Composites

6 December 2017 | By Casey Evans

“It is possible to commit no errors and still lose. That is not a weakness...that is life.” -Captain Jean-Luc Picard

Assignment

Group Assignment: Produce test coupons for your composite fabrication process(es)

Individual Assignment: Read the safety data sheet (SDS) and technical data sheet (TDS) for the resins that you're using. Design and fabricate a 3D mold (~ft2) and produce a fiber composite part in it, with resin infusion and compaction.

Class Notes

Other People

Documentation takes ages (think of my French journal). “Strong advice I keep repeating: you’ll fail if you work and not document.”-Neil. Solder mask. A MOSFET is the drain source resistance because power is I2R and power is heat. Cast epoxy film to not melt away when soldering. Final project page should show a completed project plan at this point. Sam Chin is doing a programmable music box. Data on a tape the motor moves and plucks strings. Harpsichord mechanism could be a good starting point. They pluck. Pianos hit the strings. Liz Carre. Kinetic sculpture. Liz is the person I met at the welding recitation. “Time triage.” Piezo disks make a nice voltage when you flex them. Max Allen. Go through to make sure each page shows that you can do each of the skills (one skill per page at a minimum – ie evenly distribute). Charlie plexing has nothing to do with a grid, it’s about the number of LEDs. Same line can’t be on both anode and cathode. Off on or tristated okay. It can be a grid unwrapped into a straight line.

Link on charlieplexing. Then if you have trouble routing, just add a wire perhaps, or add a second layer. One layer boards are cheapest of all to make but it’s okay to get creative. NxN grid of LEDs, knock out one LED per row along the diagonal now you have Nx(N-1). “Double sided is not a heroic thing, it’s a very modest thing.”-Neil. Ground plane for RF or something. Vias are orthogonal to current carrying. Trace width relative to current. 4pcb.com/trace-width-calculator.html Max is doing a fidget spinner. Vibha. “You should be terrified at the thought: everything’s done I just need to document it.”-Neil. Oh dear. “Take advantage of your sections for the final push.”-Neil. That is talk to your peers and not just your TAs for help. Check. I do that. Tomas recommended electric wires. Microblower to turn off. Genshaft late drop. Oh. He was in our section.

Jake Read biggest site. Machine week. Neil: “Let me tell you what you did wrong. You did great on machine week.” Jake: “HTML5 wants MP4.” Neil: “HTML5 wants MP4.” Gifs aren’t good for websites. Avconv is freely available across platform. I think you just run it in command line, I never really understood that before. Gif is terrible media quality. Generally variable bit rate is better since it cuts more where less is happening. Different from ffmpeg because they fought and split. Composites this week. Next week is just meeting people and reviewing. Ensure final project documentation includes answers to all the questions on the project presentation page. Design how everything comes together for the project. Work from supply of time, not demand of time. Write a schedule of the remaining time. Use spiral development. Not how to almost make anything, how to make almost anything. Cut way back on your ambition to finish something. 10s of dollars of things in the inventory. Final presentation is a scheduled exam. Try to finish final plans this week.

Ended with Casey last week so Andrew Bahle. Not just images, also include words. Sonar theremin module. Bit rate error. RS232. 1/bit rate. Tiny 45 RC clock is good to 10% but RS232 needs to be good to 1%. 1) Resonator on the board to determine time 2) More expensive AVR with better clock resolution 3) Clock calibration register 4) Change bit rate in the code. Aluminum casting xylophone – nah – use a waterjet. Xylophone on glass Janice Wang. Harmonize as they approach each other. Increase tempo with proximity. Aluminum is at the limit of what a ShopBot can do. The only limitation in number of terms is ultimately inductance. Magnet wire is cheap so you can just sit there spinning. Nadia->Jens->Jake. Fusion 360 high speed machining. Trochoidal milling. Danielle Aspitz. Rui Qing. Darle Shinsato. Bluetooth RN4871. Underlay in terrible shape so board not held down well. Fairly precision machining so need a good underlay and flat copper boards (not domed). It’s important to understand what a multimeter actually is and that it’s not just a magic device. 5V FTDI cable or 3.3V cable – know the difference! Don’t fry boards. NRF24 are comically cheap. Tomas earring to do fancy stuff. TinyFPGA. Verilog is like C code for hardware. ABD3.

This Week

6.152J Nanofabrication. Vik talks about it on his page. Manufacturing recitation strongly recommended. Ara Knaian. Thesis is a good model of how to write a good thesis. Google’s phone named after him. He also worked on the McDonald’s straw, or “reverse suction for axial withdrawal.” By this Friday know everything that you’re doing for your final project – working with your section to figure this out. Composites. Bike frames, kayak (overshot), hardshell briefcase. Hard robust material with better material properties than the other stuff we’ve done. Can’t bend it because there’s a fiber in tension and a resin in compression. Hence composite. Aircraft (like Boeing 787) are epoxy resin with carbon fiber. Buildings are concrete and rebar. Lay down rebar first. Adobe – clay and straw. Wood fiber is “precisely a composite in exactly the sense.” Fibers + tree resin. Woah.

Boing 787 is like a single piece of tape. Fiber comes in many forms. We’ll use natural fibers. Cotton, silk, burlap and such. They’re friendlier and safer to work with than other things. Epoxy is what we’ll use (low volatility). Polyester not as good properties and smelly. Microcracks can still maintain integrity in a composite. Barcelona fab lab skate. Kids make their own skateboards. Stack plys of wood. A composite of composites basically. McMaster Carr stocks garolite is a poor man’s carbon. Boat building is full of composites. Entropy resins is a Berkeley spin off. Resins often need good ventilation. “Biobased” may just mean sawdust. A gallon and a half goes fairly far. Lingrove (collab with Entropy Resins) makes Ekoa sold at Rock West. In the past used burlap. It’s extremely cheap but the fill factor of resin is pretty large since weave is too coarse. There are also aerospace vendors, which we won’t use this week. Probably.

Mold+fabric+resin+pressure->remove excess resin->break from mold. Use layers of fabric so that there is a radius of curvature from the point of applied force to the top of the material. Middle fibers don’t do much so you can do something like aluminum honey core where you have a few layers, a core and then a few more layers. Spar, ribs and skins as a skeleton. I should ask dad to take some pictures of the airplane he’s been building for ages. Fibers should be roughly a fiber length apart after compression, 50% in some applications, 90% in aerospace. Resin needs to permeate the fabric. Remove excess resin to improve fiber:resin proportion. Squeegee resin on and press it on. That’s what I’ve done. Hydrostatic – fill a trashbag with water and dump it on top. Vacuum. Clamp (limited because they’re at a point). Fibers are doing the work-there’s a weight penalty for excess resin. Whatever you do won’t work well at first. Open mold has a vacuum bag over it to compress. One good side one bad side. Requires vacuum. Closed mold has 2 good sides and can be used with vacuum or other modalities. Machined shape of paddle in building foam. Wet layup. Pre-preg unpopular (fiber with resin in it). RTM/VARTM pull resin through fabric. Hydrostatic 1/10th compaction of vacuum but better than just squeegee. Release film (so it doesn’t stick to the tool – could be a grease layer on wood, wood paste wax, ‘cheaper and cheerful” alternative to official release film is 3-4 layers of stretch wrap), bleeder (seals the mold but lets the resin through – fairly expensive so we make our own with a wall covering perforator) and breather (distribute vacuum suction – quilt batting works). Clear coats (of resin) make things look pretty. Autoclave – heat + pressure. Pultrision – pull composite out, so trimming is important. West Kust Surf – composite surfboards made of cardboard. Exact Flat to get fabric to exactly fit the mold. Joe helps out with the Harvard shop. Browser based geometry engine. How to cut 2D shapes to go into a 3D mold. Bring in surfaces not bodies. Select faces you want to be included. Substania Negri. Adaptive remesher. 1 minute rule on run time. Pre-flatten. Couple degrees of curvature pelt is best. Optimize. Canvas is closest proxy to the fabric we’re using. Black is sagging in the mesh. Consider making darts (slits). White is good. Scaling issues with Illustrator.

Safety: Fiberglass and carbon fiber is dangerous. Asbestos lung disease if you breath it in (gets in there and doesn’t get out). In skin it’s itchy. Respirators, goggles, gloves. Taking carbon fiber and cutting it with scissors is okay it’s fine but sanding, milling and anything else that makes short fibers you get all the issues. Slippery – fibers can move relative to each other so that it drapes. This week shouldn’t be hazardous. Exothermic though and need ventilation for the resin. This week: Laptop case/cell phone case. Folded in cardboard, milled in foam. 2 sided mold recommended.

Training Summary

I think there was some but with the paper due in STPP (Science, Technology and Public Policy) and the PSet in 6.341 I didn't even really check my schedule to see when it was. I just wrote and solved and wrote and solved. But I did make time for the recitation for some reason so here's the notes from that:

Ara Knaian. Kindle. NK Labs. Transistion to manufacturing. Can’t make an iPhone in the woods. Marconi wireless. Palm Pilot. Autonomous submarines for the Navy. Bunny Wang hacked the XBOX. Now collaborating with Ed Snowden apparently. Eink. Core technology of Motorola Z. Suction Tube for Reverse Axial Withdrawal. Skunk works. Product requirements document. Know-how & IP. “A product is not a collection of features. A product solves an actual problem faced by actual people.” Desirements. Swiss-Army assault rifle not a good idea. What is it for? Why build it? How will your product fit in the market? Block diagrams and artistic sketches. Fab light laser – Jake Read.

• Product architecture – what are we building?
• Engineering Validation test phase. Can we build one unit, by any means necessary, that meets the product functional requirements? GSWT – Grad student with tweezers.
• Design validation test. Can we build a bunch of units that meet the product requirements, look great and have stable software? Now you can send the product to reviewers.
• Product validation test. Can we build a ton of units (10s of thousands) in the real factory with a high yield at a production cost compatible with the sales price?
• Ramp – can our factory and our supply chain handle the production volumes we need? Getting the kinks out of the supply chain.
• MP (mass production) – does the world need our product? If so, what kind of boat should we buy? (ie what other projects are out there?)

Design for manufacturing – tolerances, interchangeable parts, staying within datasheet spec everytime, design for high yield, design for CE, UL, FCC, USB, etc. It’s like running a giant Monte Carlo simulation (you find errors you didn’t know existed). Design your product to pass CE standards you can probably sell it almost anywhere. UL extension cords and stuff electronics standards. FCC is radiation interference standards. The product of engineering is documentation: bill of material, gerber files, solid models, mechanical drawings, process instructions, test instructions, software binaries, designs for production jigs and fixtures.

Supply chain – component vendors (digikey, mcmaster->bad prices, if you do something nontrivial then go around them) if you’re big enough they’ll also produce custom components for you, small component vendors (sometimes cheaper to outsource), job shops (they are an incredible equalizer, manufacture processing, ex printed circuit fabs, rows and rows of hundreds of people with exacto knives fixing iPhones that have been underetched, AT&S Factory, can charge more for smaller jobs, new entrance to the market less difficult since know how is outsourced), you can email correspond with a machine shop in China (cheap and fast and high quality at first but later ones may be less so) and get quick output, Contract manufactuers (final assembly and test of your product, become middlemen with job shops and component suppliers) Older teenagers like college students are the people working here. Interested in attracting a person or showing a funny video to a friend. ShenZhen can’t afford ShenZen anymore so they outsource. Lol. Electronic production factories decent working conditions. Like a college but with no learning happening. So sad.

Product worth manufacturing. The last 1% takes 99% of the time. The product of engineering is documentation. An OEM lives and dies by its supply chain. Job shops are an amazing enabler of small OEMs. Ara’s tip sheet of job shops he likes will be posted on the class site. Form Labs electronics, Kindle, MotoZ chips. When personal manufacturing meets mass manufacturing. Like seeing someone citing your paper. Seeing people with a Kindle always makes his day. A very gratifying thing. USB PCI express have high power overhead. Ara stuff is much more efficient. People like to help people who are their friends even if money is involved.

Assignments

Individual Assignments

During lecture I remembered working with my dad on his RV4 (which got put on hold when I was born but has been slowly piecing together since my brother and I left the house). We made a fiberglass cowling for the canopy. It's been years (since high school probably), but I was able to get a photo from my dad:

This is the cowling piece by itself:

We'd just used the shape of the canopy to set the fiberglass in epoxy. I think we used scissors to cut the fiberglass mesh. My dad said he used a respirator to sand it down but I wasn't there for that I don't think. I remember having to be careful with the epoxy and maybe ventilation. Point being, I'd done this before and so I had an idea of it. But I think we either let the air take care of it or put water on it. Nothing fancy like a vacuum. Probably. I mean, the whole airplane certainly wasn't going to fit in a vacuum bag. Anyways, I decided to make a stand for my lamp. When I went to the lab to do a second run of the group project (I'd missed the first one since I wasn't on the Facebook group), I ended up just doing my project. Grace was there and suggested I cut the mold in cardboard and use the laser settings to only cut through part of the cardboard so it would bend along the lines I made but stay together. It took a few trials to get the settings right but my tests are shown below. Numbers on each piece of cardboard are: top left is power on the cut-through, top right is speed on the cut through, bottom left is power on the cut-halfway, and bottom right is speed on the cut-halfway. The trial number is circled.

At first I thought I'd do a hexagon so I made two tubes as big as the laser cutter would go (32") to then tape together but I didn't like the look:

Grace suggested a triangle and I agreed. It works well since my apartment has three best friends in it. We're a triangle of awesomeness. So I remade the tube as a triangle (recut and everything by deleting every other line in my file, very simple). At first I just taped the corners to help get what I thought would be great sharp edges.

Then I taped the whole thing since at the time Grace and I were acknowledging that I wasn't going to be able to get the mold out. I wanted to get the mold out because otherwise why did I need the cardboard? So I taped it so that in theory the resin wouldn't seep in as much. Or if we didn't bother with the plastic wrap at least it would all have a uniform appearance. The idealism I had there about how sharp the corners could be. Haha. Anyways, here is the extra tape model:

Then I added fabric. The fabric I used was rolled up and had creases every 10 inches or so. At the time I didn't think creases would look good on my finished prodeuct so I was just going to wrap one unwrinkled length around, however much that was, but Grace suggested I add more layers. I cut out a second wrinkle free segment, which amounted to about 4 layers wrapped around. Here's what a single wrinkle free segment looked like.

Note that my strategy was to wrap the fabric around the pole until I ran out of fabric. I still kind of though that I would be able to pull the cardboard out. So I wrapped it in plastic. I wasn't as careful about the creases in the plastic because it was pretty difficult to wrap nicely (I've never been especially talented at wrapping gifts either). This should have hinted at what was to come but I just figured the vacuum (if I could manage it) would get rid of all creases at that level. Haha. A better idea, later described by Gavin, would have been to do like one side and corner, then another side and corner and then the last side and corner and then put them together. I could have also done smaller strips like paper maché perhaps. Paper mache is a composite I think. But I did what I did. With the fabric cut and pole wrapped, I was read to epoxy. There were some warnings on the back of the epoxy and hardener but I also looked at the datasheets. The Technical Data Sheet (TDS) gave the mix ratio and strengths and such. This TDS is a bit uglier but a little longer with more on safety and application. The Safety Data Sheet (SDS) gave the hazards, why they're hazards and how to deal with them.

I mixed the resin with a 100g to 43g ratio as per Grace and the TDS.

Epoxying up the whole thing was time consuming and rough. I wasn't wearing a mask for a bit and got lightheaded and asked Vibha for a mask. She helped me get it on since I had gloved hands. I may have lost a bit of mental ability from that. Anyways, as I described earlier I was trying to just wrap the fabric around and around. I applied the resin with the tounge depressor at first but then I realized how big the whole thing was and just poured it on and rubbed it in with my gloved hands. Getting it to stick and wrap and stretch in just the right way to line up well at the top and bottom and still be tight on corners was...really challenging. An art perhaps. I am not an artist. Like I said, a paper maché type model or Gavin's idea of several pieces to put together may have been better. Oh well, that's another adventure. I tried. I guess it looked okay. And here I still thought a vacuum (or more likely hydrostatic since my pole was quite large) might still correct all errors.

I wrapped it in a bleeding layer and then a blotting layer. We tried to vacuum it but it was big and the bags refused to seal properly. I had at first just tried one bag, but it was too big so I taped two bags together but it still didn't seem to have a good seal. I couldn't figure out where the air was leaking out either so I was going to try again when Vibha suggested sort of inverting the lip of one bag and joining the two together like that. Still didn't work so I just kinda rolled it up and put it under a table. I applied some pressure to try to get some excess resin out but I'd applied it pretty thinly so I wasn't terribly concerned. Also, I assumed hydrostatic wasn't allowed because we were on the 5th floor and if a bag popped (I would certainly have required several) it could be rather unfortunate. That may have been a decent option or else making my own vacuum pack with the plastic wrap. But for some reason that all seemed like more trouble than it was worth to me. I don't agree anymore, but alas, it is done. Here's a picture of the vacuuming attempts:

When I came back the next day I pulled out my pole and to my dismay it was quite hideous. The edges were very much not sharp and the resin had clearly been non-uniformly applied. Here's a close up of that:

So ugly. I was trying to decide if I needed to redo it and how I would do that. I also tried to pull out the cardboard mold, which didn't work. I'd kind of expected that but it was still disappointing. I did push it down enough to test the strength of the composite itself which was about 4-6 ply. It seemed tough. Definitely sufficient to hold up a lightbulb. I did work hard on these edges. They were my reference point and they kind of ended up still aligned. Kinda.

Lesson learned I guess. Don't try to make pretty composites, especially not without some better compression than your bare hands and ambient air pressure at sea level (if I'd done it back in Colorado would it have been noticeably worse?). Another point to make here is that the wet layup took so long that some of the resin was quite hard already by the time I was doing compression. There's another reason why smaller segments probably would have been a good idea. And you might not be able to tell from the picture below but the pole is actually a bit crooked.

I like to be proud of what I create and this was decidedly not something to be proud of. I imagined bringing it home to my roommates: "Look what I made this semester" and having to defend it's horribleness. But I'm also not a huge fan of redoing things. Especially things that take a while. And especially not at the end of the semester. So I tried to think how to make it work. My original plan of a cardboard flexure lamp shade was giving me the mental image of a wrapping paper tube with a light on top of it. Horrendous. Like the picture you bring home to your parents and say "look I made this for you." Oh no. I had to backtrack. So, after looking at it and deciding it looked like a stick plus Elysa's comments that wallpaper or paint might help (and my use in high school of a thin boa to decorate the wooden dowel that was a stand for my band headdress), I decided to go on a quest for fake flowers to cover it in. But first I had to go to 6.341. Then I went to an art store, a florist, another art store, Walgreens and Target before finally heading out to Porter Square's Michael's where I found a leaf garland. Their flower choices weren't that great and it was all more expensive than I'd expected so I just went with one garland which I knew wouldn't be a complete fix. I hot glued on a garland to draw attention away from it's ugliness and make it look like it was meant to be that way.

Wow. Much better right? I didn't really think so but the other people in the lab really liked it (they also liked it before the garland so I'm not sure if I trust them). I showed the pictures to my roommates too and they liked it. I'm guessing it's because they didn't see it in person. Anyways...now I'm going to have to get the other pieces to fit the theme. I really wanted to do flexures so hopefully that will work. I spent a while running back and forth with different materials the lab has to see how they look with the pole. I think acrylic will work well if I do decide to go with a vinyl cut circuit (probably for the microcontroller part if I end up redoing that, the Bluetooth board and bulb should be more stable. Or nah, the BLE sure but the bulb may benefit from being bendy. Then I could mess with the best angle for it and such. I'm also going to need to find a diffuser for the light since I'm not sure a flexure lampshade will be sufficient. But this is where I move on to the final progress page, so look there for more information. Here's the end result of this week's project:

Here goes for the final project! For curiosity I may try standing on my pole to test strength like Neil did in class. Okay I did that and it made some cracking sounds but held. Oh it's after the last class now and we got to see some other people's projects (I wasn't going to bring in my pole - because it's, like, a pole). I like the Christmas tree. I also like that loads of other people who talked were like: "nn, edges not so swagerous." I feel you. Anyways, here's the shot:

Okay. So I decided to make a foam stand, right? Well I also think I'll cover it in composite because it will fit the theme. I was going to make an interesting root structure in CAD but I had enough trouble with a simpler model - plus I had this great idea of just putting twigs between the fabric and the mold instead. So here's what I got:

Looks nice but the scaling was off and when I tried to fix it I overconstrained it or something. The lofts wouldn't generate at the proper scale, so I just made an estimate and machined that and I'll have to cut the triangle out a bit better afterwards. Not a huge deal since it's just foam. Dave helped me set up the Shark but I didn't realize you needed to babysit it and I had to run to work all the way out at Lincoln so it was paused for most of the day. When I restarted it I forgot to turn the end mill on and it just started dragging itself through the foam. Not pretty:

I guess that's what happens when you don't give the Shark enough attention. Fortunately it did a good job cleaning it up later:

But then I had to leave again to drop off my vaccinnation records at the clinic before they closed so I could get off registration hold and Dave turned everything off after I paused it and it sent it into some terrible state where I had to restart the whole thing (I may just be incompetent but the screen was frozen and the only working button was the EStop). It took 4 hours (instead of its estimated 64 minutes) but it finally finished. Then I left the lab to go to a party (also it was 2130 and Mike only said he'd be there till 22h anyways). The party was nice, but there wasn't any substantial food, which I had been banking on. Anyways, what's left is to cut out the triangle for the stand, set up a little corridor for the wires to go through, and then actually do the composite thing. Sick! So, as a refresher, the finished stand looked like this right out of the mill:

It wasn't just popping out like I expected but when I flipped it over I saw it was due to the great tape work I'd don earlier. Also the height was just about perfect. Here's a picture of the bottom right after milling:

Once I did pull it from the excess material I had some jagged edges. Here's a picture:

So then I sanded it down to a gorgeous finish:

After that I used my pole to map out where I needed to sand. I didn't take a picture before but it was about 5mm on each side that needed to come off. The stand itself isn't a perfect triangle so I did shape it a bit ungeometrically since I wanted a snug fit.

Then I sanded out a place for the wire to go. I also realized I needed to punch a hole through the thing in order to get the wire to be discrete. Not so bad. The lab had this really convenient and perfectly sized tool for the job (the wire path making, not the hole punching...that wasn't very pretty so I tried to sand it out a bit.

Here's what the first cable I tried looked like in the path. There's more on cable selection and problems on the final project planning page.

I also checked the pole fit in the stand. It was pretty tight and I had trouble getting it out. Since I'm going to cover it in composite I want it to be a bit loose, so I sanded a bit more after this. I will try to just leave fabricnear the edges so I can sand more if needed but I guess the composite might sand or grind if I really need. Here's the stand:

Then I finally got to use the vacuum! Vebha helped me again but it looks okay. I just hope the edges worked out. I made one dart but then realized that with a circle I would need more and I'd already covered it in epoxy so I just let it be and I guess we'll see how it comes out. Here's the vacuum bag:

And now I wait. The lab doesn't open until 3 tomorrow so I will be able to continue with this then. In the mean time I worked on getting the relay operational, which is all on the final project planning page. Hopefully the rest of this part will just be checking that it fits the pole and sanding it/fiddling a bit with it. We'll see! Woah, I didn't realize that I hadn't documented yesterday. No fun. It was my birthday so I guess I just went to sleep. I don't remember. My roomates had a surprise party waiting for me when I got home. It was awesome. Anyways, when I got back to the lab I unveiled my composite from the vacuum bag and saw it was covered in all these little holes from the bleeder layer. Very lame. Here's a photo:

It's okay though I just scratched off the plastic that got stuck and rubbed in the circles a bit. From a wider perspective I thought it looked pretty nice. Fresh out of the bag it looked like this:

Then I trimmed off the extra fabric, cut out the hole from the pre-holes I'd started before applying composite, and added a foam base. Fortunately I was able to shove the pole into the stand without too much effort. It looks pretty good:

So that's the end of the composite specific stuff. Everything else is continued in the final project progress page. Check it out!

Group Assignments

I wasn't able to make it on Sunday when a few of the EECS folks got together and made two test pieces. I used a different material than the two group tokens and all three materials are shown here.

Messing around with the pieces it's easy to see that the tighter weaves need less resin to fill the empty spaces. It was hard to compare weights since the objects were so different but the wide weave burlap seemed the lightest. Part of my contribution, I suppose, was using a different material than those used for the group project (at least of the pieces that I saw done) to compare. That was mostly on accident since I liked the material I used best but I'll take credit for it anyways.