Monday saw the arrival of the latest set of printed prototype parts from Shapeways which means I've been able to have another go at building a chassis for the tiny Hudson-Hunslet diesel I've been working on. If you remember from last time while the parts all seemed to fit together I couldn't actually get it to work. Of course, since then I've bought the right tools for the job so should have no problems with clearances this time.
While I still think I could make the chassis work using the cheap 3D printed nylon material, a lot of the feedback I got to the first design was that I really should be using brass bearings. The problem is that because of the size of the chassis and my wish to have clean air between the wheels, there really isn't room to print a big enough hole to take a bearing. Of course that doesn't mean I can't have brass bearing surfaces as Shapeways offers brass as one of it's materials. Technically it's not 3D printed brass as they actually print a 3D wax and then use that for traditional brass casting. The end result though is that I can have the chassis made from brass.
On first impressions it seems to have come out really well and seems to be dimensionally accurate and a nice fit inside the 3D printed body. Of course I still need to ream out the holes and in the first photo you can see I've only done the right hand set, while the left hand set are as printed. In the second photo I've finished reaming out all the bearing holes using the 1.55mm reamer you can see hiding at the back.
Once all the holes had been reamed out I set to assembling it starting with the layshaft and worms. It took me three attempts to get the thing assembled as on the first two goes I managed to glue everything solid. I'm using Loctite 243 for this, which is designed as a thread locker but works well on none threaded parts as well. Unfortunately in this case it works too well. The problem is that I had to put the Loctite into the holes in the two worms and then push the layshaft through them. As you slide the shaft through it gets covered in Loctite which of course then gets onto the bearing surfaces and everything locks solid. On the third attempt I used a lot less and then oiled the bearings as best I could before applying power and after a little initial prodding this was the result.
I'm amazed how quite it is when run like this, but I'm also rather impressed that I got a stage further than last time and that the motor turns the layshaft without issue. I left it to run for about an hour to make sure the oil worked into all the bearings nicely and to make sure there was no remaining Loctite that could cause problems.
With the layshaft working I turned my attention to the wheels. Each axle also needs a gear adding as you fit it, and again these should probably be glued in place, but for now they are just a loose friction fit as I didn't want to waste them if it didn't work, and unlike the layshaft I'm not sure if I'll be able to get the requisite purchase to remove the wheels once fixed in place. Anyway even with just a friction fit the result is this:
Amazingly it all works nicely. This also shows just how small the chassis is as it is too small to sit properly on the rolling rod for testing. Getting this far has highlighted one problem though in that the worm nearest the pulley seems a little loose on the layshaft. I'm guessing this is because it has almost all the layshaft pushed through it so will lose more Loctite than the other worm during fitting and I used the smallest amount I could to avoid gluing everything solid again. In hindsight this was a bit of a design flaw so I'm going to tweak the design to make assembly easier, probably by dropping the central layshaft supports which will allow the worms to be fitted and then secured in place.
As you can imagine I'm seriously happy with how well it's gone so far. I need to figure out a set of pickups for the insulated wheels next and wire in a resistor so it can be used with a normal 12V DC supply (the motor is rated for 6V and I've been using a 5V supply for testing) but it certainly looks as if the approach will work even if it needs a little refinement.