Last modified: 2024-11-07 23:00:53
< 2024-11-05 2024-11-09 >Diaphragm piston air engine with rotary valve
Robert Murray-Smith posted this video: https://www.youtube.com/watch?v=pOPL0KdLXEc
Also this one about the diaphragm piston: https://www.youtube.com/watch?v=-gvplMQbqnA
I suggested that it would be cool if the valve was in the crankshaft instead of a separate part, and also that the 2 pipes going to the top of the piston should be reduced to 1 as soon as possible to avoid wasting efficiency on pressurising the pipe that is not connected to anything.
He nerd-sniped me by asking if I wanted to give designing it a go, so here we are.
The plan is to do as much as I can in one day and then stop.
I don't have any latex gloves and I don't have any dry graphite lubricant, but I do have nitrile gloves and 3-in-1, so that should do.
We'll use fasteners instead of glue.
The cylinder should bolt down to the base plate, which should have the pipe integrated inside.
The finger should be retained against the cylinder housing by a cylinder bottom-cap (with hole for con rod) that screws on and seats against a tapered surface, squeezing the finger in between.
We'll put some provision for bolts in the flywheel.
Actually, maybe turn the crankshaft through 90 degrees, so that it goes through the base plate like it does on the Wig Wag, instead of across the base plate, and then we can easily put the valve on the crankshaft.
Making the rotary valve seal will be tricky.
One idea is that if we make it have a smaller diameter then there is less sanding to do to clean it up.
We can run a reamer through the female part, and then if the other part can be turned on the lathe then it shouldn't be too much work.
One issue is that if the valve is on the crankshaft then any misalignment between the axis of the bearings and the axis of the valve (on either the housing or the crank) will cause it to either bind or leak.
We've got a 40mm stroke.
This is what I've got so far for cylinder and piston:
The piston has a flat surface so that it prints without support. It will use 1mm pins for the pivots.
The cylinder has a hole going through to the base plate at the very top, it has a tapered seating surface to hold the glove finger, and then a thread to retain the cap that holds the glove on, and a couple of mounting holes for bolting it down to the base plate.
I don't know if the tapered glove-seating feature is too large to actually stretch the glove over. Might be OK.
I considered putting a flat surface on the wall of the cylinder to match the one on the piston, to reduce the free space around the piston, but decided to ignore it for now.
The piston is 16mm diameter and the cylinder bore is 19mm, so a 1.5mm gap all around, should be plenty.
I consulted ChatGPT about con rod length, and a longer con rod is preferable because:
And the only downside is a longer rod makes for a longer engine.
I've gone for a 60mm con rod, but will see how crazy that looks with a 40mm stroke, maybe it is bad.
I think it'll be fine, if the stroke is 40mm then the maximum horizontal offset of the crank pin is 20mm,
so we have an angle of arctan(20/60) = 18 degrees. Maybe the con rod is too long if anything!
Having the crankshaft sit vertically means we can in principle block off the bottom of the base plate and fill the valve area with oil. Although wouldn't the oil then just get blown into the cylinder and out the exhaust?
The cylinder is printed, I managed to stretch the glove finger over it, it fits pretty well. It felt like there was not enough space around the finger inside the cylinder, but I blew into the hole and the piston got thrown across the room. Great success I think.
Actually I think the flywheel and rotary valve can all be made in one piece, at the cost of slightly more trouble if I want to turn the valve on the lathe. Which I think I'll plan not to have to do. And just need a small washer to space the crank pin bearing up away from the flywheel surface.
Here it is with con rod, flywheel, valve:
Just needs the base plate, and a nut for seating the glove on the cylinder.
FreeCAD 1.0rc2 keeps turning my transparent parts opaque again. https://github.com/FreeCAD/FreeCAD/issues/17728
Here's a first pass at the pipe on the inside of the base plate:
I would prefer smoother transitions between the "collector" part and the pipe, but maybe this is fine?
Another way to model it would be to do "subtractive pipe" for 2 short sections of pipe, ending up tangent to the main pipe. This will do for now.
One crazy idea is to put the crank at the top end of the cylinder, with the valve just going directly into the cylinder, and then a weird-shaped conrod to connect it to the bottom of the piston. I think save that for V2.
I'm now procrastinating about where the inlet/exhaust holes are going to come out to the outside.
Lol, I think I made a blunder on the timing. The crank pin wants to be 90 degrees away from where it is.
I'm going to print this base plate:
Will probably need to redo the flywheel due to the timing blunder. Might be able to test it by manually drilling a new mounting hole though.
I used a short piece of 0.7mm lockwire to make the pin at the small-end. And for a first pass I'm going to use a cable tie instead of the retaining nut as I still haven't printed that. Just got 20 minutes left to wait for the base plate to print.
With retaining nut:
I am pretty confident about the valve timing thing, so I'll just fix the model now, and then when the base plate is done printing I can test it with what I've got and kick off the print for the retaining nut & fixed timing.
It was almost as easy as just changing the datum plane, but that then means the hole intersects the seam and it throws off some of the following geometry, oh well.
OK, small error. I put it all together and tried to whizz it around and the piston fell out of the cylinder at the bottom of the stroke. Cylinder needs to be longer. The retaining nut might solve this all on its own, but I'll make the cylinder longer in CAD just in case.
I'm tempted to reduce the diameter of the bore slightly as well, would give it less wiggle room near the bottom of the stroke. Best just leave it I think.
I have 30 mins left on the flywheel & retaining nut print, but then it will be time to cook tea. I think just set the new cylinder printing straight away, and test it all out later.
Whoops, I forgot to move the air hole, what a blunder.
The new longer cylinder has the same problem!
The retaining nut seems to give a very good seal.
I'll try a larger diameter piston, and a squarer front on it. The reason to change the piston is I'm running out of time today and it prints faster than the cylinder.
It still doesn't work. The piston still gets cocked, and now there's more friction as well. So to avoid reprinting basically the entire thing, I'm going to reduce the stroke. I just need a longer con rod and a flywheel with the hole in a different place, and if then the friction looks like a problem I'll need a new piston as well.
Still might be able to get it working today. Going for 28mm stroke now.
It works!! Amazing. Now to put together a video clip demo and blog post...
https://www.youtube.com/watch?v=kuQJRl1gI3k
< 2024-11-05 2024-11-09 >