节律性泵

Summary

This is a powerful pump with a set of fast, sharp, moving cogs. Use it at your own risk and I assume you recognise that I accept no liability for what you do with this design.

The ready-made STL files are suitable for a 15mm OD, 12mm ID PVC hose (search for "12mm ID x 15mm OD - PVC Clear Hose Vinyl Tubing" - don't go for less than 5m), and can pump 6+ litres / minute with a head over 4 metres - the motor torque (once the cogs are released and greased) for pumping is good - easily managed by a consumer power drill, but will obviously be affected by the viscosity of what you are pushing through the pump. The print bed will need to accept about 220x220 mm with those files. One can easily make smaller pumps using the scad file - I have not tested that fully.

Here is a more recent video showing the pump shifting 12 litres of water from one bucket to another in 120 seconds: https://player.vimeo.com/video/864764282?badge=0&autopause=0&player_id=0&app_id=58479

The drill was mainly operating at a medium speed, so it can pump faster, but gets increasingly noisy to do so.

Available variables include

• the stack count - one can chain (vertically) the pump. One (1) is all I have used.
• the cog shape profile,
• the number of outer ring teeth,
• the number of sun teeth, (the ratio between ring and sun teeth will affect torque)
• the number of planets
• the gear height (either side of the pipe channel)
• the pipe outer diameter
• the pipe inner diameter
• the power drive bolt size (as an ISO M rating)

Additional settings are less important - but affect the module size to be used for the planetary cogs (default 1), and the power cog (default 3), lid thickness, the amount of the circle to be used for the pipe, and so on.

The pump design allows for stacking - but be warned the torque requirements will obviously be multiplied for each stack. However, using the mod. 3 central cog will help transfer a lot of torque.

Also, the initial release of the cogs after the print will likely need a far greater torque than a power drill can supply - it's better to do this manually, and safely, clamping the pump by its edge and then using a suitable means to release it. It should be trivial enough to print out a release cog (the same as the power cog, but with a transverse hole for a suitable steel rod as a double-ended lever) to assist, but I didn't do that. I should have.

This project depends upon two awesome OpenScad libraries

• the iso bolt library at https://www.thingiverse.com/thing:2158656
• the polygear library at https://www.thingiverse.com/thing:3833810

My current OpenScad file is not written for customiser, but if you are familiar enough with OpenScad, you can find the settings on lines 250 - 261

There are three models to print: the main is the pump itself, which needs to be print-in-place, and the tolerances are way down less than 0.1mm, so you may need to (carefully) release the gear train - it's going to be about 120kg force to release (use WD40 or a PTFE release spray to help), and it's a tricky release. Make sure you have prevented bridging during your slice as well - you want a precision closing radius and resolution of 0.0001.
Using a strong infill (triangles or hexes of some sort) is strongly recommended: I used PLA+ for this print, and I printed it slowly (for accuracy and noise reasons); the main pump took about 24 hours (using a Bambulabs X1C).

For the driver/power cog, especially, you want thick walls - maybe 7 - and a good infill for dealing with distributing stress - such as triangles or hexes. The pump definitely benefits for that also - but the power cog really does need it.

Once the cogs are free, you should grease the entire gear train with lithium grease or such. After that, having run the gears without any pipe, the gears should run very well indeed.

Then there is the driver cog - this is fixed to a long M8 bolt, which you can then use to connect to a power drill, or to another motor.

There is the lid - you will want to print two of these. Exposed cogs are really NOT friendly to human flesh. I found out the hard way - you do not need to. Print two of these - they are friction fitted, and have a hole for the driver.

The model itself needs to be printed-in-place - you won't get the cogs into position otherwise.

Feeding in the pipe - Choose one of the pipe entry holes, and push the pipe in, then from the other side (keeping clear of the planets) pull through about a metre of pipe. Then, slowly rotating the planets (by hand), push the pipe in place as the planets close over the pipe. Finally push the pipe through the second hole. (The pump can be run both ways).

Make sure that the pipe is centred along the walls of the pump, and is not interfering with the cogs either side of them.

If you watch the video, you will see that the pipe will be tugged through the pump - so it is a good idea to clip the pipe on the input/pipe-entrance to prevent it being tugged through.

Update 1

The files haven't changed - but FYI, there is a weakness in the driver cog. The torque on the pump will strip the bolt-head side of the cog. It is easily remedied by using a large washer on both sides of the cog and tightening them well.

Update 2

Advice for use - once the pipe is threaded, run the pump one way only. If you switch directions, the pipe may fold in on itself, and then get caught in the gear train, and it will be damaged. Also, using pipe clips works well to prevent the pipe from travelling through the pump, but use a silicon ring or similar grommet to protect the pipe from the clip, otherwise it will scratch, and eventually tear the pipe. Don't put a clip on the output end - the pipe will naturally stretch and shrink, so it's a good idea just to keep the output/pipe-exit unclamped.

Final Thoughts.

This was a really interesting project. I built this with the purpose of pumping up plasticised (with PVA) plaster of Paris into a lath+plaster ceiling cavity in order to reinforce it. In the end it did not get used for that purpose, because it turned out trivial to lift the floorboards from above the cavity, and reinforce it by pouring down rather than pumping up.

Regardless, I learned a lot about OpenSCAD printing complex machine parts in the process. There may well be a time when I will use the pump anyway.