Week 08 - Electronics Production

Group Assignment:

• Characterize the design rules for your in-house PCB production process: document feeds, speeds, plunge rate, depth of cut (traces and outline) and tooling.
• Document the workflow for sending a PCB to a boardhouse
• Document your work to the group work page and reflect on your individual page what you learned

Please refer to the group page to read about our group assignment.

Individual Assignment:

• Make and test a microcontroller development board that you designed

Learnings from this week's group assignment

The main learning from the group assignment was to familiarize myself with the Roland mill and the workflow from KiCAD to the PNG image and finally toolpath. I also learned about the good finish a 1/64'' end mill can achieve in milling traces and how poorly the v-bit performed in comparison.

About our PCB production process

At Fab Lab Ísafjörður we have two main ways of producing PCB.

• Milling copper plated sheets
• Cutting copper film in the vinyl cutter and transferring it to a nonconductive base sheet

All components are placed manually and either soldered individually by hand or by applying soldering paste and heating.

Exporting the files for PCB milling

The whole process of creating PCBs and milling with the Roland mill is described in a series of tutorials linked here.

There are some changes from the previous versions of KiCAD to the latest v9.0.

For exporting the copper layer and edges go to: File - Plot. In my case, I have a one sided PCB so I only need the F.Cu layer and Edge.Cut. I selected Edge.Cuts to be plotted on all layers - by that I get both layers in one svg file.

Now some manual work needs to be done to create png images that can be translated to g-code.

I'm going to use a 1/64'' (0,4 mm) endmill for the traces and a 1/32'' (0,8 mm) endmill for cutting the outlines.

Because I want to cut outside the contour line, I created a path from the contour and added a 0,8 mm offset to it.

Then I changed the page size to match the outlines of the cut.

Afterwards, I exported the two different colours into different png raster images with 1000 px resolution

Then I opened the files in paint.net to change the colours.

Setting up the milling machine

Here is the setup of the Roland MDX-20.

The machine has to be turned on and set to machining mode by pressing the "view" button.

Warning

The X/Y Zero is not where you would expect it (in the lower left corner of the plate). There is an offset of about -30/-30 to it.

Before abjusting z-zero - put the milling bit into the collet - move the spindle above the plate and lower the z-axis with the arrow buttons. Loosen the stop-screw of the collet again and carefully drop the bit the last few millimeters.

Typical speed is 4 mm/s and cutting depth for the traces 0.1 mm.

In the computer you have to start the FabModules JSON server and next the FabModules homepage and select png.

Then select Roland-mill and Traces 1/64.

Adjust X and Y to set the origin to a suitable point on the stock.

Finally calculate and check the toolpath. The predefined machine settings should work fine.

Milling the PCB

You can click on "send" and the operation will start right away.

Milling the traces took 20 min.

Then I changed the tool, readjusted z-zero and opened the Edge.Cut image and repeated the process.

I had to split the holes and the contour up into seperate files, because the ordering of the traces didn't work in the way, that the holes would be cut first.

Here is the board out of the machine, with clean cuts.

Soldering the components

Here I started soldering the first small components onto my PCB.

Then I continued with all the parts I could find.

At this point I realized, that something was wrong with my board. There are some traces connected to each other, where they shouldn't be connected.

That might be caused by some tolerance issue from exporting the traces and the spacing between the traces set to the same width as the end mill.

I was able to resolve it by cutting the copper with a steel ruler and a sharp knife. Next time I have to remember to double check this when calculating the toolpath.

Then I discovered one more issue, which was a connection missing between the 5 V USB power and the 5 V circuit on the board, because it can be either powered by 12 V or USB. I solved that using a jumper wire.

When plugging it into the USB port the green and yellow light show that there is voltage present. However, my idea was to use the yellow LED only for indicating 12 V power supply. Somehow the 5 V can flow the opposite way through the voltage regulator and powers the yellow LED circuit with 5 V, which makes it light less bright then with 12 V.

The missing components for the SDI12 interface haven't arrived yet, because they are not part of the standard inventory. I will finish the board later, but it should work for some simple testing.

Programming the SAMD11

Because I didn't know how to program the board I searched and found this page showing the different types of programmers. So first I need to programm a bootloader to the chip to make it recognizable as usb-device.

Burning the bootloader

It seems, that I have two options.

• SWD D11C
• swd+uart rp2040

After talking to my instructor, he gave me the Fab SWD programmer and I used these instructions to learn how flash the bootloader to my SAMD11C.

Here are the main steps:

• Install Fab SAM Core in Arduino IDE
• Connect Fab SWD programmer to USB port
• Connect external power to SAMD11C
• Connect both boards via SWD (four wires)
• Select Board and Programmer in Arduino IDE
• Double-check all settings
• Burn Bootloader

First the PWD clock frequency wasn't selected right and the bootloader burning failed. I found this page with a more detailed description and all settings explained.

Settings in Arduino IDE:

Terminal after succesfully burning the bootloader:

Once the process is finished, you can connect the PCB directly to the USB port and it should show up as a COM-port device.

Sending a program

I made this simple code to listen to the pushbutton and responde with a blinking LED.

To make the program work I had to figure out the right pin numbers and used the pinout diagram from Adriáns page

//Defining which pin to use for LED and button
#define LED 15
#define BUTTON 2

//Variable to count number of button press
int cnt = 0;
//Timer before executing the blink
int cntpause = 0;

void setup() { 
  pinMode(LED, OUTPUT);
  pinMode(BUTTON, INPUT);

  for(int i = 0; i < 3; i++){
    digitalWrite(LED, HIGH);
    delay(500);
    digitalWrite(LED, LOW);
    delay(500);  
  }
}

void loop() { 
  //Check if Button is pressed -> counter +1
  if(digitalRead(BUTTON) == LOW){
    cnt++;  
    cntpause=0;
    while(digitalRead(BUTTON) == LOW){
      //wait
    }
    digitalWrite(LED, HIGH);
    delay(50);
    digitalWrite(LED, LOW);
    delay(50);      
  //Check if button has been pressed at least once and if timer has reached 200 -> then start blink
  }else if(cnt > 0 && cntpause > 200){
    for(int i = 0; cnt > i; i++){
      digitalWrite(LED, HIGH);
      delay(500);
      digitalWrite(LED, LOW);
      delay(500);     
    }
    //reset counter and timer
    cntpause=0;
    cnt=0;
  //Check if button has been pressed at least once and increase timer counter by 10ms
  }else if(cnt > 0){
    cntpause++;
    delay(10);
  }

}

The code should count how often the button is pressed and return the number counted by blinking the red led.

Tip

Sometimes, when playing around with the board I couldn't send a program anymore. In that case I just burned the bootloader again and it worked fine after that.

Here is the example program running on the board, showing the function.

Issues with the board

There are two issues with the board I discovered while programming and testing.

1. The small 5 V voltage regulator get's very hot, when the board is connected to USB. In that case it draws a lot of current (250 mA).

I think I might need to add a diode at some place to prevent the voltage regulator from reverse flow.

1. The push button has no pull-up or pull-down resistor. The digital input pin of the SAMD11 is very sensitive and when connected as I did, the state of the input can be indifferent or triggered by noise. I could trigger the button by moving my finger close to it - so I accidently made a touch sensitive button... The right way to do this is to use a pullup or pulldown resistor.

Fixing the board

In regards to the two issues above, I found a solution to both of them.

1. I added a small diode (0,5 W) to the output side of the 5 V voltage regulator. That stopped the part from consuming way to much current and heating up, when powered by the USB.

2. I added a 10 kΩ pulldown resistor to the pin of the push button. I also changed the code a bit, because I found an error there as well. I need the input to be HIGH to trigger the counting, not low.

Here is the updated code and the PCB with modifications.

//Defining which pin to use for LED and button
#define LED 15
#define BUTTON 2

//Variable to count number of button press
int cnt = 0;
//Timer before executing the blink
int cntpause = 0;

void setup() { 
  pinMode(LED, OUTPUT);
  pinMode(BUTTON, INPUT);

  for(int i = 0; i < 3; i++){
    digitalWrite(LED, HIGH);
    delay(500);
    digitalWrite(LED, LOW);
    delay(500);  
  }
}

void loop() { 
  //Check if Button is pressed -> counter +1
  if(digitalRead(BUTTON) == HIGH){
    cnt++;  
    cntpause=0;
    while(digitalRead(BUTTON) == HIGH){
      //wait
    }
    digitalWrite(LED, HIGH);
    delay(50);
    digitalWrite(LED, LOW);
    delay(50);      
  //Check if button has been pressed at least once and if timer has reached 200 -> then start blink
  }else if(cnt > 0 && cntpause > 200){
    for(int i = 0; cnt > i; i++){
      digitalWrite(LED, HIGH);
      delay(500);
      digitalWrite(LED, LOW);
      delay(500);     
    }
    //reset counter and timer
    cntpause=0;
    cnt=0;
  //Check if button has been pressed at least once and increase timer counter by 10ms
  }else if(cnt > 0){
    cntpause++;
    delay(10);
  }

}