This weekend was one I couldn't wait for. Zahrah had been talking about taking a trip to her hometown Fujairah since Fab Academy started. We were pretty rushed about it, since we were all a little behind on assignment work - a combination of delays because of a late start with the shopbot and everyones' designs not being quite ready.
Early in the week we had loosely planned for Zahrah to take us (Wendy, Francisco, Darshan and myself) for a weekend though it was only until 6.30pm on Thursday (the start of the Arabic weekend) that we actually made phonecalls about accommodation.
In a last minute frenzy we packed up all our crap and left the lab. Unfortunately Francisco couldn't join us. But I was happy that we managed to leave at all given how late we left it.
It took a few hours to get from Dubai to Fujairah due to construction, and Zahrah drove us past a major oil refinery. I wound down the windows and the smell of the oil in the dark was rich and strangely beautiful (to me at least) - just a few megalitre tanks of the liquid gold that the whole country floats on. Both figuratively and literally.
Because of the timing Zahrah's mother very generously allowed Darshan and I to stay in the guest room - basically a party room separate from the main house decorated with ornate furnishings and a chandelier. Late into the night Darshan and I talked about the future of manufacturing in space and other obscure technical things Fab people talk about.
In the morning we went to the beach and took a boat tour around some islands in the gulf. Zahrah and I rode around on jet ski's while Wendy and Darshan went for a swim. I hadn't yet been in water in the Emirates, so it reminded me of how casual it is for Australia to have beaches.
For lunch we met up with Zubair, who drove down from his hometown Ras Al Khaimah a few hours north. We had a picnic of biryani in a traditional large dish by the beach. The arabic way of eating uses a dish over 1 metre in diameter, and everybody sits around it and eats from the same plate. It was possibly one of the nicest vegetarian dining experiences I've ever had!
Our last stop in Fujairah was for Wendy to get henna done by some of Zahrah's friends. We were originally going to have a fake tattoo of circuit traces done on my forearm but black henna wasn't allowed at this place, and also men weren't supposed to be there either.
So instead Zubair took us to a mosque and we ate icecream sandwiches to kill time.
Unfortunately after this Zahrah had to leave since her mother was flying out the same weekend we decided to visit. So Zubair took us late in the afternoon to Ras Al Khaimah, the northernmost emirate at the pointy end of UAE and Oman.
The city had no tall buildings and was divided by a main river going through it, and the two halves of the place were connected by a few major bridges making it feel exactly like Perth. It even had an attempt at a small high-rise development on the less busy foreshore just like Perth's been trying for about 15 years. I'd be ready to say this place had a better food culture and family vibe however.
We booked a few rooms in the local hotel and crashed after an exhausting day.
In the morning we drove up to the mountains. It was hectic, because they were made from a jagged slate that cut through the sky almost 2km above sea level. For someone who was impressed by large expanses of Australian flatland this was really something else.
In typical Emirati style though, a ridiculously well built road was carved out of the mountain and an incredible zipline network was set up for tourist fun.
From the plateau we could see down into a village in Oman and a concrete factory.
As always, there is never enough time. Given the amount of stress occurring this week we were all pretty glad to be out of Dubai for a bit!
##What's this week all about then?
This week was about Moulding and Casting, with the specific requirement of using a three-step process in order to build the part.
This meant that we were required to machine a mould, cast it with silicon, then use that silicon mould to cast the real part from urethane resin.
Because of this requirement, techniques like making a mould from a printed part were excluded (admittedly not challenging enough) and it was an excellent chance to properly master 3D machining.
I was really at a loss for ideas of things to make for this assignment, but I settled to try and make propellor blades for the windmill. I initially wasn't going to commit time to the blades since they would be intricate to design and likely require a lot of strength.
They were easy enough to source from HobbyKing so if all else failed I would just buy them from there, but I'd at least give them a go seeing as I couldn't think of anything else.
Additionally, if they came out well I could help Abdulla out since his project also needed propellor blades and it would actually be kinda cool if we had the capablility to build blades in any custom sweep and twist we needed.
So in a nutshell; Things to make -
Initially I designed my part in Fusion to be printed on the Formlabs printer so I could get the sizing right and figure out how strong the part might be when made out of resin.
Being low RPM I had to give it quite a twist angle and being resin I also wanted to keep at least 6mm thickness at the root of the blade where it attaches to the hub.
I'm not much a fan of mucking around with lofts but it was inevitable with such a specific shape. I created four offset planes, and Abdulla sent me a profile he generated using a NACA profile generator.
NACA is a standard airfoil profile library and we wanted one with good wing stiffness and balanced lift:drag coefficient.
I then sized up the profiles on each of the offset planes by precision eyeball and lofted the profiles together.
The compounds used had some particular safety considerations to be respected - humans are not particularly well designed to accept these complex substances and so even small exposure would seriously compromise the human immune system.
Each of the different compounds were two part, we had three types of silicon: Dragon skin 20, Sorta clear 37, Mold-max 60; and one type of urethane resin: Smooth-Cast 327, with a whole bunch of different colours you could add.
Each of these has different mix ratios, and they can be either by weight or by volume. Pay attention to this!
+ Dragon skin 20 has a 1A:1B mix ratio by volume and weight, and has a pot-life of 25 minutes
+ Sorta-Clear 37 has a 1A:1B mix ratio by volume and weight, and has a pot life of 25 minutes
+ Mold-max 60 has a 100A:3B mix ratio by weight, and has a pot-life of around 40 minutes
+ The Smooth-Cast 327 has a 1A:1B mix ratio by volume, or 115A:100B mix ratio by weight. It also has a pot-life of 20 only minutes.
The pot-life as stated is pretty dubious, so I generally try to stay well below it. Also, large batches for some reason begin reacting sooner, so I stick to having small batches unless I'm really ready to put out a fire.
You should really take a look at the Material Safety Data Sheets before you begin using them (I've linked them in the paragraph above). You also have to be really careful about handling bottles that have been opened, since when you're working with gloves the outside of the bottles get contaminated in compound which the human skin's defensive mechanisms really don't enjoy.
As a minimum you need gloves, eyewear and enclosed shoes. You should be in a well ventilated environment and to avoid spreading highly dangerous compound all over yourself you should wear a lab coat. We also had plenty of paper towels handy.
We did a bunch of tests and practised pouring while trying to minimise bubbles. In the end I settled with the medium hardness silicon for no other reason than we had the most of that compound left of the three. I did some small tests with each of them, and I found it made little difference for my design since there wasn't much fine detail, but I did need a good enough hardness to maintain compression around the entire outside of the mould.
Some of the ways to reduce bubbles didn't make much sense to me (I never really got the hang of mixing without bubbles). Apparently you slice the mixture instead of stirring, but there was no way I could find to do this slowly enough without folding more bubbles into the mix.
Another method was to pour a skinny bead from a height and let the bubbles pop on the way down. With this one I could see that the bead formed spaghetti when it landed and intuitively I would imagine that just traps more bubbles in it. It does work to pop most of the bubbles that were in the cup however.
A few techniques I found to be effective involved the way you agitate the mixture in the mould: Pouring a priming layer, then smearing it around with a tongue depressor into all the places bubbles would aggregate. This pre-wets the surface so there are no bubbles near the surface of the mould, and is good because even if you dump the whole cup in your part won't have any surface bubbles (so long as you painted it well enough).
Another technique was to pour the mould half way full, then repetitively up-end one side of the mould then the other, allowing the liquid to travel from one end of the mould to the other. This agitates the bubbles to come to the top because the increase in liquid velocity. This at least lifts any bubbles near the mould away from it giving you a thick skin in the area where the urethane will go.
This is also an embodiment of my inability to not intervene when food is cooking.
I didn't include a lot of pictures because my hands were full and also the white moulds are pretty boring to look at.
I also needed to address the problem of strength, and being resin I was concerned about the part shattering if anything went wrong (especially given high RPM).
Laterwards when I actually cast the part I realised that urethane is quite ductile and not brittle at all, but to attempt to give the part some rigidity and at least prevent catastrophic brittle failure Wendy and I talked about laying up the part with a skin of linen.
To confirm whether or not this would be effective I prepared some composite sandwich tests. I used the linen squares in two tests of two ply, two 0-deg and two 45-deg orientations. I also prepared one single ply control sample so I had something to compare to
I then soaked the specimens in urethane resin and let set in sealed conditions. The results looked promising - the single ply respectably durable and then the two-ply specimens considerably stiffer. The effects of fibre orientation were unclear but I was pretty happy that two plies of any configuration would do.
## Content warning: Heavy Mental Computation
### Just read the pictures (left) if you prefer your brain to remain in the usual two pieces
Next I would have to design the mould. This would be a relatively simple exercise if my part was trivial from the Z-direction, however I realised a string of problems for many days with the CAD of this mould.
This was the single most difficult problem I've ever had in a modelling situation ever.
At first I'd figured it would be a reasonably simple part to design a mould for - no crazy Z-direction features, no deep or thin features, easy radii to machine, no over-/underhangs.
However this was not at all the case. To summarise my oversights - the sweep and twist of the wing shape meant that I didn't have any clean references to create flashing lines for the mould. I also had an issue where since the sweep of the wing is "plunging" the depth of the mould couldn't simply be flat, it was also a twisting and plunging spline that followed the wing profile in X-, Y-, and Z-directions.
It's **really** hard to explain, but it took me a few days to figure out something that could work. Looking back and knowing what I know noe there's a lot of other things I could have tried, none of them *quite* perfect however, and I'd spent enough time on it that I just moved on.
First, in order to obtain a clear reference for a parting line I had to manually tweak each profile sketch to include a reference point along the smooth side of the sketch. If you think about the profile of the airfoil, because it is round on the leading edge it's basically one face with only one corner.
I had to manually draw the second corner in, on the rounded edge, so the computer could define and split the part into two.
I modelled a mould using this parting line, but then the next problem was that the mould was *unmachinable* . I did a curvature analysis and there were some concave areas that an end-mill wouldn't reach.
The next thing I had to solve was making the parting lines extend horizontally from the part, in order to isolate two out of the three axes. (The simplest way I can put this is that with the surface flowing in all three directions it was impossible to machine, so I had to limit one of the axes so machining complexity was only in two of the three directions).
I actually couldn't find any good ways to do this with my current suite of solid modelling experience so I had to branch into some surface modelling to get this to work.
Using surfaces and by projecting the parting line horizontally, many hours of lofting and stitching meshes, I finally was able to make the mesh watertight and built a mould that was both machinable and held a chance of keeping a good flashing seal on the mould.
## Safe content zone - Machining the moulds
Next up came an actually very enjoyable part of the assignment - machining the mould. Having a satisfyingly challenging part to machine, I arranged the machining job so that all four moulds would only use one block of machinable wax.
Two things - even though the wax is reusable I still didn't want to waste two blocks of wax when I could get away with one, and the second pair of moulds were produced so that if I ever wanted to produce blades that spin in the other direction I would have the one mould that was capable of casting blades of both directions.
Arranging this is Fusion I had one set of moulds on the top side, and then flipping the block I had the other set of moulds separated by 4mm plunge tolerance in case I machined too far (likely).
This was a really cool experiment since flipping the block wasn't part of the directive for this week but something I really felt the need to learn about.
First, and after much deliberation shared with Wendy, I drilled datum holes on both sides of the wax block. I was pretty rough about it (an attitude inherited from my time notching and welding racecar chassis by eyeball) but I got it pretty close. Then I set up a fixture that would allow me to accurately flip the workpiece and maintain datum.
Using the SRM-20 is just the same for setting origin for circuit boards, so I just used V-Panel to maneouvre to the datum as per Electronics Production week.
Wendy had recommended I use a ruler she had just bought to space the workpiece from the edges of the bed - I had a better idea. Since she wasn't expecting the ruler back until I was done I taped the ruler into the machine on one edge of the block, along with a lone defenseless pencil sharpener on the other edge, that way the part always had an corner to reference from.
Next I had to figure out which direction the software would flip the part in, so I designed a quick double-sided test. I used the Roland SRP Player software to generate toolpaths, had to add the profiles for the 3mm straight shank end mill bit I was using and set up the job for high-3D machining and two-sided machining. The software takes care of the roughing and finishing paths, but I could add other things I needed as required.
Once I did the practice run, I also was able to determine the offset from the edge of the part. I was happy I had more than 1mm on each side, but realistically my recommendation to anyone trying this is to give the wax at least 3mm to avoid cracking when removing the silicon from the mould. It's not a huge problem since you can simply tape the edges or seal it with some laser offcut scrap, but for the sake of strong moulds the walls should be at least 3mm.
The settings I used for machining were defaults from SRP player, and the toolpaths generated were roughing and finishing for both sides (four in total).
After machining the top side and needing to flip the part, I realised I wouldn't be able to use double-sided tape to stick the second part down since it was machined away and not flat! To keep the workpiece in place I fashioned a perimetric fixture using another pencil sharpener and offcuts from the laser. I tried to use tape on the very fine edges of the block, but being wax it didn't take too well with such little surface area and so I had to rely on a very close-fitting fixture to keep the block in place.
After flipping the workpiece I also found my origin was 1mm over in the x-direction. This is likely to do with my shonky drilling and not that the jig wasn't correct. Since the x-direction wasn't super critical and I gave myself 6mm z-direction to play with (as opposed to 3mm in the y-direction) I was comfortable that it's not going to matter too much if I missed and moved the origin of the work coordinates in VPanel by (0,0,1)
Much to my surprise (yes, I did just machine a free-floating block unrestrained) the workpiece kept in place and the part machined perfectly!
I took the block out and cast the two sides with silicon to prepare for casting in urethane.
## Ohh shiiit
This whole time, I'd designed the mould to be machinable by only 3mm end mill. I wanted to minimise risk (and undue effort) of tool-changes by sticking to one size of tool. I had a feeling this would bite me in the butt and after casting the first silicon moulds it almost came true.
I looked closely at the machined wax, and could see that the blade profile looked a bit off. The deepest areas of the mould hadn't actually been machined, and the surface was flat in areas that were potentially tight for a 3mm end mill.
I suspected my worst fears, that the bit couldn't reach the bottom of the mould, and that the toolpath generator decided it best not to go there. This could possibly mean that only one of the blade directions would have a sealed mould, but could also potentially mean that neither moulds would succeed.
I really wanted to avoid this, but I was about ready to (crazy bit)
Turns out I was wrong about this (in a good way), and it brought to my attention an interesting dilemma where it's quite difficult to see what you haven't machined i.e. material you *haven't not removed*! Realising that the flat spots in the machining were all at the same height as I went in to generate a new tool path for the crazy bit, I found the cut depth for each toolpath was set to only 50% the depth of the part!
This was set by default by SRP Player (what a f*cker) because since I had set it to calculate two-sided machining it assumed I only wanted to remove material to half way before flipping.
## Improving the Mould
Some ways I could have improved the seal of the mould would have been by insetting one side by a few mm and offsetting the opposite mould. This would have allowed a wall for sealing the part as opposed to a seal made up of nearly horizontal surfaces. Having the mould in this male/female style would be much better, but I initially didn't try it because I didn't think the wax could be machined in this way.
Having more locating lugs would have been nice, and also if I had filled the moulds the whole way up with silicon the moulds for the urethane would have been more rigid. However I didn't want to waste the silicon on my part since it was so large.
## Yeah, well.
All things considered, I'm actually quite proud of this failure for the week. It forced me to think about some seriously complex stuff and I somehow kept it together.
If I were forced to make propellor blades in the future I now fully appreciate the ability to cast from a 3D print. I understand that a lot of my designs are challenging to machine and this whole week has been a worthwhile exercise in clearing up some of those issues.
As usual, the design files for this week can be found at Design Files. However, assemblies can't yet be archived in Fusion 360 so you'll have to go to the direct links for the flip test files and the blade mould files if you want to have a play.
##Things that I've given up on:
###Printing the case in my preferred colours/materials
I had to settle for PLA/PHA due to time. With my flyout day closing quickly I had to settle with the fact that having the entire case printed in Nylon and ABS was just not going to happen.
###Luggage rack design
I wasn't particularly impressed with how long it was taking to do shopbot week, and being severely behind in logistics for the move to Iceland had me bummed out. I don't think I'm going to end up using the luggage rack in it's current embodiment, I'll have to figure something else out and quickly!
###Relationships, just generally
Gah. It was a pretty harsh week in the romance department also. What a let down. To get through it I just powered straight into the machining for this week and I think the solution will to be not sitting still in the lab for too long.
## Wrap it up Carl!
So lastly I noticed quite a bit of traffic to this page over the last week, I'm really glad at least some people are following along with my journey! Here's a map of where you're all from. Thanks for following along, it means a lot.
It's really awesome to see how the Fab network is unfolding and I can't wait to finally meet some of you in Toulouse this summer :D
Replacing your mate's welding rods with sparklers