Week 4 (Electronics Production)
Week 4
First, I prepared an MDF bed, then used painters tap and supper glue to secure the copper plate on the surface. Now the reason why I’m using a V-Bit is because I did a lot of research and found out that technically speaking V-Bits are better suited for PCB carvings. Cylindrical endmills are structurally weaker, but they are easier to use. Most people just buy an inexpensive set of low diameter endmills and change it every time they snap. I don’t have that privilege so I have to take the safe and harder rout. A big challenge I notices was finding good documentation for PCB carving with V-Bits. This made everything more challenging because I’m a noob with CNCs and on top of that I needed to learn the software to design a proper manufacturing process.
After learning how to use the software I made a special board to test my initial ideas. I decided to first test the V-bit carving toolpath. It failed miserably. I couldn’t think of anything causing the problem but after watching a lot of videos I notices some parameters that could be causing the problem. I notice a parameter in Carveco that said “MAX width and depth”. This helps you know how much the v-bit will go in the material to try and carve the selected vectors. The separation between my vectors was too big for my v-bit to hand so it used a bigger stepdown to try to make the selected carve. After noticing this I spent a lot of time watching the little documentation I had. Everybody was use PCB dedicated software which means that the software was interpreting differently the v-bit.
I believe that there is a way to make universal the use of a V-bit in any CAM software. The solution I came upon was making the software believe I was actually using a low diameter endmill. The parameter that are really important to play around are the stepdown and the stepover. Stepdown is how deep the bit will go per carve and the stepover is how much the bit is going to slide after each pass. I made a small drawing in SolidWorks to make fast adjustments. After doing this the CNC worked flawlessly. Now the new challenge is increasing exponentially my Shapeoko Pro’s precision.
I decided to use the FTDI PDB to test some parameters. This board is really hard to engrave due to the small traces the integrated circuit need. If I can engrave this PCB then everything is possible with my CNC. After doing some tests I was noticing some problem with the stepdown of the V-Bit. After squaring everything I was still having problems (See PCB Precision Test Picture) … For some reason the engravings changed depending of the area of the copper plate I was working on. The finish is not that smooth too. Y decided to manually control the CNC with my Xbox controller and carve a straight line (right click and open the picture on a new tab to zoom in) to see if the problem was do to squaring.
The answer is "yes". I clearly saw a difference in thickness throughout the line (this happens because I’m using a V-bit) and one more thing. At first, I didn’t notice but a really big problem was cause by the MDF square I was using as a bed. I squared and clapped the boars only on 4 sides of the MDF. By doing this the center of the board was deflecting in the middle, creating a "frowny face". Machine squaring the board by cutting a rectangle on it helped but at the same time the "frowny-face" got worst. This happens because the material got thinner in the area, I did the clearance toolpath. Less material is restating the deflective force therefore it’s easier to bend.
Now I need to machine square the CNC’s MDF bed and design a smaller bed that clamps to the T tracks my CNC has. I had a really hard debate deciding what material I should use. I want something more permanent so MDF is out of the question. Plywood would make it hard because it is manufactured by putting layer on top of each other. This was complicating everything because in Honduras the material I have access too are limited. Then my friend Dan Meyer saved me without knowing. Dan sent me a couple of 2 colored HDPE boards. This was going to be the chosen material to use. To do this perfectly I need to make some counterbored holes on the HDPE.
To eliminate alignment errors, I decided to design and print a hole jig to make the initial holes on the HDPE boards with a drill. I clapped the jig on the HDPE and added a drill jig to make sure the holes are perpendicular to the surface. The outcome was perfect.
I manually drilled the counterbored holes. The first holes I dril1led with the jig gave enough clearance for the fasteners threads to go through the HDPE and they also help self-center the bigger drill bit. I marked the drill bit with painters tape to try to make the same stepdown. Then I did a test on the T tracks with a big smile. This upgrade should correct all the previous issues. Now the PCBs should come out pretty clean.
Week 4
Getting to know this kind of MCU is a little problem due to its lack of detailed documentation. Most people like keeping it safe by restraining their designs with older architectures (Atmel and Microchip). Due to the shortage of semiconductors the demand for this older MCU models had an exponential increase. This limited the fabrication of critical components for many electronic devices. Atmel and Microchip MCU’s still are hard to find. On the other hand, ESP-32 MCD are easy to find.
I’ll follow my ex-classmate and now instructor + friend from Spain Adrián Torres foot steps. I’m going to build the UPDI (USB-Serial FT230X). ). I’ve never used a hot air gun, soldering paste nor low temperature solder. I’ll learn to use them when they arrive to Honduras and use this technic with all my PCBs. I’m also planning to sue my raspberry pi 4 to set up my UPDI.
Dan and I have been working on an incredible side project that we hope will help everybody on one way or another (for more information please click “Remote FABrication Initiative”). I finally manage to receive Dan’s mail from Chicago! The package took longer than expected but at least it's in my possession. Every arrived perfectly with no visible damages. Dan sent me a lot of components so that I could make many boards throughout FAB Academy. I’ll be making the USB-FT230XS-UPDI and Serial for this week. Dan sent me 4 of each in case I mess up something because it will take too much time to get new components if something goes wrong. Hopefully everything will be ok and the 2 PCBs will come out flawlessly.
I don’t have a special magnifying glass with white led lights to assist me. I needed to make sure that I had enough natural light to have better visibility. I decided to setup my soldering area on my backyard and wait for a sunny day to arrive. I was extremely nervous because I only had 4 FT230X with me. Two tries for each PCB if I mess up. This was my first time playing around with SMD soldering hahahaha. I did a little experiment to see how the soldering paste works. I used TS391LT low temperature soldering paste to work with. This paste has a different behavior to what I was expecting when it’s exposed to heat. The paste gets a little dark and burns a little when it’s exposes to the hot air. Then it turns shinny and it settles almost by its own.
After doing this I needed a game plan to add soldering past just on the needed areas. The areas for the passive component (resistor and capacitor) were no problem. The problematic area was going to be the traces for the legs of the integrated circuit. After some minutes of critical thinking, I came up with a crazy idea that worked. I added a vertical line of soldering paste that covered all the traces. The I removed the extra soldering past with a needle. This separated the traces and prevented a possible short-circuit between the legs. I messed up my first UPDI PCB. After soldering its really important to check the traces with the continuity mode in a multimeter before connecting the PCB. I tried to see if my Raspberry Pi4 was able to detect the board but it didn’t. To do this you enter the Linux terminal and use the command dmesg -w . I checked again with the multimeter and found a short between to traces. I tested my new knowledge with the FTDI and came out flawlessly. Then a tried again with the UPDI and came out flawlessly.
Make (Almost) Anything
