18 septembre 2017

Cutting Aluminum

Last weekend, I tried to cut an aluminum sheet 2mm thickness. This was my first test with CNC to cut aluminum, as I used to cut carbon sheet and fiberglass mainly .

Result is good so far, next trial will be with a 4mm sheet. It needs to add some oil spraying during the milling.

My settings:
- drillMill: 1 tooth, M2x8
- cutting speed of the Kress: between 5 and 6 (RPM?)
- cut advance: 200mm/sec
- layer: 0.2mm

10 septembre 2017

Tests to improve print quality

This week I made several progresses which improve the print quality, especially regarding retract and ghosting.
First, I installed and test the BondTech BMG extruder. It is a must, compact, very efficient , light, it has only adavantages. With it, and the mirror on bed, I could make my best first layer.  I did first calibrate the width of the line. Once done, this line is very clean. Retract has been set at 3mm. There is still somme stringing (oozing), but it is reasonable.

Then, I tried  to  remove the ghosting effect (aka ripple or vibration). I split it in two: first part appears when the print head accelerate, second part is all over the wall, I 'll call it woobles.


To reduce the wooble  I added   TL-smoother (2) between the TMC2100 driver and the stepper (on X and Y) . Picture show a comparison before (left) and after(right).
We can see that the right one has less woobles. So the TL-smoother seems to have some effect on the TMC2100, however the TMC2100 becomes much more noisy...

Second, to reduce the ghosting, people on forum always says: "reduce acceleration, reduce speed".  So, that what I finally  did because I could 'nt find any vibration problems on my printer mechanic.   But, this time, instead of changing in the Marlin firmware for all movement, I changed it in Slic3R. The last version (2.7.0)  allows to set an acceleration for each movement type. So I set the acceleration for perimeters to 100 (instead of 500). And the result is better, left=before, right=after

23 août 2017

Bed adherence

After years of using blue tape for bed adherence, I decided to try something else which do not make any mark below the print and which would work at least for PLA and PETG...
So, my friends of the Grenoble maker club, recommended me to try using mirror  with head spray. They advertise it me so well that I decided to try.

As I had some old mirrors around, with good  dimensions, I simply lay one done on the Tevo Tarantula aluminium bed (without the PEI) and fix it with clips. Then I spray the surface with hair spray .
That's all.

First test with PLA succeed with no problem, the bed was a 50°C, adherence was great , no warping nothing. After print is finished, then I needed to wait until the bed is completely cold, or put the mirror + parts  into a fridge. When cool the printer part can be very easily removed.

Then I tried exact same method with PETG, with bed heated at 75°. It also works great ! The part I tested was  7x5cm so not very large, but it worked fine. Same as for PLA, wait until the bed is completely cold to remove the printed part.

In addition, the mirror surface is perfect, and bed can then be easily leveled.
So all benefits, I think I  can now forget the blue tape ... ;)

14 août 2017

Vibration, ripple, shadow, ghosting

I have own several printers and the ripple effect always  appears on most of my print.
How to get rid of it ?

The ripple or vibration or shadow effect looks like this:

Look on the surface where it is bright, do you see all this little waves ? They are vertically aligned. The surface should be perfectly flat and clean.
This test has been printed with my last printer, home made, see TevoTarantula in an earlier post.

Thanks to "megablue" who did post this very demonstrative test. You can find it here on thingiverse.

To remove this effect, I tried several things: switch from delta (the worse for ripple effect) to Prusa, change rods on OboXL3D (big ultimaker like), switch from bowden extruder to direct, switch from Direct to the E3D Titan Aero...

Here is the list, mechanical, electronic and settings:
- belt tension
- axis movement softness
- remove any play
- replace motor (Nema 17 1.8° and 0.9° per turn)
- stepper driver (A4988, DRV8825, TMC2100)
- various electronic boards (Mks base 1.3, MKS base 1.5, MKS SBASE, Arduino + Ramps, etc...)
- frame: delta(Kossel XL), ulimatker like, Tarantula Tevo (prusa), Minifab, Anet 3d(prusa), TevoTitan(prusa)
- bowden to Direct extruder
- Direct extruder to Titan Aero
- decrease X, Y acceleration
- decrease X, Y jerk
- Increase E jerk
- increase E acceleration
- reduce X,Y max speed

My last print improved, even if not perfect. What I changed ? Filament mainly...
Here is the result:

So to conclude, it seems that the best solution is still to reduce speed as much as possible, and avoid abrupt acceleration as much as possible, because the extrusion can't handle it quick enough..
Do not hesitate to post a comment below if you want, there is may be a good solution to make technical parts looks better.

09 août 2017

Testing a very low cost powerfull chinese servo MG996R...

Well, I am dispointed by this very low cost chinese servo MG996 , which is supposed to provide a 9kgcm torque...
It is less powerfull than a Hitec 5485HB   with 6.5kgcm....

In fact, both servos do not comply specifications announced, even the hitec one. Be warned.
But the Hitec is still better, with less power consumption.

See the videos of the various tests:

MG996 + 2kgcm
MG996 + 4kgcm
HS5485HB + 2kgcm
HS5485HB + 4kgcm
HS5485HB + 6kgcm

06 août 2017

GT2 pulley

After some search I found this way to make a pulley of various size , for a GT2 belt with a 2mm  pitch.

First, you need to know the number of tooth, and the pitch of one tooth (2mm ). Then compute the diameter of a circle which will contain a polygon. The side of the polygon is the pitch, and the number of side of th epolygon is the number of tooth. This site is great  to help that computation:
otherwise this formula:
Inradius r
  • r = (1/2)a cot(π/n) = R cos(π/n)
where r  is the distance between the center of wheel, and the nearest point of the side, and n the numer of sides of the polygon:

Once r value is found, then the wheel can be created under CAO software with this method:

1) make a solid wheel a little bigger than r, thinckness 8mm is fine for GT6
2) extrude draw on the surface , one tooth  like this:

The distance between both extrmimites of the tooth should 2mm. The diameter should be .78.
The center of the circle should be at r distance from the center of the pulley. 
3) do a circular repetition of the function , based on the center of the wheel

31 juillet 2017

TaranTitan: second version

I call it TaranTitan as it is started from ideas of the Tarantula printer and use the E3D Titan extruder.
The TitanAero is a direct extruder (no bowden) small and pretty light. It should bring only benefit to the quality of the print.  Based on my previous design Tarantula++obo, I made the follwing changes:

  • retrofit from 32 bits electronic to 8 bits (mks-sbase) to Mks-base 
  • retrofit the stepper motors from 0.9°/step to classic 1.8°/step
  • remove the external box and fix all electronic and Raspberry to the frame. This save a lot of space on my bench
  • replace the dual head extruder by a Titan Aero from E3D
  • improve the part which fix the Y axis to the frame
  • replace the  TFT32 by a graphic LCD display. It allows to see print progress even if print was launched from the raspberry.... the TFT32 does not
  • add some LEDS on both sides of the frame
  • simplify the Y-motor fix
  • replace power supply by a good meanswell 350w, much more silent that the Amazon power supply at 19$...
  • AND Finally, the most important change, use the new driver TMC2100 which are much more smooth and silent than previous driver used on the mks-sbase (drv8825)

The result is this 3D printer, which now provide very good quality prints  :)