I started experimenting with resin casting around 1992, and it took me a long time before I found a process that gave me consistently excellent castings. Inspiration came mostly from studying commercial resin parts; I never found a manual or article that describes the vacuum casting process in detail.
There's an interesting cultural aspect to resin casting. It seems that vacuum casting is the European way of resin casting. In the USA pressure casting is the preferred method. I'm describing vacuum casting here. |
I made a graphic of the steps of the process that I use. I keep improving the graphic; this is the eleventh version.
There's one step that I haven't figured out completely, and that is post-curing at elevated temperature. Generally speaking, post-curing improves material properties of thermoset plastics. But I haven't noticed that effect with post-cured polyurethane castings. More experiments are planned. |
This is my no-brand Chinese single-stage rotary vane vacuum pump (Wikipedia). As far as I know, a rotary vane vacuum pump is the minimum that you need to achieve the vacuum level required for vacuum resin casting; a membrane vacuum pump is likely not good enough. Currently, similar no-brand pumps are roughly 150 euros in the Netherlands.
My pump came with a vacuum gauge that's pretty useless for this purpose, since I'm most interested in the 0 to 2 mbar (abs) range. But gauges for this range are expensive, and I learned to listen to the pump to judge the vacuum level. Plus you can see the resin bubbling vigorously when the vacuum is deep enough. The black vertical cylinder is the oil mist filter, that my pump did not have originally. I tried several other solutions, but nothing would catch the fine oil mist generated by the pump. This filter does its job properly. I duct its exhaust into the ventilation system, since it could contain unhealthy components. |
Here's my 9 liter vacuum desiccator by Kartell. The bottom half is made of polypropylene, the top half of polycarbonate. The dome used to be fully transparent, but it's coated with tiny drops of resin, because the resin bubbling can be violent. Plus I also used it for an industrial project, further dirtying the inside. When I bought mine some 20 years ago, it was around 75 euros.
I'm using thick-walled 12x8 mm PVC tube, otherwise the tube will collapse under the vacuum. |
I've used three types of addition-type silicone rubber in the 10 to 15 Shore-A range, bought from Dutch suppliers. From left to right: PolyService 8510 (orange, 13 Shore-A), Silicones and More Pink 10, (pink, 10 Shore-A), Smooth-On Mold Star 15 Slow, (green, 15 Shore-A) bought from Form-X.
The difference in hardness (softness) between the three is noticable. For example a Pink 10 mould is really floppy, and should be supported lateraly during the cure of the castings. On the other hand, it will release parts with strong undercuts. PolyService 8510 was my favorite, with good all-round performance, but the price increased rather drastically. Pink 10 is rather easily damaged: sharp edges of the casting will cause scratches inside the mould. Mold Star 15 is more viscous, but the moulds came out fine. |
I've used Smooth-Cast 305 for nearly all my castings. The main reason for using 305 is the 7 minute potlife, that is just right for my process. Most polyurethane casting resins have a 3 minute potlife, and that's too quick for me. I also like the slower cure of 305 because it means most castings won't heat up during the cure, which leads to shrinkage. A drawback of a slow resin is a longer demoulding time, in my case it's around two hours.
I always add a tiny amount of special polyurethane coloring, to change the color from white to a light gray. Details are difficult to see on white parts, and gray solves that. |
Here's a typical project, a set of small U-2C inlets. The master parts are seen at the rear, on their casting blocks. The parts are positioned vertically because of the openings. At the front, the castings with and without the gates can be seen. |
The photo shows three parts to be copied, fitted with a casting block and a piece of 1 mm plastic card in between (gate). The soft silicone rubber accepts large undercuts, as you will see. |
Here's a wheel, prepared to be copied. It's a pretty radical mold when it comes to undercuts, but for me this the best way to cast a good resin wheel. The spoke holes have been closed with Kristal Klear, which makes a very thin film when dry. |
A view inside a typical mold, held open to show the interior. Note that I do not make cuts that reach the exterior of the mold, because they would form paths for air leakage during the vacuum process. |
The resin has been cast and cured. The foaming of the resin under vacuum left some flash on the top face of the molds. |
The result: five nice castings. Note the heavy undercuts on most of these parts. Releasing the parts can be a bit of a struggle, but the rubber takes the abuse well. |
The parts come out perfect, consistently, without air bubbles. The castings blocks show a bit of variation in their volume. They need to have some thickness (strength), because they are the part that you grab to pull the part out of the mold. |
Usually I place my filled molds in a row after filling with resin (absent here), to hold the rather floppy molds to their original shape, while the resin cures.
The mold second from the right shows clear signs of being at the end of its life. The silicone rubber slowly discolors to a whitish color. Another sign is that releasing the castings becomes more and more difficult; the release properties of the silicone rubber deteriorate. |
There's a third indicator of mold life. After twenty pours, my castings start showing a surface texture that the master definitely does not have. Here's an example, showing the 3rd casting (rear) and the 25th casting (front). The difference between the two castings can be seen clearly. The texture can be sanded off, but it's a lot of work and damages the details. I chose to use my moulds for a maximum of 20 castings. |
Here's the result of a full day of casting - quite satisfying! My slow resin requires ~2 hours of curing, which determines the casting interval. But all castings were 100% perfect, not a single air bubble to be seen. |
A few words on the subject of shrinkage. Shrinkage can occur both in the mould and in the casting:
regarding the mould: one of the reasons for choosing addition cure (platinum catalyst) silicone rubber over condensation cure (tin salt catalyst) silicone rubber is that the former has close to zero shrinkage. The latter shrinks during cure, and amazingly continues to shrink during its life. Therefore my choice is simple: I use addition cure silicone rubber, and the mould will not show shrinkage.
regarding the castings: polyurethane resin itself shrinks very little during the curing reaction. However, there's a big 'but': when the resin heats up during the cure, because the cure is exothermic, it expands, pushing out some resin from the mould. When it cools down, the natural thermal shrinkage occurs, making the part smaller than the mould. The resin's potlife mostly determines the shrinkage: a fast cure means it gets hot, and a slow cure means it will hardly heat up. Fast curing resins can get so hot that they form steam bubbles in the centers of the castings! I use a 7 minute potlife resin (SmoothCast 305), and I never felt a temperature increase in the small parts that I produce. Therefore the resin hardly shrinks. SmoothOn lists less than 0.1%.
I tested the above theory by measuring my largest casting, the ALE-2 chaff pod. The master is 87.5 mm long, and of the five castings that I measured, four were 87.7 mm, and one 87.9 mm. I blame the length increase on temperature differences of the days I cast the mould, and the days I made the castings. Silicone rubber has a large coefficient of thermal expansion: I found values ranging from 200 to 300E-6/°C. Compare that to polystyrene 70E-6/°C and aluminum 21-23E-6/°C. That means a 10 °C difference makes a 0.25% larger silicone rubber mould, and that equals the size increase from 87.5 to 87.7 mm of my ALE-2 pod.
After sixteen years of on and off use, I replaced the vacuum pump oil. It looked quite horrible, with more than half the oil looking like the 'mayonnaise' you get when a car engine has a blown head gasket, and the engine oil and engine coolant mix.
I can't say exactly how much time the pump ran, but 5 hours total is my first estimate, 100 runs of 3 minutes each. That is less than a single overnight vacuum cure of a composite part. I read a warning that polyurethane resin vapors contaminate vacuum pump oil. But I don't have a way of checking that. |
This close-up shows how the oil separated after 20 hours. The top part looked like regular oil to me, then a layer that looked like 'mayonaise'. At the bottom I see some white parts. Weird! |
Interestingly, four days later the division line had moved down, from 140 to 105 ml, suggesting that the oil-water mix was slowly separating. However the separation stopped roughly at this point. |