Week 5

Fabricating PCB

Jari Pakarinen, Fab Lab Oulu staff and Fab Academy 2017 alumni, gave us the local lecture about PCB fabrication. There are several techniques to create PCB. As examples of the techniques of subtracting, vinyl cutting copper sheet, laser cutting and milling circuit board. As examples of techniques of adding, painting with conductive ink, sewing with conductive thread (E-textile) and plating for mass production. For this week’s assignment, I used milling machine to fabricate PCB. PCB consists of layers of different materials. Milling machine removes copper layers from the board to create paths and pads with remaining copper layer. We have Roland SRM-20 in Fab Lab Oulu.


As a milling bit, it is possible to use both flat bits and V-shape pointed bits. The diameter of the flat bits is 0.4mm, the diameter of v shape is from 0.2 at point to 0.5mm at upper. I used V-shape pointed bits.


To fabricate PCB, we need to prepare file as a specific format for the milling machine. For Roland milling machines, file should be prepared as .rml format. It is possible to convert the file in Fab modules.

Converting files

First, I downloaded PNG files of Trace and Outline from Brian's website. I went to Fab modules to change the format of the files for PCB fabrication with Roland SRM-20. I imported PNG file of trace by selecting input>image (.png). The image appeared on the screen.


I selected Roland mill (.rml) as output format and PCB traces 1/64 (0.4mm) as process.


On the right of the image, there are sections where I can change settings for manufacturing. In input section, I can check the size of the PCB and change dpi. With Roland SRM-20, it removes copper layer around the parts which are shown as white in the PNG image. If the image shows opposite colour, I need to invert the image. In output section, I chose SRM-20 as machine and set x-, y- and z- position as 0. Here I can change, such as cut depth, tool diameter, offset, etc. The milling machine repeat following the tracing paths according to the settings. I can also set offset overlap (%), which decides how much the bit overlap the trace paths in each time. I used 0.13mm for cut depth and 0.4mm for tool diameter. I clicked “calculate” in process section, then it showed the tracing paths.


I saved the file as .rml format by pressing “save” button. Next I converted outline image. I chose PNG image for outline as input and Roland mill (.rml) as output format.


I selected outline 1/32 (0.79) as process. As I did for the trace file, I selected SRM-20 as machine and set x-, y- and z- position as 0.


I set cut depth as 0.6mm, and stock thickness as 1.7mm. The thickness of the board is 1.6mm, so with setting thickness as 1.7mm, it will be definitely cut. With this settings the bit cuts 0.6mm in each time, and will repeat three times till it cuts out 1.7mm thickness. The tool diameter can be set a little bit smaller than actual tool diameter. The offset will be divided into both side of the outline, so if I set tool diameter smaller than the actual one, it cuts out more than the machine expects. If the components are placed close to the outline, it causes problems but “cheating” machine a little bit is sometimes useful, for instance when I get error of calculating outline by Fab Modules recognising it is too close to cut. But this time I set tool diameter as 1mm which is actual tool diameter. I pressed “calculate” and saved the file as .rml format.

Settings for milling machine

First I fixed the position of the board on the wooden board with double-sided tape. Copper layer of circuit board is about 0.03mm, so it is important to keep the board flat when we manufacture PCB. I carefully applied the double-sided tape without overlapping each other.


I placed it near the lower left corner of the wooded board and set the wooden board in the machine using screws.


Trace should be done first. I attached 0.2-0.5mm milling bit using hex key.


I opened VPanel for Roland SRM-20 in the computer which is connected to milling machine.


I moved x- and y- position to the lower left corner of the circuit board by clicking the arrows, and pressed “X/Y” button on the right of the screen to set x-, y- origin.



I carefully lowered the z- position till the milling bit came to very close to the surface.


I loosened the screw which hold the bit using hex key and carefully placed the bit on the surface of the board. I fixed the z- position by tightening the screw again.


I pressed “Z” button to set z- origin.


After I set z- origin, I moved z-position slightly higher from the surface to avoid scratching the surface when I started operating. The milling bit can be easily damaged by dropping, so be careful not to drop it!
I pressed “CUT” button in VPanel, and deleted all the previous cutting files. I clicked “Add” and selected the trace file .rml format. I pressed “output” and tracing started.


After milling was done, I vacuumed dust on the surface.



Next I cut outline. I changed the milling bit to flat one for cutting. I adjusted only z- position following the same steps as above. I pressed “CUT” button and deleted the previous file and select outline file as .rml format.


After cutting was done, I vacuumed the dust again. I removed the PCB from the board using the spatula kind of tool.


I should always make sure it is cut through before try to remove. Once I tried to remove by force, and broke it. Later I found out the bottom was not cut through. If I had checked and found out it was not cut through before I tried to remove it, I could cut it again from the same x-, y-, z-origin.

Fixing the problem

After I checked the board with microscope, I found copper layer was not perfectly removed and tiny copper lines were remained on the PCB. There are several possible reasons:

• When I screwed the bit, I might moved up a little bit z- position from the surface.
• The board was pulled up a little bit when other people took their PCB and I set z- origin on the higher point.
• The machine is not accurate.

I decided to make a new board with different settings. This time when I converted the file in Fab modules, I changed some settings to make sure it would be milled properly. I changed cutting depth from 0.13mm to 0.14 so that it would trace deeper, and tool diameter from 0.4mm to 0.35mm so that it would trace with more offset overlapping.


I followed the same steps as above and made new PCB.
Here is the comparison of two boards. The difference can be seen clearly.

Soldering the components

I cleaned the PCB with water and scratched the surface with a steel wool to remove dust and photo sensitive layer on the top of the PCB. Also I removed the extra copper with the cutter.



The board is ready for soldering.
I collected following components and put on the double-sided tape next to the names so that they will not be lost.

• 1x ATtiny45 or ATtiny85
• 2x 1kΩ resistors
• 2x 499Ω resistors
• 2x 49Ω resistors
• 2x 3.3v zener diodes
• 1x red LED
• 1x green LED
• 1x 100nF capacitor
• 1x 2x3 pin header

First I wet the sponge. I should always clean the tip of the iron with the sponge so that extra solder is not on the iron tip. I set the PCB on the PCB holder so that it does not move while soldering the components.


I checked the components and make sure their locations and orientations. ATtiny, LEDs and diodes had to be soldered in the specific orientations. I referred to the following schematic and board layout from Brian’s website.



There is no specific order to solder, but as Neil explained in the lecture, it is easier to solder the components which are in the middle of the PCB first, outside ones later, and the lower height components first, the higher ones later. I started soldering 49Ω resistors (R3, R4) and capacitor, then ATtiny45. I soldered 2x3 pin header at last.
To solder the components, first I heated the pad. If the pad is not hot enough, solder sticks on the iron but not on the pad. Then I attached the edge of the solder wire on the border of the iron and the surface of PCB. I wait a little bit till enough amount of solder melted on the pad. Then I took the wire off and slightly slid the iron to spread solder equally on the pad. I placed the component leg carefully on the pad with the tweezer. I heated and melted solder without adding solder to fix the component. I soldered the other leg and then soldered the first leg again to make sure the leg is attached on the pad. The components can be damaged by heating too much, so I should be quick when I heat the pad with iron. Also for the components which has more than two legs, I should solder diagonally to avoid heating one side concentrically. It is also easy to take balance of the component and keep it flat on the surface.
During I was soldering, copper was peeled off. Probably it was because I heated PCB too much and the glue was dissolved. I set the iron 350℃ but probably I was too slow to solder.


I made another PCB and soldered with 340℃ iron.
After I soldered all the components, I checked again the locations and orientations of the components comparing the schematic, PCB layout and the photo. I also checked my PCB visually to make sure all the legs were connected on the pads with enough amount of solder and all the components were flat on the surface. After that I checked the short circuit between VCC and GND using multimeter.

Programming a programmer

This PCB is going to be an AVR programmer which can program a microcontroller. To be an AVR programmer, this PCB need to be programmed by a working AVR programmer. Also for this process, I referred to Brian’s website.

Setting development environment

First, I downloaded the latest version of CrossPack. CrossPack is a development environment for AVR microcontrollers running on Mac OSX. It allows to compile .hex file which is going to be used to program the ATtiny45.


After install CrossPack, restart Mac. I skipped this process and had a trouble to compile .hex file later. As it is written in Read Me, after software was installed successfully, I needed to restart Mac to complete the process. Second, I downloaded firmware source code from Brian’s website and saved it in my FabAcademy2018 folder. I opened Terminal and typed following command to change directory to the source code directory.

cd fts_firmware_bdm_v1

I typed following command in Terminal to build .hex file.

make

In the source code folder, .hex file was created.
I opened the Makefile and checked the programmer.

PROGRAMMER ?= usbtiny

I was planning to use the usbtiny to program my PCB as it was default, so I closed it without change.

Programming ATtiny45

I plugged a working AVR programmer and my PCB into USB hub. As I mentioned above, before I figured out installation of CrossPack had not completed, I couldn’t run make command. So first I tried with Windows computer to program my ATtiny45 to make sure the PCB is fine.


Later it succeeded also with my MacBook.
I ran following command to program ATtiny45 with the contents of the .hex file. Several progress bars were appeared.

Make flash

I ran the command to set all the fuses except the one to disable the reset pin. Several progress bars were appeared.

Make fuses

After I programmed the ATtiny45, I checked it to make sure it is working. I opened System information (Utilities>System information). I chose USB from Hardware section in the left and found usbtiny. I first plugged into USB3.0 but it did not recognise USBtiny. I changed USB port to USB2.0 and it worked.

Characterising milling machine: Group website

Process

To complete group assignment of this week, I worked with Heidi and Peetu. We examined how cutting differs with different settings and tools. We used Roland SRM-20 for this test.


We downloaded PNG files of tracing and outline from Fab Academy website. We tried five different settings of milling bits and cutting depth.


Our assumption was when it cuts deeper, the width is wider than it supposed to be, and when it cuts shallower, there will be copper left on the surface.


Here is the result. With the tool diameter setting 0.4mm, the thinnest gap the milling bits made was 0.016 inch = 0.4mm (fifth from the right) with both v shape milling bit and flat milling bit. It was because the thinnest tool path made in Fab Module was 0.4mm. It is possible to make the tool path for less than 0.4mm lines if smaller tool diameter is set in Fab Module. When the tool path is made, V shape milling bit might make a gap thinner than 0.4mm, but it is better to use 0.4mm as the thinnest gap in PCB design.
The thinnest trace was hard to measure. There might have been tiny differences in our manual operation and it did not give the clear results, but with flat milling bit, 0.14 mm cutting depth, the thinest trace was 0.004 inch = 0.1mm (fourth from the left). Basically the tool diameter doesn't matter that much to create the thin trace, compare to creating the thin gap, since the milling bit removes the copper around the trace lines. However, as our result shows, the outcomes can be affected by operation of the machine and too thin lines can be removed. It would be safe to use certain thickness, according to our result at least more than 0.1mm, for trace lines in PCB design.


After cutting boards, we looked at the surface with microscope which has measurement scale.


Tool difference
With v shape bit, we could see two lines on the edges of the trace. Since the bit is v shape, the tracing width is different between the deepest point of the bit and in the point of the surface. The point which is closer to the surface, the trace width gets wider. On the other hand, the trace lines of the flat milling bit is just a single line on the edge.



Shallow cutting depth
There was copper left on the surface of some parts of PCB. However, probably it was not because of the tracing depth, but because the surface was tilted, since copper was left only some parts.


Deeper cutting depth
It looked different, with the deeper cutting depth settings, more the copper layer was removed and the tracing paths looked wider. However, it was hard to measure the exact width of the tracing paths with microscope.

Comment

When we cut the outline with tool diameter 0.79mm, one edge was not cut. It was also mentioned in Nordic review. It was because in Fab Modules there was not enough space in the bottom and Fab Modules recognised that the cutting line is too close and cannot be cut with 0.79mm milling bit. So we changed the tool diameter to 0.5mm and it cut. It was a little bit difficult to characterise the machine. Milling process requires precise operation and it was difficult to set the same environment for each settings. Changes can be made from very small difference of each operation.

Reflection

This week, I made four PCB. PCB manufacturing process is not so tricky but the copper layer is very thin and requires precise operation of the milling machine to create PCB. If there is a slight mis-adjustment of the height of the milling bit, the copper will not be removed properly. It was a little bit challenging for me before I got used to the level of the preciseness, ended up making four PCBs. And it was difficult to figure out what exactly was the problem since less than 0.1mm difference can cause a failure. I could only come up several possible reasons for the failure. During the programming process, I was also having a trouble to set up environment for programming. There was not so much information regarding programming with OSX. But I enjoyed the process and not I am more confident in PCB fabrication process.

Files

Here is the files for trace and outline of programmer (.rlm format).

• rlm file for trace
• rlm file for outline