drivers being digital circuits the lowest voltage step it can
produce depends on the battery voltage level and on how many bits the sine generator operates. This formula defines the amplitude of a quantum q. where Vm is the supply voltage (24Vdc)
n – number of bits the stepper driver operates which is usually 8.
The sinewave phase voltages
and currents are generated using a quantized signal, which means the voltage is
increased and decreased to form a close to sinusoidal shape as shown in this
figure. The amount of increase and decrease of the phase voltage cannot be
smaller than one bit resolution of the driver, one quantum q.
Now having 8 bits (most
common resolution of stepper drivers in 3D printers) to generate a sinewave
from a 24V motor supply voltage we have the minimum voltage step about 94mV.
Next table shows what current step this means for different steppers and how many steps are used to generate a sinusoidal shaped current at low
speed.
To have smooth and noiseless
control we need as many steps used for the sinewave generation. In case of the
1.6Ω stepper we can see that only 25 steps out of 255 are used in full current
and 12 out of 255 at half of the motor current. Therefore, I have chosen the
2.4Ω version (LDO-36STH20-1004AHG) for a lower quantization error. With this approach even stepper drivers with lower performance can drive the Orbiter extruder with excellent performance.
Gearbox & gear Ratio
Every good preforming
extruder uses some type of speed reducer. This is because stepper motors do not
have enough torque especially at low speeds to push filament with high enough extrusion force.
There are lots of
discussions about which gear ratio is the best, some say 3:1 some 5:1, 30:1,
Orbiter has 7.5:1…let me say, all assumptions are completely wrong, forget
about gear ratio!
What matters is the resulting
steps/mm, how many steps are converted into 1mm of linear motion. Higher the
step/mm ratio results in higher torque / extrusion force but the maximum acceleration is
reduced. There is a trade-off between extrusion force and maximum acceleration.
The sweet spot for Nema17
pancakes used in most extruders is about 400steps/mm. The smaller 36mm round
Nema 14 stepper has about half of the torque, therefore the sweet spot is around
700steps/mm (micro-stepping set to 16). See next chart for steps/mm VS resulted extrusion force and max acceleration for different stepper currents (calculated using LDO-36STH20-1004AHT stepper performance). Please note the acceleration is just an indicative value because it depends a lot on friction in the filament path, backpressure of the nozzle, how much weight / filament spool the extruder has to pull.
Having the drive gears with
12mm in diameter the gear ratio needed for 700 steps/mm is ~7.5:1. With this
setup we can have good extrusion force and good acceleration, above 8000mm/s2.
There are several ways to
implement a speed reducer gearbox. Most commonly used is a single stage spur
gear reducer. Like the 3:1 gear in the BMG style extruders or 5:1 of the Sherpa
mini etc.
The gear ratio needed for
the Orbiter extruder can be achieved by several reducer topology, let’s take a
look of the pros and cons of each from an extruder performance perspective.
From this analysis point of view the best
solution for a compact extruder design using 36mm round Nema14 stepper is a single stage planetary reducer
design.
Backlash – worse backlash we have using dual
stage spur gears. This is because the first stage backlash is multiplied with
the second stage reducing factor. Best in class would be the cycloidal reducer
manufactured with very high precision. Single stage planetary gearboxes are not
with zero but very low backlash.
You might think backlash matters for an
extruder design. Well, the truth is that as long it's low enough it affects nothing.
The backlash affects only retraction behavior and ads up to the retraction
distance. So as long as its lower than the retraction distance the printer needs it will
affect nothing. Even for dual stage spur gears the backlash do not account for
more than 0.1 - 0.2mm.
The measured backlash of the Orbiter v2.0 is ~0.06mm.
Note: if you set to high pressure advance causing retraction movements, high backlash can cause nonlinear behavior of the pressure advance. I would disregard this fact since a good PA calibration should not lead to retraction during print movements.
Ease of manufacturability and sensitivity to
tolerance - Spur
gears are known to be easily manufacturable. Planetary gearset is made out of spur gears. Worm gears have to be very precise
otherwise will show inconsistent extrusion artefacts. Cycloidal gears are the
worse from this point of view, they have to be manufactured very precise
otherwise it will produce small output shaft speed variation (wobble) which can be seen
as inconsistent extrusion artefacts.
Occupied space – best in class is
worm gear reducer, the others solutions are roughly similar.
Complexity – Lowest complexity we
have for worm gears, and dual stage spur gears. Planetary reducer is a little
more complex. Most complex is the cycloidal reducer with double disks.
Efficiency – best efficiency we
have from planetary and two stage spur gear reducers. Worm gears are known for prone efficiency. In certain applications cycloidal reducers are very
efficient, in case of a compact extruder design with a simple cycloidal design
the efficiency is affected by the many friction contact points between the
moving and stationary parts.
Back-drivable – from my point of
view, this is one of the most important characteristics which influences the
extruder performance. Worm and cycloidal reducers are not back-drivable
therefore pressure in the nozzle cannot be used to accelerate retraction. On
the contrary planetary and spur gear-based reducers are back-drivable and the
retraction can be accelerated twice as fast, considerable reducing stringing
behavior and increases extruder responsiveness.
Lifetime – Worm
gears are among the worse, they being so compact the forces acting upon the
gear tooths are the highest, this is made even worse by the higher gear ratio
they usually are designed for. Best in class would be cycloidal drives, known to
be very long lasting due to their rolling friction like rotation. Second best
is planetary reducer, the forces are shared between the planet gears (three
planets in the Orbiter means three times lower forces and wear for each spur gear tooths).
Unlike other single or double spur gear based extruders with the Orbiter planetary reducer solution we never had or been reported any defects with grinded planetary gears. This proves also the high reliability of planetary reducer solution.
Note: Single stage spur
gear reduction was not included into this comparison since it would not fit in the given space.
Filament exit guide
One of the weaknesses of the Orbiter v1.5 is
the plastic filament exit guide and its distance towards the drive gears. This highly
affects extruding performance of TPU. It was a result of the mold manufacturing
possibilities of the v1.5 housing.
Best material for the exit guide would be
stainless steel, this has high wear resistance against abrasive filaments.
With this solution the filament path is very restricted and the filament has nowhere to go but into the hotend.
This also means
manual change of filament is slightly more difficult. I recommend to use
automatic filament load and unload by using macros triggered manually or via
the filament sensor add-on.
I personally open the latch only in case of
filament jam.
Retaining tension screw
Many users complained that they lose the
spring and the small washer from the tension screw, including myself.
Together with LDO we decided to do something about it and came up with a new tension screw assembly design with retaining feature as some of our users suggested.
The picture shows the new design, we designed
a new custom-made screw and adjustment knob. The knob width is increased for
easier adjustment and manufactured out of aluminum for low weight.
Final words
We put all we learned about extruders into
this design over the past years. I’m really thankful for the great
collaboration with LDO Motors and Bondtech, without them this project would not
exist.
Last but not least thanks to all of you, 3D
printing community, users of Orbiter extruder. My sincere hope this story helped
everyone at least a little to understand better what really matters for a great
extruder design.
The Orbiter project grew a lot in the last two years. As usual every seller wanna get in, flooding the market with lower quality clones. I felt always sorry for exited guys sharing their new toy and seeing its a clone most of the time without them even knowing it.
To combat clone industry we have registered the Orbiter as trademark of LDO in China. We cannot stop cloners but we can stop them using the Orbiter name, I assure you we will.
...I can assure you this is not the last story
of the Orbiter extruder family 😊...