All cyclones need a good seal, or the air will just go around into the vac instead of swirling around.
1 Like
The walls of the bucket do flex inward. But when they flex, I think that the seal between the mdf lid and the bucket has small gaps. I caulked them together, but it did not stick well to the plastic bucket. I’ll have to get a better bucket/lid combination.
Barry is right. The point of the cyclone is that the dust gets thrown against the sides until friction slows it down and it falls out of the airflow. The airflow speed needs to stay fast to be effective against the smallest dust particles.
The three 90* bends on the PVC are not doing your airflow any favors either. I’d try just attaching the vacuum hose to the top of the cyclone in as straight a connection, or as gentle a curve as you can maintain, and see if you get better results.
And, just be sure we’ve covered the very basics, the tools are attached to the longer black hose in your photograph, and the vacuum source to the white plastic pipe, right?
1 Like
Yes. 
I just had another look at the design in a slicer. The top part has a kind of a downward spiral, but that spiral effectively only works for half a rotation. And the diameter is quite large, so I’d expect air speed to be relatively low in that section. If the dust particles are not pressed against the wall, or they don’t move fast enough, I think they’ll just get sucked up into the vertical pipe instead of dropping down into the bucket.
The whole thing took 2.5 days to print, so it’s not that easy to just try a slightly different design…
Some cyclones use a “neutral vane” to avoid turbulence where the airflow comes in and keep the speed up.
My suggestion for retrying the vacuum connection shouldn’t require any reprinted parts as long as your hose fits in the top of the cyclone. I hope some tuning gets you to a place where the time you’ve already invested in printing pays off.
1 Like
My $0.02… It is a chip separator, not a dust separator. The dust is still going to get through. This looks like it is still letting a lot of chips through though. There is a lot of “art” that we explain with science behind dust collection. Getting enough air flow, and enough pressure, and speed (which are somewhat independent properties) is needed for the design.
I don’t have a funnel/cyclone, I have a Thein baffle, which I talked about here:
Since you have a CNC, maybe it would be easier to test theories out on something like this instead? Hopefully it works OK if it’s a lot smaller.
I have that same HF dust collector and I turned mine into a 2-stage with a Dust Deputy (first with a homemade thein separator, which worked great) and it really works well. It is WELL WORTH the time and effort to make it into a 2-stage. That being said, the little Home Depot versions of cyclone separators that snap onto buckets would work really well for the CNC also. I have one of those hooked up to my chop saw at the moment.
I’m sorry that the cyclone didn’t work out for you! I whish I could tell you exactly why my works, and what is faulty with yours - but as jeff says, undertsanding the sience behind it is impossible for me.
What I have experienced though - is the following:
My cyclone setup does indeed gather most of the dust - both fine dust, saw dust and chips. I’m suprised by how fine the dust in the bucket actuall is. It glues itself to the sides and the underside of the lid.
When there was air leaking different places in my setup, the vacuum bag got much more dust and debris. Making a proper lid and enforcing the inner sides of the bucket made the operation behave in a good way, and the vacuum bag is now nearly empty, even after lots of use. I think the air leakage might cause the cyclone effect to weaken a lot.
One thing I wonder about your bucket: might it be too small? Or can the rectangular size impact the cyclone’s swirl?
I’ve used a 30 ltrs brewing bucket. I think it’s size is perfect. Small enough to fit in my tiny chicken coop room in the shed, but big enough to accomodate the cyclone swirling. Here’s from the norwegian shop: https://www.europris.no/p-gjaeringskar-30-l-125377-125377.html
1 Like
Thanks for your feedback!
I actually found out that a hardware shop nearby is selling a very similar bucket to the one you use, with a matching lid. I was planning to buy one tomorrow, and now I definitely will.
I expected the bucket size and shape to be irrelevant, because once the dust goes into it, it should just fall down and stay there. No idea whether that is true or not.
1 Like
Just happy to share!
It seems to me that the “outward” force inside the bucket is quite strong. The fine dust is plastered along the sides and under the lid. It seems like the force of the swirls is very strong and pushes everything outward and downward. Can it be possible that this swirling effect have to happen both inside the cyclone, and in the bucket? As already mentioned - understanding the mystical forces of a shop vac is not easy!
Idea from way out in left-field. Any chance the printed cyclone is too porous, therefore leaking/losing air speed? I 3D printed some whistles for my daughter’s volleyball team which worked terribly until I coated one with thinned down white glue. While the whistle looked okay, there was enough of a gap between lines of filament that air wasn’t going where it needed to for the whistle to work. Sealing it up made a world of difference.
1 Like
Might make a difference! If leakage impacts the cyclone effect, a porous print would certainly be worth sealing in.
I would say no. I have multiple printed parts for my dust collection system and none of them leak and I’m only using ABS.
I think the Cyclone effect is important in the bucket so a round bucket is better than 1 with corners like in the picture.
I have been searching a lot on DIY cyclones to go with my MPCNC but none of them has buckets with corners – all are round.
I’d assume that ABS is less porous than PLA?
One way to check if there’s any air coming through the walls is to use the “dry wood test”. If you wet one side with water and dishwasher soap and try to blow it through with your mouth. You should be able to see bubbles on the other side.
Success!! My bucket imploded too!! 
I went and got that 30L round bucket, attached the cyclone to the lid, and now it works perfectly. I spread out the same dust on the floor, vacuumed it up, and pretty much all of it ended up in the bucket. The filter was still very clean.
I used two buckets. Under normal operation it does not implode. But when I close the inlet with my hand, the vacuum gets too high and the inner bucket starts to crumple up. I’ll make some reinforcements soon.
I also sealed the cyclone where the two halves are screwed together, not sure if that made a lot of difference. The two parts already fit together very tightly.
I did not change the white pipes with the 90 degree bends. It might be even better without those, but I’m very happy with the performance now. (I don’t have enough hose to leave out the white pipes.)
2 Likes
For science, I also made a small test to see whether the round shape is important. I placed my rectangular bucket inside the round bucket. The top of the rectangular bucket is about 3cm below the lid. (Seems more in the picture below.)
I vacuumed up the dust again, and it worked fine! So it seems that my key issue was the bad seal of the lid.
1 Like
As long as your second bucket is the same as the inner bucket the inside one can’t collapse. The outer one will keep the inner one from deforming. I’ve been running mine like this for years. That’s also why the dust deputy kits come with two buckets. Try to only let your 5 gal bucket get half full. After that the vortex inside can suck up some dust into the vac. If you can swing it, get rid of the 90s, they’re slowing down your airflow, but will work for now.
Good news, Ill print a cyclone as well if I find some PLA…
All are being used up for facemasks (Covid-19) right now…
We have been printing like crazy for a few weeks…1 face shield takes just shy of 1 hour…
It takes around 750g of PLA and takes between 2 and 3 days of continuous printing. It’s supposed to be possible to print it completely without any support, but I had to use the “support on build plate only” option to make the first few layers of the top part print well (the amount of support material is very small so it hardly impacts printing time or amount of PLA needed). The bottom part prints fine without any support.