In the third part of this blog series we go in-depth into composite layup and mould making. This was such a big part of the project that we’ve had to split the blog post in two, with the first part going through our attempts with 3D printed moulds and the second half showing what we did after learning every possible way not to do it.
Once we had our initial concept enclosure and deck fully modeled in CAD, it was time to take our 3D model from the computer screen to a finished carbon fiber part for prototype testing.
For this you need a mold. The process of making a mold seems simple. You can either make a mold directly which is just a smooth surface in the shape of the outside of your part, and the composite material is laid directly into it, or you can make a plug/pattern/blank which looks like the final part, you then form a mold around the plug to create a negative pattern, and this then becomes your final mold/tool which you lay the composite into to create a clone of the initial plug. If you are confused imagine the conversations we were having at the beginning about plugs, patterns, tools, blanks, molds, negatives, positives... So to help understand here is a quick summary of terminology:
- Plug/pattern/blank/positive – a solid replica of the final part, which you use to make a mold/negative.
- Mold/tool/negative – a tool that you lay composite fabric into to form the final finished part. This is normally in the shape of the outer cosmetic surface of the part, and may include the entire shape of the part or require multiple pieces to allow for the finished part to release from the mold after it has cured.
Our plan for the boards construction at the time was an enclosure with a big internal overhang, which would have a flat 2D lid screwed on top of it to seal the enclosure off, and then there would be a thin single piece carbon fiber deck bolted on top, with custom shaped plastic or urethane blocks supporting underneath each bolt, because adding a curved deck with concave and drop to a flat enclosure would leave lots of gaps in-between. To make the deck we would directly make the mould/tool for the carbon fiber to be laid into to form the finished parts. However the enclosure was more complicated due to having an internal overhang which would require a two piece mould to release it. So we decided the enclosure would first have a replica plug made, and then we would use that plug to make two moulds/tools which bolt together for the layup and then unbolt to release the part. The reason we didn’t decide to directly make the two tools for the enclosure at the time was because we thought getting two separate hand finished tools to align to each other without a seam would be difficult, and a plug could be used to form both tools from the same hand finished part.
Our first attempt at the enclosure plug was a solid piece of MDF wood which is soft and easy to shape. The idea was to shape it by hand into the approximate shape of the CAD model, to fast track bashing out an initial prototype of our chassis so we could start testing the concept. We successfully shaped the overall bulk of it, however when it came down to the fine details our skills ran out, so we gave up on that approach.
Plan B was to 3D print the plug out of plastic (PLA) filament, and then sand the printing defects smooth. Due to the limited bed size of our FDM printer (Prusa i3 Mk2), we had to print the part in 4 sections which were glued together. After a few days and kg’s of filament we had 4 separate pieces which would form our first plug.
Once the four separate parts were printed we glued them together using standard Araldite epoxy. This gave us a plug that roughly resembled our design but there were a number of issues. Firstly the bed of the printer was not perfectly flat, and during the print the parts also shrunk at the corners. We also found that the pigment used to colour the filament had quite a large impact on the way the parts printed, leading to more discrepancies. This made the 4 separate pieces not line up naturally, so we had to use a straight edge during the gluing to align them which left some gaps between the prints as well as some misalignment that we would have to try and sand smooth.
However when we first started sanding it quickly became apparent that PLA 3D printed parts are surprisingly abrasion resistant and don’t sand easily. So we started looking at the options available to build up the surface instead of removing it. It was really important that we not just correct defects but made the whole surface extremely smooth and slippery with no 3D print lines that could mechanically lock into the tool. We ended up trying a high build spray-on primer which helped but not as much as we would have liked because we then ended up having to deal with 3 different hardness materials that we were sanding which is the 3D print (hard), bog (softer), and the high build primer (very soft). This coupled with the fact that the 3D prints were so uneven made it very difficult to achieve a consistent surface.
(photo shows a mishap where the plug was left partially sitting in the sun near a window on a hot day, causing the print to melt and forcing us to go through the steps all over again.)
After we did our best to make the surface smooth we had a sheet of polycarbonate plastic waterjet cut to create an accurate divider that would create a parting line along the edge of our plug where the first half of the two part mould/tool would be formed with a flange that the other half could bolt to. Unfortunately while our fancy waterjet cut piece of plastic was super accurate, the 3D printed plug we made was not and so some decent gaps had to be filled. We scratched our heads for some time on how to go about filling those gaps. Initially we tried plasticine but ended up discovering a wonder-material called filleting wax which is far superior. Its quite amazing how compliant and formable filleting wax is and this allowed us to easily fill the gaps between the polycarb sheet and our plug.
The last step was to apply a release agent to prevent the resin from bonding to our plug. We overlooked this step and were too eager to lay up some fiberglass, so we read on the internet that Vaseline can be an effective release agent and proceeded to apply a coat of it to the plug.
Finally, it was time… We dug into our box of composite goodies not remembering exactly what we ordered since it had been so long already since the start of the project, and began chopping things up in preparation for the layup that would form the first half of the mould/tool. This included what we thought was a fine gsm plain weave fiberglass (but was actually nylon peel ply fabric designed not to stick to resin), as well as release film, breather cloth and vac bag.
We mixed up some polyester resin (we used polyester and fiberglass for the tools to save cost), and with paint brushes we placed our first piece of “plain weave fiberglass” onto the plug and proceeded to wet it out with resin, then repeated with a few more layers. It quickly became apparent how difficult the triangular pockets were going to be to work with as the fabric just wanted to drape over them, causing voids at the corners. So at this point we decided to add the release film, breather cloth and vac bag and see what would happen after we pulled a vacuum and let it cure for 24hrs.
The result actually looked pretty cool (and believable – we still had no idea it was nylon peel ply we used), but it was obviously super flexible as we could tell by trying to lift one of the corners. So we got out the heavy 600gsm plain weave fiberglass and added some additional layers ontop before we tried to release the part. We placed foam ribs between the layers to bulk up the part to make it more rigid and we really had a hard time getting the fiberglass to conform around the foam due to its thickness and inability for plain weave to stretch. Here you can see how it looked after curing:
We had a successful release from the polycarbonate sheet, but unfortunately when we tried to remove the 3D printed plug not only had the resin completely adhered to it, but even our efforts to remove the plug by destroying it resulted in the nylon peel ply layers doing their job and separating from the surface.
A complete failure resulting in an unusable mould and total destruction of the plug.
We decided to have another crack at 3D printing the plug as this was going to be by far more cost effective than other methods. This time applying everything we had learned from the previous attempt.
Significant refinements were made to the 3D print settings and the same red PLA filament was used for all 4 pieces, this time a far better surface was achieved with much better tolerances that would save a heap of work in the steps ahead.
We got mad trying to sand all the tiny detail lines and pockets we designed into the parts so we just filled them with bog to get rid of that problem. We upgraded to a proper primer designed for this application called Duratech Surface Primer, which was much more effective at filling the 3D print lines and could be sanded to a really nice smooth finish with a lot less effort.
In the background can be seen our attempt at also 3D printing a single piece deck mould/tool directly.
Once we applied the surface primer to the plug and let it set, it was sanding time. If you’re not familiar with refining surfaces, the process is quite tedious and time consuming. 120 grit had already been used to rip through the bog underneath and achieve the overall shape, so we moved up to 240 grit on the softer primer to not burn through it too quickly or leave deep scratches. We then went to 400, the goal is to remove all the scratches from the last grit each time we go up a level. Next was 800 and lastly 1200 at which point the surface was feeling extremely smooth.
After that we decided to go one further step and rubbed a cutting compound into the surface which effectively works like even finer sandpaper. We then had to seal the surface (TR Sealer Glaze) because the primer is still porous meaning resin could soak into it. The sealer was applied with 2 coats and then buffed smooth and following this was several coats of TR Honey wax which is a proper mould release agent. Unfortunately we did not know that you have to buff the wax off before applying the next coat, so we ended up just smothering 6 coats of wax into each other.
We used the same technique as last time with a polycarbonate sheet and filleting wax to create the flange for the first half of the mould and during the layup we used real fiberglass this time! But unfortunately we only had 300gsm plain weave so it still didn’t want to conform to our part and we decided to add relief cuts to the fabric during the layup to help the plain weave conform without leaving cavities. We used a lot of layers of the heavy 600gsm ontop of that and then applied the release film, breather cloth and vac bag.
24hrs later anddd boom – we had a successful release! (sorry no photos were taken).
But the surface was shit.
There were so many pinholes and voids where air pockets got trapped or the glass bridged over a corner, despite pulling quite a strong vacuum on the part. The good news is that our plug was mostly intact and with some light touching up (ok we completely sanded and resurfaced it) it was ready to go again.
This time being many youtube videos wiser we had some new ideas on how to go about the layup. Firstly we ditched the stubborn plain weave and opted for chopped matt fiberglass instead which not only conforms far better but is significantly cheaper as well. Secondly we sourced a polyester gel coat resin, which is basically just thicker and what it allows you to do is paint on some layers of resin to the surface and let it partially cure before adding the fiberglass on top. What this does is ensures the surface of the part is free from air bubbles, reduces finer details, and provides a tacky surface that the first layer of fiberglass can stick to so it can be finessed around the finer details of the part and stay put.
We also learned how to use the TR release wax properly, which is to allow the wax to glaze over and then buff it back to a smooth surface with a rag between each coat, until 6 or so coats were applied.
After another layup attempt the resulting finished mould/tool was infinitely better although there were still some issues.
While the gel coat provided a perfect surface in some areas, other areas had these strange ripples, and we still had voids around the sharp corner where the flange meets the plug, with the very thin layer of gel coat just collapsing into the air pockets behind it. After much research we learned that we simply needed more layers of gel coat and more time for it to cure before starting the fiberglass layup on top of it (4 coats allowing each to tack off before applying the next works well), along with being more careful not to disturb the gel coat layer with too much movement during the layup.
Unfortunately though that was the least of our problems. When we removed the vac bag and the polycarb flange, the release agent was so good that the plug just popped out of the mould, and it wouldn’t go back in the same way. It was as if the mould had internal stresses that caused it to warp to a slightly different shape, so the plug didn’t sit right even with a lot of pressure on it, and it was basically not possible to get it to register back into the mould for us to lay up the second half of the two piece mould and get an accurate seam around the part.
It was at this point we realised we had just put too much faith into the accuracy of a multi-piece 3D printed part and this plug/buck method of making the tool.
And that was the last straw. It was all or nothing this time. We had a CNC milling machine sitting behind us laughing at our futile attempts with 3D prints, so we abandoned that method and went all out with attempt #4 which will be covered in the next part of Developing the Radium board.