electronics design

This week started up by designing a PCB for an ATTiny44 chip in Eagle. The board needed to have 1 led and 1 switch. I decided to put two leds and a switch instead. Eagle went pretty smoothly. The main things I learned were :

- use the tool in the menu bar to check for errors Tools > Errors
- export high res pngs or else the fab software won't be able to read it.
- make sure your traces are >= .03"

Below are some images of things that went wrong.

Mistake #1. For some reason the bottom line of the outline wasn't thick enough to be read by the fab module. I thought it looked funny in the tool path, but then I ran it anyways. I ended up making another file with a thicker bottom line. Then Che-Wei told me that you can just black fill everything you don't want there, and the software knows to just cut once along the line instead of doing offset paths.

Mistake #2. Another problem with my first board was that the traces were too thin. I made them .01 in eagle. Too SMALL! I then tried .3 and that worked. I think you can get down to .25, but thinner than that you are risking them breaking off.

Mistake #0. Here was a board I designed and cut a couple weeks ago. It was my first time using eagle and I exported a PNG of 72DPI. You couldn't tell this would happen from the Fab preview. Even though the file path looked crazy. I cut it anyways because I didn't know better. But now I know to pay attention to the file path, it is actually showing you what it's going to do. It's not magic.

Recap. Here are my two boards, the bottom one has traces that are .03". You can adjust your trace size with the "wrench" looking icon and a menu pops up and you can select your desired width. From there, just click on the tool paths that you want to make thicker.

On the left below is my final board, on the right is my final board with a simple program running. I used the hi-lo tech attiny tutorial to use the AVR MKII to program my board. The program reads the input from a button press to toggle which led goes on.


	int led11 = 3;
int led10 = 2;
int switchIn = 7;

int activeLED = led11;

void setup() {    
  
  pinMode(led11, OUTPUT); 
  pinMode(led10, OUTPUT); 
  pinMode(switchIn, INPUT); 
  digitalWrite(switchIn, HIGH);
  digitalWrite(led11, LOW);
  digitalWrite(led10, LOW);
}

void loop() {
  
  int buttonState = digitalRead(switchIn);
  
    if(buttonState)
    {
      digitalWrite(led11,HIGH);
      digitalWrite(led10,LOW);
    }
    else
    {
      digitalWrite(led11,LOW);
      digitalWrite(led10,HIGH);
    }
  
}

Straight into LTSpice extra + unresolved.

The first thing that I decided to do this week was start playing around with LTSpice. I got really excited by it after the great tutorial Brian ran. One of the reasons I wanted to try out LTSpice was for a piece that I am working on that will show how a quartz crystal oscillator works in a wristwatch. These components are in so many of our devices, but few people, including me really understand how they work.

One of my first steps in making this piece was to learn about how a crystal is made to oscillate, and then how that oscillation is divided into a time signal that can interface with the mechanics of a watch. Once I did this (I can go into this somewhere else because I learned some interesting stuff, but it's not that relevant here), I designed a circuit to try this out for myself. I based this circuit off of this circuit. I wanted to pair this down even more so that I could model the bare minimum required to make a crystal resonate.

I thought this would be super straight forward, but it turns out that they don't really give you an 'off the shelf' crystal in LTSpice. Well they kind of do, but you can't just specify a frequency and voila it works. Instead, you have to build a resonator out of resistor, inductor, capacitor circuit. So you specify all of those values. I plugged these into an online calculator and got close to my desired 32,768Hz, which is the frequency used in a watch crystal.

This is testing out an RLC circuit, I couldn't figure out how to get values out.

This is testing out a "pierce oscillator", a circuit that uses an inverter to jumpstart the oscillation. I couldn't get LTSpice to show me anything here either. For this, you can actually use the oscillator part, but you have to give it resistance, inductance + capacitance value. It's essentially interchangeable with an RLC circuit.

This is something I would still love to get working. This could be a very powerful tool, but I didn't have time to dig much further this week.

« Back