Saturday August 16th, 2014
Today, I finished the initial PCB draft. This required deciding what pins on the microcontroller would be used for which hardware interfaces. The PCB is designed to be mounted on the back of the LCD unit. There are 10 pins in the interface between the two, and I figured that they would be sufficient to handle the weight of the PCB with the other components. The mounting holes for the hardware are on the LCD unit. In theory, if the LCD is mounted to the enclosure, the PCB will not need any explicit mounting because it is firmly attached to the back of the LCD via the 10 pins and the associated solder joints. The two pieces will be connected with a set of header pins.
Also, I was able to arrange the components on the board close enough together to reduce the board foot print. This is great, because it will reduce the cost of the boards. It also means that the device will fit in smaller enclosures. The hardest part of this was routing all of the traces without any vias (it was not mandatory, but I kind of took it as a challenge). At first, I had the LCD module backwards, and the routing was really easy, but once I realized my mistake and flipped it, it became an awful mess. I am so glad I can do this on a two sided board! (The boards I drafted for classes were only one sided.)
Here are some screenshots of my work:
Electrical Schematic
PCB Draft
Notice all of that empty space on the left side of the PCB draft? That is about 1/5th of the width I originally allocated to the board (that could amount to a substantial cost savings). Of course, I will be eliminating that wasted space before sending this to a commercial PCB maker. I am also considering adding a ground plane to at least one side of the board. You might notice that some of the silk screen (the text and outlines) stuff overlaps a lot. This will make no difference as far as functionality goes, but I also plan on doing something about it before sending it to a commercial PCB maker. (I'll probably reduce the font sizes first and then move the BATT text to the top of the battery module and the U0 text to the bottom of the voltage regulator module (LD33V).
(In case you didn't already know, all of these files will be released under an open source license at the conclusion of the project. Also, in case you were about to ask, I did all of this in KiCad. KiCad is an open source PCB drafting suite. I don't know how it compares to commercial software like Eagle, except that free is way cheaper, especially for commercial use.)
Here is my todo list on the PCB. As items on the list get done, they will be removed and I will add anything interesting to the above section.
- Order prototyping parts (photosensitive board, developer solution, and etchant; this is waiting for Kickstarter funding)
- Create rough homemade board prototype
- Make any necessary revisions (and maybe re-prototype)
- Send out revised design to be prototyped by a commercial PCB maker
- Make any necessary revisions (and maybe re-prototype)
- Get estimated for mass production of boards from a medium scale PCB maker
Rather then paying for all the stuff to manufacture low quality home made prototypes, you should use a proto-pcb company like OSH Park (www.oshpark.com) where you can upload your Eagle .brd file or Gerbers in a .zip file and they will create professional grade prototypes in a smart purple solderscreen in batches of 3 at just $5 for 3x 1x1inch boards the turn around time within the US is about 10-14 days from order to delivery. Your design would prob cost you less than $3 for 3 prototypes and you can order 3,6,9,12,etc... Much much nicer than home made etched boards and cheaper.
ReplyDeleteAlso on your early prototypes you could add a prototype area to the board using up that space on the left so that you can add components to the circuit and connect via jumper leads if needed, then build it into the next revision.
Some examples of their PCBs : http://jason-webb.info/wp-content/uploads/2013/01/Picture-457-1024x768.jpg
Good luck.
The reason for the initial homemade prototype is the short turn around (I should be able to make a board in a few hours). This first prototype will be for testing the electrical design (make sure I did not make any mistakes) and component placement.
ReplyDeleteThe second prototype phase will indeed use OSH Park to make three professional prototypes. Once I have made sure the design is electrically sound, a 10-14 day turn around will be entirely acceptable. (In fact, my worst case cost estimate uses OSH Park prices for the board. The worst case estimate assumes that I will not need enough boards to order in bulk from a large scale producer.)
As far as price goes, it will be about comparable or maybe slightly cheaper for the homemade board, but you are absolutely correct that the professional boards will be way nicer. (It also helps that I actually have a piece of photosensitive board that is somewhere around 7"x4".)
As far as room for more components, I won't need it. The current design uses all of the available pins on the MSP430 (the unconnected pins are all for programming the device and cannot be used for GPIO). Also, that empty area is worth almost $2.00 for OSH Park pricing, so I am better off eliminating it unless I find a pressing need for it.
Thanks for the comment!