Work on the Synaptron Mega, based on our Synaptron Micro design, is moving forward. You can see the first prototype in the post Synaptron Mega Enters Development Phase. My initial tests focused on basic functionality and firmware modifications. The prototype performed well. With minor hardware modifications I think we could have moved forward with a production version, but I felt a strong desire to make changes before completing my tests. The image above is a first attempt at board layout for the changes, and is probably pretty close to actual size. The pin-to-pin connections have not been made, and that constitutes another 4-8 hours of work.
There were several areas I wanted to address after reviewing the first iteration of our prototype. The bullet points are from my previous blog entry. In red are my additional notes.
- Tighten up the TSSOP footprint models (the pin landings are too long).
I made some modifications to the footprint library for two of the TSSOP parts, but will likely need to shrink them again based on the new overall layout. Making them too large allows them to shift during assembly and uses much needed space on the PCB. Making them too small can create production errors, and I still have to assemble the prototypes by hand (under a microscope).
- Use a larger footprint linear regulator or consider a switching regulator. With this design I assumed the on-board 5V regulator would be used to run a motor’s quadrature encoder, and if the extra current load is causing some heating at 12V, at 24V it’ll be worse. Maybe even twice as bad
The main goal here was to allow the use of a 24V motor supply without causing serious overheating of the on-board regulator. The original design had a low cost 5V regulator in a TO-252 package. In previous designs I’ve had good experiences with the Infineon TLE4270 in the TO263-7 package. This part is specified to work up to 36V and has a maximum voltage rating of 42V which matches the H-bridge IC (also an Infineon product). This regulator IC costs more, and so I made some other tradeoffs to reduce the component count to accommodate the increased budget for the regulator.
- Change the 0402 discrete components to 0603 to give them a little more thermal mass (there’s plenty of room).
I made this change, and with the additional attempt to shrink the board, the whole “there’s plenty of room” opinion is out the window.
- Consider shrinking the board. This design needs to remain on a 2-layer PCB, and the mounting holes need to match standard cooling fan mounting holes. Shrinking the board may be tough, but I’d like to see it about 20% smaller.
I reduced the board from one that can carry a 50mm square fan, to one that can carry a 40mm square fan. If you compare the screen capture above with the photo of our prototype in the previous post you can compare board size by referencing the mounting holes in relation to the 20-pin header on the left of the image above.
- See if a bigger terminal block will fit.
I was able to fit a 10+ amp terminal block on the part. More importantly, the one I was using did not allow for very large wires to be used, which was the main problem I wanted to address.
Can I make this design work? If I went to a 4-layer printed board it would be pretty easy, but I’m shooting for a 2-layer printed circuit board for cost reasons. With only two layers I’m not sure I can pull this off. There will be a need for more modified part models, connectivity changes, and I have to use tiny traces and vias (connections between layers).
I do feel better about the new direction this design is pointed in. It is more aesthetically pleasing (to me), and the engineering tradeoffs in size, cost, and performance are much closer to the original design concept. We’ll see if I can make it work.