After spending quite a bit of time designing at testing my own 3d printer impeller, and noting that such designs are essentially impossible to buy online, I've put this page together in case anyone else wants to purchase one. It is principally aimed at people who want to construct their own dust/chip extractor and filter systems, for instance woodworkers using CNC machines, table saws or other workshop tools. I can print up to 320mm diameter maximum (about 12 1/2"), and am happy to do basic customisation for size, direction and fitting onto your motor. More significant customisation, I will have to charge for. My own impeller is D250mm, and for one around that size I would charge £60+postage from the UK (about $78+postage). Postage for this size to the US would cost about $21, or to the EU €8.2.
Modelled on high efficiency centrifugal impellers with gently backwards curved vanes. See images for the internals, and the complete design. There are 8 main vanes, and another 8 smaller aerofoil shaped vanes as shown. Standard D250mm design is around 95mm/3.74" tall and weighs around 785g / 1.73lb.
Please note, these are not highly engineered designs that have been through multiple stages of FEA analysis and testing! Although I am a mechanical engineer, I do not have access to industry level fluid flow software etc. I did research the topic of designing such impellers as best I could, but there is very little detail available on actual implementation. The conclusion I came to is that the flow is complex and not something that can be easily distilled down to simple design principles. Further, the variety that exists even with high efficiency impellers like these, tells me that there really is no simple industry standard. So naturally, I eyeballed it :). I made a few smaller models, varying parameters such as inlet/outlet area ratio, and moved forward with these as best I could. That said, the shape of such an impeller, with forward curved vanes etc, is always going to be better and more efficient than one with simple welded metal blades. And having used my own impeller for a few months now, I can say with confidence that it does the job quite nicely.
See pics below for underside. I have designed the impeller to have a removable insert, torx shaped (dimensions available upon request). Three reasons for this. First, it lets me print it with thicker walls and greater infill for the insert than the impeller, allowing the interface between impeller and motor shaft to be much stronger. Second, it makes it replaceable without having to reprint the entire impeller (in case of wanting to move it to another motor, or damage). Third, ease of assembly for me. In my case, as per below, I had to use two separate inserts. This was a result of the shaft of my old washing machine motor being rounded over and not entirely uniform, due to the difficulties I had removing the belt pulley. So I added the first inner insert, glued into place, span up the motor then cut it to round and to the correct dimension like on a lathe. Glue the two inserts together, and finally the impeller proper can be slotted on top. In my case, an M8 rod goes through the impeller and into the motor shaft. Happy to do a custom design on the insert for you, to fit your own motor - either 2 parts as I used, or just a single insert if your application if it is easier to fit for.
V shaped concentric cuts on the base are only there to reduce risk of the base warping during print. Will not affect function.
As mentioned, my own impeller is 250mm diameter and runs in free air at about 3500RPM, which is the max my old 500W washing machine motor can manage. My actual setup is multi stage, pulling from a CNC machine into a dust cyclone, leading to course filtration, then the impeller and washing machine motor, then fine filtration, on to a pair of parallel vacuum cleaner motors (high pressure, low flow, low electrical efficiency), and finally a real UK made HEPA filter (never buy Chinese, I became sensitised to beech wood dust because they just don't work). The vacuum cleaner motors were part of my first design, but I wasn't happy with the flow rate or power usage, so when I bought my new 3d printer I designed in this impeller. I run all three motors when in use, typically with the vacuum motors on low power and the washing machine motor at maximum 500W.
Caveats and disclaimers
- I am entirely comfortable running mine at 3500rpm, and any that I produce should have no problems going faster, but when testing it is ENTIRELY your responsibility to test and run them up in a safe manner. That means not standing around on the plane of rotation, so that if they do fail, you are not in the firing line. If there is a failure, I will re-reimburse in full for parts bought off of me - but any damage caused by the failure is your responsibility! It should go without saying that a large impeller spinning at thousands of RPM demands respect, but if you are here and reading this then I would hope you know that already....
- Given that they are printed in PLA plastic, if you are extracting anything abrasive then obviously these should be added after the filters...
- Do not expect a super pretty print! This is a huge part to lay down, and so print speeds have necessarily been sped up. Even then it takes 1 1/2 days to print. It is intended to be functional, not ornamental, and all the print settings have been modified to that end. That said, I do have a pretty good 3d printer now, so it wont be horrible either...
- If you have speed control on your motor and are able to give it a more gentle spin up, that would be helpful. If not, and the impeller is going to be subject to large and sudden acceleration, then it might be necessary to reinforce the inserts. Get in touch and we can discuss ways to solve this. One option is to print a plastic shell and fill it with carbon reinforced epoxy - I can do that for you. That said, the plastic by itself is pretty strong, and replacing the inserts can be done relatively easily at a later date if necessary, so not a problem to just give it a go.
- For highest flow is it best to have a shroud of some sort around the outlet. Snail shell design is the most common (translates rotation momentum and exit pressure into linear flow). It might be possible for me to design and print something to your specifications, especially if the impeller you want is smaller, but it might also be easier for you to create something specific to your own project!
- Each impeller will be printed to order, so please understand it will take a couple of days to get it in the post.
1.8mm minimum walls, typically 20% infill for the impeller and 80% for the inserts.
Material will be PLA, for this application I think it is actually the best option, being reasonably strong and impact resistant, not to mention quick and easy to work with. Of course, this does mean that it should not be used in an elevated temperature environment. I could probably print in PETG if required, even carbon reinforced PETG but I would have to charge! Of course, polycarbonate or nylon would be best but it is a pig to print without warping, and I would have to charge many, many time more for the final product.
Printed with a 0.6mm nozzle and around 0.2mm layer heights. This does mean the layer lines will be quite visible. I "could" do the impeller with a 0.3mm nozzle and give it smoother surfaces, but the final product would be slightly physically weaker and, of course, take much longer to print. The only downside of a 0.6mm nozzle is slightly rougher surfaces, I doubt this will significantly affect the air flow but if you wanted smoother there are a few ways to go about it. Manual sanding is of course one way, but it might also be possible to spray automotive primer along the air flow pathways. Even spin it up to a low speed and let rip with the spray can. Another option is the get some nitrile gloves and epoxy resin (thick, add powder if necessary) and manually smear it over all relevant surfaces. Polyester resin might also stick, as PLA is polyester based, but I have not tried it myself.
Printing is done on my heavily modified Wanhao D9. In fact, it was such a rubbish printer I've had to replace almost every functional component, and it now bears only passing resemblance to an original D9.
For other makers out there with their own machines, I have added an old version of this design onto thingiverse: