JLCPCB metal 3D printing keyboard

About the case design

This is the design of the case that will be produced by JLCPCB metal 3D printing.

Some parts such as the protrusion on the inside are different from the final shape.


Last time I made a case which was made by resin, I was personally concerned about the thickness of the bezel width, so this time I tried to make the bezel as thin and slim as possible. In addition, the mounting method has been changed to reduce the height of the entire keyboard. As the result, it became as integrated shape with only the bottom, not the top and bottom split case.

As I wrote my previous article, my ultimate goal is to manufacture the case by cutting aluminum, so I designed it with an awareness of the shape that can be cut. However, with 3D printing, the price varies depending on the volume of the output, so we adjust the shape of the order data by removing parts that have no functional problems.

Futhermore, due to budget constraints, we decided to open the bottom part and cover it with FR-4 (glass epoxy, the meterial used for circuit boards). If you print the bottom, the price will more than double, so I took the plunge and remove this part. The change to this frame shape will becomea problem later, but it will be described later.

Cover with a bottom plate made of FR-4 without printing on the bottom

Flat switch plate and case top for easy cleaning
The front (right side of the image) is 12mm in height and thinned as much as possible.

Board design change due to change in mounting method

The previous resin case used urethana foam and a gasket mount that fixed the FR-4 switch plate between the upper and lower cases. 

For the reason, the outer shape of the switch plate had "ears", but this time, the mounting method has beem changed to a fixing method like an O-ring mount, so the ears for the sandwiching are no longer necessary.

Regarding the change in shape due to the fact that the ears of the switch plate are no longer needed, we will not reorder the PCB, but will cut the perviously ordered FR-4 switch plate with a cutter and file the cross section. Acutally when I first placed the order, I decided on the shape, including the screw holes, so that I could try out several mounting methods, so this time it's a living form.

After removing the ears with an acrylic cutter, the end face was prepared with a file.

The PCB of the main unit also has a shape change. Since the previous case had a certain thickness in thee bezel, the PCB protruded the USB connector part in a convex shap.

Because the bezel was thick, the board shape of the connector part was made convex

This time, the bezel is thin and the PCB is inserted from the top of the case, so it is necessary to eliminate this convex shop. *If there is a protrusion, it will interfere when inserting from the top of the case & it is necessary to dent the shape inside the case.

This time, the PCB is inserted from the top side of the case, so it is changed to a flat shape so as not to interfere.

Change MCU to ATmega32U-MU

Although it is not directly related to his case shape change, the MCU has been changed from ATmega32U- AU to ATmega32U-MU. Although the ATmega32U-AU and ATmena32U-MU are functionally equivalent parts, they have no following differenes.

  • MU(7mm*7mm) has a smaller component size than AU(10mm*10mm)
  • MU has no soldering legs, so it is difficult to solder by hand.
  • The unit price of JLCPCB's PCBA service is overwhelmingly cheaper than AU. AU about $20, MU about $4

ATmega32U4-AU Image source: JLCPCB

ATmega32U4-MU Image source: JLCPCB

Actually, I was planning to order and mount the PCBA with ATmega32U-MU for the first board order, but when I ordered it, JLCPCB ran out of stock. So I changed the design data in hurry and made ATmega32U-AU by myself. Changed to the soldering policy. The inventory was revived when i was reordered, so I restored the design data again.

Precautions when ordering PCB boards

In this order, I tried decorating the board, but when I uploaded the Gerber data on the order screen of JLCPCB, I was puzzled by the preview being displayed with the black resist on the right side(back side) disapperaing. I thought that the design data was incomplete, so I changed the data serveral times and uploaded it, but the same phenomenon occurred. *The center of the surface on the left side of the image is circular and yellow, which is the correct state.

In the preview, the resist on the back side has disappeared, but if you check the Gerver Viewer on the lower right...

However, when I open JLCPCB's Gerber file Viewer and checked it, I noticed that the resist layer was hidden. I don't know the detailed cause, but this kind of phenomenon occurred when the Cu(copper foil) layer was arranged in a fine shape all over.

For some reason the registration layer was unchecked

If you check it, it will look correct

Just in case, enable layer display on Gerber Viewer. The preview image on the ordering screen did not change even if layer display was enabled, but I was able to confirm that there was no problem with the data, so I placed the order in this state.

I will post pictures of the actual product later, but the board I received was as designed and there were no problems. However, depending on the order data, the preview may be correct(the order data is incorrect), so if you cannot solve the problem or are worried, it may be a good idea to contact us!

Ordering JLCPCB metal 3D printing and correcting data

Now for the JLCPCB metal 3D print order for the main case. After uploadinf the design data optimized for 3D printing, the engineer in charge of comfirming the data contacted me that the output might be distorted. I received the same message when I printed out the rsein, and I replied thhar some distortion was acceptable,  so please print out the data as it is, and I started the production procedure. 

However, after that, we received another concern from the factory engineer that this frame-like shape was expected to cause a large amount of distortion.  At this point, I understand that the distortion is not tolerable, so U canceled the order.

Consideration for the cause of distortion

According to the engineer, the bottom part os small and has a frame shape, so there is a cconcern that the output will shrink toward the center (or spread in the opposite direction). In the laser melting method(SLM), which melts and solidifies stainless steel powder with a laser, the modeled object is slightly distorted due to the temperature difference between the hot part and the cold part. There are similar concerns with resin output of stereolithigraphy, but in this case of metal printing, the temperature due to laser irradiation is so high that it cannot be compared with resin, so if the shape is not appropriate, the distortion will increase. *If the shape is appropriate, it is an output method that gives very high accuracy. 

From here on, it's just my guess, but another  possible cause of excessive distortion is the output direction of the modeled object.

JLCPCB's metal 3D printer has an output area of 250x250x300mm. Since the long side of this case exceeds 250m, it must be installed in one of the following two direction with respect to the stage.

  1. Vertical placement with the right of left side of the case facing down
  2. The long side of the USB connector (rear) side is grounded diagonally to the stage.

In either installation method, the installaton area for the stage is small, and the bottom opening is free for printing. I am assuming that is what causes the distortion. Perhaps there is some twisting in addition to the contraction toward the center.

Even in the second installation direction, the open part of the bottom is output in a free state, so it is distorted

Respond by correcting the data

The best solution is to print the bottom part with appropriate thickness without removing it, but I can't do that because I'm framed for budgets reasons.

As a final solution, I divided the frame shape into four parts and output each side as one part. After receiving eash side, cut the threads and combine them with set screws. By the way, the fee is calculated by the volume of the output product, so it doesn't change whether you divide it. Rather, it was a little cheaper for the lower hold of the screw.

back and front. Although it exceeds 250 mm, it can be output by arranging it diagonally to the stage.


The parts on each side are fixed with M3 setscrews and combined.

When I uploaded the data divided in this way again, this time the data was approved without any problems. The warning mark in this image is due to the fact that the diagonally divided part was judged to be thin.

Successfully passed manufacturing inspection

By the way, I thought that outputing a part with only one side might distort (warp) during modeling, so I upload the following data with two sides connected with support material just in case.

Data that joins the back and front parts with support material and makes it a cylinder as a whole

Data in which the sides are joined with support material

However, these didn't pass the manufacture inspection. This is because, as started in this JLCPCB print guideline, metal printing does not support printing multiple parts that are connected with support materials. Make sure you check the conditions beforehand.

Please kindly note: FDM with ABS material and SLM with 316L meterial are not supported in this way.

In the case of metal 3D printing, data with support material does not pass the manufacturing review

As a result, we settled on printing the parts on each side individually.


Just when I thought that production had finally started, the following text was displayed on the order screen. It says that the delivery time will be longer than expected.

Audit failure reason, but message about late delivery

Recently, the production of SLM-316L material is under heavy load, the build time will be add 1-3 business day! Hope for your kind understanding! thanks! 

As shown in the following video, even JLCPCB, which has a large number of 3D printers, only have one printer that can output metal. So this kind of thing will happen if orders overlap. Relax and wait.

After few days. the status changed to In Production and production started.

However, the display showing the progess of manufacturing stopped at the first step (Data Processing) for about two days, so I wondered if the engineer was busy and could not update the status for the web, but the display changed. I continued to wait patiently.

It seems that my prediction was correct, and the status suddenly changed to waiting for shipment, and I was notified that the production was completed. When manufacturing resin parts, the status was updated from time to time, so it seems good to know in advance that this may happen in the case of metal output products.

The next progress was updated and comfirmed after shipping notification was received.

You can see that the output work was done in Data Processing ~ Printing

Arrival of substrate after specification change

As for the JLCPCB metal 3D printing, we had to recreate the order data and as mentioned above, the delivery was delayed, so we received the board with the changed specification earlier. The following three types of board were ordered. *The board and the 3D printing cannot be shipped together.

  • Bottom plate x 2 types
  • PCB that becomes the main body of the keyboard

This time it's not the familiar blue box


I tried using the option to remove the order number, which I did not use in the previous order, and the option to have papper sandwiched to prevent scratches when delivering the board. 

Please note that the option to insert paper cannot be used for boards that use PCBA because uneven parts are created (in the first place, it cannot be slected on the order screen).

Use the option to have the paper sandwiched to prevent scratches and the order number removal option

Boards using PCBA cannot use the option to sandwich paper

Copper foil (Cu layer) is used as one element for decoration, but it becomes silver (HASL wit Lead), silver (LeadFree HASL), and gold (ENIG) depending on the surface finish such as lead-free. ENIG is a process called electroless gold plating, and it's expensive, so I haven't tried it myself, but as far as I can see, it seems to be gold.

Experiment with decorative substrates using copper foil, resist, masks, etc.

The logo, which was silk-printed last time, is made brighter with copper foil this time (it's hard to understand because the photo is poor...)

The shape change of the USB connector part is also manufactured as intended.

Changed shape of USB connector

As shown in the following photo, the ATmega32U-MU has a much smaller component size than the ATmega32U-AU, so I think it will be an effective option for keyboards that do not have room for board component layout.

The bottom is ATmega32U-AU footprint, the top is ATmega32U-MU (installed). one size smaller

Arrival of JLCPCB 3D printed stainless steel case

This time, I wanted to receive it as soon as possible, so I used FedEx International Packet (delivered in 3-6days) instead of OCS Express (delivered in 4-8days). For comparison, the shipping cost was about $8 for OCS and about $15 for FedEx.

I received the delivery notice on 24/10 and arrived at hand on28/10. It was delivered to me by Yu-Pack as a domestic agent,

Regarding customs duty and comsumption tax, it seems that a payment form will be sent at a later date.

Shipped on Monday and arrived on Friday

Another layer was covered with packing material, preventing the model from moving.

Confirmation of model

Here is the sculpture I received.

It has a more solid and solid feel than I imagined, giving the impression of a lump of stainless steel. This 316L is called surgical stainless steel, and it is also used for jewelry because it is less prone to corrosion and rust and less likelty to cause metal allergies.

Almost no distortion to worry about

There was almost no distortion, which I was concerned about, and the printing was very accurate.

In the photom it's light gray, but in real life it's very cool because it's close yo the space gray colour in Apple's product lineup. I would probably polish the surface to give it a silvery sheen, but I like the darken tone of this matte so left it as is.

temporary assembly

Accuracy has been achieved, so it fits perfectly at the stage of temporary assembly

If it were made of aluminum, it would not be possible to put out this weight with this size due to the specific gravity.

In the case of metal 3D printing, electricaal discharge machining is used to separate the model from the stage (wire electrical discharge machining). As a result, the surface that was in contact with the stage during printing will have sawtooth marks as shown below. However, since the surface trreatment is applied after cutting, it does not have the impression of being rough, and it is relatively smooth. 

oblique cut

It's hard to see, but there are traces of cutting parallel to the long side.

If it is installed at an angle to the stage, this mark will be slanted, and if it is installed at a right angle or horizontally, it will leave a mark of teeth that are perpendicular or horizontal to the modeled object.

The red straight line is the wire for cutting and the direction of travel. When placed diagonally on the stage, the cut marks remain diagonally

The other sides have a cast metal-like texture. Is it like this becuase the stainless steel powder is melted and solidified by the laser?

cast iron texture

The output mark remains diagonally, but it is an inner part, so it does not bother me

Additional work and assembly

After confirming each part, we will assemble it. Each side is assembled with M3 setscrews, and a bottom plate made of FR-4 is used to secure the entire frame for rigidity. As a piolt hole, a φ2.5 hole was provided in the ordering data, but the accuracy was perfect, so it was easy to cut the M3 tap.

I was worried because stainless steel is hard, but the tap cut easily

Combine with the M3 setscrew that was lying around at home

Sorry for the out-of-focus

Only 4 setscrews are used, which is the minimum required, so if you use only these screws, the entire frame will be loose. For thar reason, we combined it with two FR-f bottom plates, which are more accurate than 3D printing, to fix the entire frame and increase its rigidity.

The plates are attached together with double-sided tape.

Rubber feet on the bottom plate

The bottom plate and case are also fixed with double-sided tape. Also suitable for screwing

This completes the assembly of the case. With only 4 setscrews, it was wobbly, but by combining it with a plate, we were able to secure rigidity and it became ticking.

Top side (There are traces of double-sided tape on the inside of the back part)

bottom side

The weight has been increase by about 100g from the stainless steel parts alone, making it even heavier.

Plenty of weight for being thin and small

Combining the case and keyboard body

Next, we will combine the case and keyboard body.

This time, the internal structure is simplified to make it as thin as possible. Therefore, we designed a case that use an O-ring mount as a cape method. Nornally, an O-ring like the one shown below is wrapped around the outer circumference of the gap between the switch plate and the PCB of the keyboard body, and the keyboard body is fixed by friction between the O-ring and the case.

However, this time, when I happened to be wandering around a home center near my home, I found a super-cheap product called a mesh holding rubber ( a product for fixing the screen door to the frame). I thought that this could be an alternative to O-rings, so I bought it as a trial.

There are also variations in the diameter of the mesh holding rubber, so If you are using an acrylic switch plate or a low-profile keyboard, you may be able to use it.

This time, will will use Cherry MX compatible key switches, and since we are using a 1.6mm thick FR-4 switch plate, we will use 3.5mm diameter rubber.

Urethane foam is pasted on the top of the case to prevent the PCB from sinking too much.

The distance between the switch plate and the board body is 3.4 mm, but it can be easily wrapped around because it is crushed moderately.

If I would it as it is, the rubber would be crushed too much dur to its hollow structure. Therefore, I decided to wind a thick water string (kite string) lying around the house about 2 times to increase the volume, and then wrap the rubber on top.

Increase the volume with water thread so that the rubber is moderately exposed to the outside. The knot was in the way, so I cut it after this

Now the rubber protrudes moderately from the outer shape of the PCB and can be fixed to the case. They are evenlt spaced inside the case so they stay in place even when you type or turn them over. You can easily replace the key switch,

I haven't test yet, but it mat be possible to ground it to the case without the need to winf the string with a thick holding rubber such as 4.5mm or 5.5mm.


Complete keyboard "inn v2.1"

Here's an overview of the complete keyboard: the width of the bezel has also become thin to my liking, and I was able to eliminate the dissatisfaction of the previous case that led to the change. As for the colour of stainless steel, it is unexpectedlt very astringent, and I really like it.

The bezel is thin and refreshing

I want to try beige and dark color keycaps

The tilt angle is 3.4°. The first prototype was easy to use, so I adopted that angle as it is.


Bottom. The gap between the side parts is because the set screw on hand is too short and can not be tightened.

Impressions atfer completion

The case of this time is based on the size of the first prototype case made by processing aluminum angles. As a result, we achieved a significant increase in weight with almost the same size, and were able to acquire a sense of stability and rigidity.

The bottom is the very first prototype made with aluminum angle. very light

Personally, my ultimate goal is to manufacture a case by cutting aluminum, but if it is for personal use and not for sale, I feel that metal 3D printing is sufficient,

Above all, this compact keyboard has thin bezels and a low height, but the fact that it is so heavy due to its specific gravity has led to more satisfaction than I imagined. Unexpectedly, it was a discovery that if I made it with this design using machined aluminum, I might be dissatisfied with the weight.

Even with the same volume, stainless steel weighs about three times as much as aluminum.

This time, due to budget constraints, we used a frame-shaped case with an open bottom, so we had to print in separate pieces and assemble it. However, due to the output area, the limit is the diagonal size of the stage whose long side is 300 mm or 250 x 250 mm in the Z-axis direction, so if it is not divided, the limit may be about 65%.

If you are interested, why not try making a case with JLCPCB metal 3D printing?

JLCPCB, where I made PCB and 3D printed this time, also distributes $54 new user coupons!