Zephal Technologies

Three lessons learned from making a large 3D printed part

When I was asked to make a door tray using 3D printing, the difficulty went up a notch compared with the small parts (no more than around twenty centimetres) that I was used to printing.


This part was 75 cm long, around three times the capacity of the printer I used, the Anycubic Photon M3 Max.


A lot of mistakes were made in the course of this project, and if this article helps you to make large parts more quickly without wanting to throw your printer in the bin, it will have achieved its objective!

1

Separating a large part

Creating the basic shape

The customer supplied me with his left-hand door tray and wanted a right-hand door tray, which was symmetrical. So I had to recreate his part in 3D from the original. Armed with a ruler, a metre and the indispensable calliper, the tray was recreated in Fusion360.

It was crucial to place the connecting elements in the right place so that the tray could be fixed to the storm door. A large number of measurements and photos were taken to recreate the room.

Tip: To recreate a part whose dimensions are difficult to obtain, an effective method is to recreate two elements that are easy to measure (in this case the two holes/notches at the top left and right of the tray), then take a photo of the tray and use this image as a reference by aligning the elements in the photo with the elements in 3D.

Separation strategy

To separate our model into several parts, the first thing to consider is the maximum print size of the machine. In this case, it was an Anycubic Photon M3 Max with a print volume of 298x164x300mm.


Our part measures 743x160x85mm. The cutting strategy was relatively simple, given that only the length was too great. The model is therefore split into three parts, each about 25cm long.

A preferable cut in terms of overall stability/ease of assembly would have been a cut at the central rib (between parts 2 and 3 below). However, the resulting parts would have exceeded the maximum size allowed by the printer, and they would have had to be re-divided to compensate, which would have eliminated this advantage.

What not to do when dividing the model

Once the model was cut into three pieces, I made three mistakes:


  1. Not modelling a connection interface, i.e. everything that allows the models to be positioned correctly in relation to each other and fixed together.
  2. Trying to print everything at once by positioning the models in a non-optimal situation for printing.
  3. UV-treat parts before connecting them together.


First attempt at printing the door tray, a failure
First attempt at printing the door tray, a failure
UV treatment and early exposure to heat
UV treatment and early exposure to heat
Unintentional bending of an element
Unintentional bending of an element
This detail did not stand up to the way it was printed
This detail did not stand up to the way it was printed
On the other side, this has created crevasses
On the other side, this has created crevasses

As a result, delivery of the tray was delayed. In financial terms, it meant the equivalent of €70 and 29 hours of printing went up in smoke.

Creating a connection interface

In order to link our bin sections together, we had to create a connection interface. The previous test used only glue on a small section, which didn't hold up. To obtain a precise and solid connection, we used a positioning tab on which glue would be applied, as well as M2x10 CHC screws all over the inside.

As well as solidifying the assembly over its lifetime, the screws allow the glue to be applied to the surfaces without coming off until it's dry.

Note: Some screws, such as these M2x10 screws, are difficult to find in supermarkets. You can find them on specialist sites such as RS Pro.

The advantage of resin 3D printing is its precision of 0.1mm. So, as long as we're not talking about mechanical parts requiring precise adjustments, the modelling of elements can (almost) always be done without tolerances.

2

Printing direction

Part positionning strategy

For the failed bac, everything was printed at once. How was this possible? By positioning them in a non-optimised way. To fit them all into the print volume, they were placed vertically.

As a result, during the slicing process (the division of the model into printing layers), large areas of resin were printed all at once without sufficient structure to hold them in place.

This phenomenon is particularly visible in the top left of the animation above. We are moving from a matrix of supports to a large area of material. As a result, when the platen rises, the force exerted on the printer's FEP film is very high and the bond between this layer and the previous one tears.

Under layer

Layer too large just above the left one

The internal and external links to this geometry did not withstand the shock

So how should you orientate your room? It depends on a number of parameters that need to be optimised according to your needs:

  1. Cost
  2. Printing time
  3. Final quality (the better the part comes out of the printer, the less finishing work you'll need)

Reference model

Standing

On side

45° Inclination

2° Inclination

Printing strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Printing time

Final quality

Reference model

181 mL

0

10,9 €

X

X

Standing

460 mL

61

27,6 €

29h

-

On side

442 mL

59

26,5 €

17h

+

45° inclination

543 mL

67

32,6 €

29h

++

2° inclination

400 mL

55

24 €

28h

+

It's always heartbreaking to see so much resin going into the racks... But we've got no choice!


With this comparison, we can identify three valid solutions for our part of the door tray:


  1. To optimise cost, the 2° slope is the winner. At ‘only’ 55% of the resin used for the substrates, it allows us to print the part for 24€ while avoiding the inconvenience of 100% vertical printing. On the other hand, the shallow angle risks creating a staircase effect on the surfaces. Nothing too serious, but extra sanding time is required.
  2. Printing the part on its side optimises printing time. In fact, the 11h saved is far from negligible, especially if the delivery is rushed. What's more, the overall quality is equivalent to tilting by 2° and costs just €2.50 more.
  3. If you want to be sure of getting the best possible quality, the 45° tilt is ideal, because you can be fairly certain that you won't run into any printing problems. The extra cost is high compared with the 2° tilt (+€8.60) but can be offset by a higher rate of successful prints.


To reprint the tray, I opted to print on its side so that I could start printing in the evening and pick up the part in the morning. The print went off without a hitch.

3

Best supports strategy

Now that we've determined the orientation of our room, let's look at the supports themselves. In most of the slicers available (ChiTuBox, AnycubicPhotonWorkshop, Lychee...) the supports can be configured.


In AnycubicPhotonWorkshop, for example, we can choose light, medium or heavy supports.

To determine which structure was ideal, this test piece was printed three times using three different types of substrate.

We had :

  1. Heavy’ substrates with a low density of substrates
  2. Medium’ media with a medium density of media
  3. Light’ media with a high media density.

To test the strength limits of all these substrates, the part was printed in a non-optimal direction (large surface with nothing underneath to replicate the error made in point 2).

The light media parameters have been modified to give a higher density of media than was originally the case.

Results at the end of the machine. From left to right: Heavy, Medium and Light substrates. The Medium supports did not hold and the model broke during printing.

Fault zones, from top to bottom: Heavy, Medium and Light supports.

Printing strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Dimension after UV

Final quality

Reference model

11,4 mL

0

0,7 €

56 mm

X

Supports "Light"

45,1 mL

75

2,7 €

56,06 mm | +0,06 mm

-

Supports "Medium"

26,4 mL

57

1,6 €

55,97 mm | -0,03 mm

- -

Supports "Heavy"

30,5 mL

63

1,8 €

55,90 mm | - 0,10 mm

+

We can see that both the ‘Light’ and ‘Medium’ parts have major defects. The ‘Medium’ part broke off completely, while the ‘Light’ part had a large crack that rendered it defective. Only the ‘Heavy’ brackets held up. On the other hand, the part is less accurate dimensionally, with a 0.1mm difference in width compared to its nominal dimension. These dimensional differences will be the subject of a future study.

Application to the door tray

To study a real case, let's look at the tray. In the study in point 2, ‘Light’ media were used. If we try all the versions, we get the following results:

Reference strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Final quality

Reference model

181 mL

0

10,9 €

X

Supports "Light"

442 mL

59

26,5 €

-

Supports "Medium"

301 mL

40

18,1 €

-

Supports "Heavy"

358 mL

49

21,5 €

+

As a result, the most advantageous substrates overall are the ‘Heavy’ ones. They use 57ml more resin than ‘Medium’, but the print is more likely to give a good result that won't have to be caught up in finishing.

The major drawback of the ‘Heavy’ is the larger mark left on the part, but given that a large assembly such as this will require treatment and sanding afterwards, this is not a drawback here.


With the results of this study, we can compare the first printing strategy, which was a failure, with this optimised printing strategy:

Partie 1

Partie 2

Partie 3

Part 1 - Blue

Printing strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Printing time

Final quality

Printing strategy 1

460 mL

61

27,6 €

29 h

- -

Optimized printing strategy

358 mL

49

21,5 €

17 h

+

Gain

102 mL

12 pts de %

6,1 €

12 h

+++

Part 2 - Grey

Printing strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Printing time

Final quality

Printing strategy 1

546 mL

64

32,8 €

29 h

- -

Optimized printing strategy

339 mL

41

20,3 €

17 h

+

Gain

207 mL

23 pts de %

12,5 €

12 h

+++

Part 3 - Red

Printing strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Printing time

Final quality

Printing strategy 1

172 mL

26

10,3 €

29 h

+

Optimized printing strategy

172 mL

26

10,3 €

17 h

+

Gain

0 mL

0 pt de %

0 €

12 h

/

Total door tray

Printing strategy

Quantity of resin used

% of supports

Price (Resin at 60€/L)

Printing time

Final quality

Printing strategy 1

1178 mL

57

70,7 €

29 h

- -

Optimized printing strategy

869 mL

42

52,1 €

2 x 17 h

+

Gain

309 mL

15 pts de %

18,6 €

- 5 h

+++

Conclusion

By optimising the printing method, we were able to save 309ml of resin for the same print, resulting in a significant 26% reduction in costs. At the same time, the quality of the print has improved, giving a more qualitative result, with no major printing defects to make up for.

Finished door tray
Finished door tray
Door hatch, rear view
Door hatch, rear view
My hand in comparison
My hand in comparison
A part ready for assembly
A part ready for assembly