As a g700 owner, I have seen differences under normal machine conditions for dust collection vs sucking up dust out of a bucket (or in my case, floor sweepings from machines that aren't hooked into my system), so you're probably seeing more passthrough than you would under normal tool use. From a physics standpoint, under-sizing the system for cost savings (using 6" instead of 8" tubes) dramatically reduces your cross sectional area, which reduces the volume the system can handle by a square (only 9pi vs 16pi CSA per tube - or 18pi vs 32pi per system) means that your system can only actually handle roughly 56% of the flow of the g700. It also reduces the distance from the cylinder wall to the outlet intake tube by a fair margin, causing some of the particulate to get caught up in the escaping airstream. Those 2 reasons are surely large contributing factors as to why you are seeing more particulate making it through the whole system than what is seen in the commercial units.

I don't know if you tried it already, but aerodynamics plays a huge role in this whole system. And studies have actually shown that that 3D printed propellers, turbines, impellers, etc... can have as much as a 50% reduction in efficiency, and an increase in really turbulent airflow due to the layer lines. I don't really know all of the mechanics and intricate details of how one of these systems work, but I bet that if you sanded the impeller, the cones, and everything that's has to do with airflow, down to a smooth finish it will greatly increase the efficiency and the end result. I enjoyed the last video quite a bit and also enjoyed this one. Great job!

I think you should continue to iterate and test and document for the audience. Great job.

When cyclones are designed the relationship between the cone lenght and exhaust pipe lenght going into the cone is very important. As I recal it is called the return length in the cyclone. The lenght basically determines where the clean air is picked out of the stream If you have it too long or too short it will feed from the dirty air, so it is very important. You could also play with the exhaust pipe diameter to see if a smaller and thus closer to the center feed will help picking out the clean part of the stream.

This is awesome - thanks for posting this follow-up video!

This was awesome. Thank you for doing it, I for one found it very helpful and interesting.

Awesome videos and thank you for tackling this project. As someone who is looking at dust collection solutions as well as a 3d printer, I find this very interesting and something I may pursue.

two amazing videos. love the way you structured your experiment and fondings and your making abilities are amazing im so glad to have found your channel

Great video man! Keep them coming!

Great video. Since you already have all at hand, I would love to see the approach with the separated bins. The results will be much better.

Amazing to see an update!!

For series with shared bins you need to add rubber flaps so that the air can't flow through the bins. If done with limited pressure it fall through as they collect. For the larger particles you need a larger low pressure / flow zone so they can fall out

This is a great series. I would love to see you carry it on and get it perfected. I don't have anything further to add than the comments below on changes. Great Job! Fun Video!

Was TOTALLY worth it and interesting!

Fantastic project and great videos. Well presented, good flow (see the pun there!) And entertaining. Thanks

One thing when you're doing those tests, don't just dump those large handfuls in. Make it emulate more closely how a machine would produce the dust.

I cannot stress the value of this stuff enough: investigate "heat shrink sealant tape". They use this stuff for sealing the shrink wrap around BOATS. The brand I use is HIPPWRAP. Stuffs as sticky as duct tape, shrinks to fit like heat shrink, peels off clean, even after heating, and is air and water tight. I use it on all my dust collection joints. Stuffs marvelous. eBay.

I found this pretty cool. i modeled this up and ran it on my CFD sim and i found the tube diameter is too small. based on the particle size and centrifugal forces it needs to be around 7.8 to 8.6in to get proper separation before getting pulled back into the exit port . based on sims 5.4m/s is about the sweet spot for this build also shrinking the exit by 3/4-1in will help as well. adding a single 3in long 1/2 tall bar inside the tube to help pull the spinning material into the exit will help with larger chunks but the main problem is the material is too close too the air exit a larger diameter will help separate the air better with a peel at 22m/s that's a 9.8-10.4in diameter. the higher the flow the more separation is needed in sim 98.5% collection down to 40 micron is doable in just the 2 stages but honestly the first stage design was a bit lack luster only really helping with very large particles. if you want i can refine the designs you have a bit send a few changes

Great videos. I know it is a very difficult quest, but two of the big variables: the length of the separation tube and the diameter of the tube.

I remmember from other builds Your problem is probably the joined buckets . The airpressure can such back out stuff from the bucket when there are two openings

That water trap idea works very well. I have built a diy dust collector with a water trap and i can say it is very good, you only have to take care of the small water drops that escape into the system using a sponge trap or a small seperate water trap. There is hardly any dust left in the exhaust after the water trap stage

A little ridge on the trailing edge of the dumps into the collectors to catch the dust flying around on the inner diameter would catch those light pieces. Like a scoop.

Great video, I loved it thank you!

I would be interested to see if there would be any difference placing this setup on the suction side of the blower so it's all running at negative pressure rather than the positive pressure you're running it at now. Maybe it won't matter, but many cyclones run on the suction side, so maybe it does matter.

The research and effort you have put into this project is commendable and applaudable. higher air speed will keep more particulate in the airstream. Rapid drop in pressure or airflow will allow the suspended particulate to fall out of the air stream and be deposited into the collection bins. The key is to measure the air speed as they exit the bins before the canister filter. A larger bin or barrel will have a much more significant air speed drop reducing the generated air stream pressure allowing the particulate to fall out of the air stream. Volume and velocity are relevant at the point of air speed drop which in your case is the collection bin. Possibly a simple fix would be to baffle the area between the collection and the filter. 🤔 Or possibly have 1 large collection bin with a greater volume than the smaller food safe containers. On my system I use (2) 30 gallon trash cans that have a diy cyclone cap. I split the single main supply line to the containers into 2 lines significantly dropping the air speed and related volume and I have gone 9 months to a year of daily use before needing to address my filter.

Great iteration from @underdunn and your creation

I know my system pulls more air out of my 4 in port if i open another part way. This is so cool.great job.

Nice follow-up video

Great videos! Wondering how pressure relief valves would work on your dust bucket lids? You can make them with a bolt, some big fender washers and a spring - print a gasket for the washers with TPU or just use whatever gasket material you have sitting around (foam, etc.). Would make it easy to adjust pressure by tightening or loosening the nut or adding stiffer/looser spring.

next update video is gonna be putting them in series, bust all exits have their own bin, so the path of least resistance does not interfere.

I'd like to see it using something like a Record Power CamVac vs a traditional dust collector. The overall CFM would be less but the static pressure would be increased dramatically. I feel this would be closer to the original Harvey.

When my son researched his collector build he found it necessary to have a grounded copper conductor running through it to prevent electrostatic sparks from causing an EXPLOSION!!! Make sure you protect yourself!!!

I have been tossing around a few ideas on this for a few months as well, scaled down for a shop vac I have been working on a version with a tangential inlet opposite of the first dust port to force the primary cyclone into the first dust outlet. I have also been looking at some of the bagless upright vacuums for inspiration just trying to scale it up and make it less compact. SimScale CFD is free for a few CFD test models BTW.

I'd love to see one more impeller test where you go more aggressive on the pitch than the commercial. I also agree that series (vs parallel) would help but would make the canister situation tricky. I'm curious how much is getting stuck in the system. Perhaps you could exhaust to a bag filter and weigh the bag before & after to see what it collected vs what's in the bins. Did you ever chamfer the edges of the bin tube or consider making some sort of catch-lip that would knock things down better?

Wow, thank you for this and the previous video. Ever since I saw Under Dunn's video I've been wanting to create one of these. I've even gone so far that I've made a 3d model, but never gotten around to print it. Maybe this is the motivation that I need in order to finally do it? Some ideas though, consider using print in place threads, I've used in in loads of models and removes the need for glue and other fasteners, just print ant it is done. Also, as you sort of mentioned, offset the dust-hole for the second phase so that it spirals down into the bucket. No need to have it as wide as the pipe, just offset it to one of the sides, at least that is my plan if I ever get around to create one of these.

Great set of videos, and I’m gonna do some testing on an in-line design versus the traditional cyclone and see if I see any results that are worth mentioning. That said, I wouldn’t necessarily measure efficiency, in a matter which you have use simply because we know that some material is getting left behind within both the hoses and the extractor. In measuring performance, I think you would have to test the outflow side and capture those particles. Great job and I look forward to more of your content

The greater turn angle was definitely an improvement. I think what is going on there is by getting it to the outside sooner, it has less chance to go straight out the back. I still think there is gain to be had by playing with the outlets a bit more. I am also wondering what would happen if the rear exit port tube was longer. The velocity is going to be greatest straight off the fins and the further it goes back the more opportunity it has to come back to the middle. If the rear exit tube was closer to the fins, I think it would pick up less fine dirt.

I think definitely test out 8 inch tubing and also improve your weighing methodology. As someone else mentioned only weighing the filter canister makes more sense than pouring out the drums as you have particles stuck to all those surfaces and you can even see big plumes of fine dust when you pour into the smaller bucket. I think one more round of iteration and testing and I would definitely buy your plans for this

when watching the flow through if you made the exhaust tubes a bit longer so they went over the end of the cones a little bit it should stop the particles flowing back out instead of dropping down into the bin you could make it a sliding fit so you could change its lenght also the sides of the outlet for the fines sounld be as large as the clear tube

Hi, I have two different systems with cyclone separator, for your system the motor with speed controll will help you to find the best efficiency.

Great content and awesome build.. yes tweaking or even twerking may improve the process.. but that is what is great about product development. I am with the 8" design.. if implemented,. would buy your plans for sure!

Quick thought - if you run a cheap'er sacrificial filter in front of the HEPA you might significantly extend the life of the HEPA and only have to pay for a couple of sacrificial pre-filters a year. Thank you for these videos!

Hmm, what about flaring out the back part of the tube to get the particles further away from the exhaust pipe? And as you said in video 1, get the ports on the underside bigger so that parts can just fall straight down.

I wonder if having an aviation/race-car style NACA duct opening into the canisters would improve performance.

Hi mate, just another quick idea, could you maybe put some small filters between the seperator and collection buckets to dissipate some of the pressure coming back out of the collection buckets?

He did the ideas! You get a like, you get a sub

Paralel separators effectively doubles the surface area of the path, series keeps the area the same but has more resistance. I think series with individual collection bins would be better than parallel as the air is getting filtered twice instead of spilt in half and filtered once

just for the future, they make super glue/CA glue desolver..think most companies call it a "debonder" good to have around incase you glue fingers together (don't ask how I know LOL)

I think this is the same as a cyclone. But I could see some improvements. First part at best simulates a 90° bend. That catches the bigger particles, but than there is nothing to guide it down to the bucket. Same could be achieved with a T pipe fitting and a smaller pipe crossing the T. The smaller pipe could be conical and it would accelerate particles more at it's end. At the end air will need to bend 180° and all bigger particles will be thrown to the bucket. I imagine this T on it's side, the particle exhaust going straight down to the bucket taking advantage of gravity as well. Second stage looks great. Gets the air spinning but so will a cyclone. What I'm thinking here is to get particles out faster by cutting a longitudinal slit. Something like a vegetable slicer. That would allow dust to exit the spinning cycle and just fall to the fine dust bucket. If the static presure over this slit is not uniform, a blow back may form, pulling dust from the slit in lower section. Now I don't have hands on experience with neither. But the physics are somewhat the same, and cyclones are very old news and proven by time. And yes this look better, more futuristic. I would definitely like to see one modified to be mounted on a wall.

I think everything you have is correct and functional but the collection ports and catch being too small and not vertical walled to the deck. The slow air coming in off the pipe is siphoning without those vertical walls breaking it up. The venturi of the original, 6" in to 5" off the blower then split to 2-5" to 7", might be better served with the downsized centrifuge going 5" in to 4" off the blower then split to 2-4" to the 6" centrifuge. I'm just building my dust collection system, so this will be a fun project. Again, great work!

Metered orifice air bleeds from the collection bins to the final filter or potentially a third collection bin, to minimize reversion of the bins?

Just wondering if there's a measurable weight increase in the final canister filter with each test? If so, comparing the weight of material from the separators with the increased weight of the final filter might better represent the percentage efficacy of the separation system. As the result will not be affected by dust caught up in the ducting which never makes it to the separation system. Just a thought.

If you put them in series and you have to separate bins, it would work. But you have to air can go through the bin and it’s obviously not going to . That’s it being better than parallel is still relatively unlikely, due to the higher airvelocity. However, at the same velocity of having them in parallel , aka half. You’d probably capture 90% on the first one and then say 50% on the second one which might be slightly better

Just an FYI, super glue tends to melt PLA, which makes it great for bonding, but not great for teardown.

If you look at the diagrams you'll notice the exit port for the fine section is fully offset to one side. IMO you need to refo that part of the base to get full performance out of this. The coarse section should probably be full width ss you said in the last video. My 2c.

So they are horizontal at the moment, what would vertical ones do, think crude oil fractional distillation

What about modifying the base units to catch the particles on a tangential path heading down to the canister? Like a P shape, rather that a T at the discharge opening

Great experiment and thank you for providing the files. Did you ever look at extending (inside) the exit tube so that it protrudes deeper into the main clear tube. Wondering if this would increase (or decrease) dust collection of the fine particles

Maybe if you hybridised the system on the in-series setup you could get better results. Instead of having the air cycle in and out through the same inlet, you could an inlet into the drum which would enter at an angle and then have a vertical outlet for the air to move on to the next step of the system. But your system would probably need to be redesigned so that the first step would filter large particles and the second step would get the finer ones.

maybe extending the outlet port a bit further into the separator would help. Dust would need to make a S turn to make it to the filter.

Would it not be better to have you motor after the separator than before it were it would pull it through instead blow it through?

I do not know if you noticed in your video but the dust before the cone was collecting after the first opening. So, I would either advance the unit or lengthen the cone so it disrupts the flow before the first opening. Last if you had a larger hose coming in then going out it will increase the air flow. Air flow is always constricted when it goes from large to smaller size which is why there are dealers who suggest large tubing for maximize air flow.

Great 3 part series, have you ever thought about selling a kit to build the centrifugal separator? A whole lot cheaper than the Harvey.

Love the video... 2 Questions. Since the percentage was less then the first test would you say that the original 3d parts do not need to be changed to the more aggressive designed part and what is the difference between using your cyclone or just adding a filter to your Harbor Freight setup instead of a bag?

you can use oring , more part but easier to make prototype for testing, put vent on your canister, put your systeme vertical, taper to reduce speed of raw

A larger diameter tube might help. E=MV squared The more velocity you impart to the dust the more energy it has so the easier it separates from the air .

I’d love to see what would happen if you added a 2in line from the buckets connecting to the 4in line between the separator and the filter. That would remove any back pressure from the buckets, I imagine you don’t want clean air coming back through the hole that the dusty air is meant to get thrown into

You may want to make the fine dust drop off section larger (Full width of the tube). Looking at the Harvey product you can see the tube becomes a flat area where dust hitting that would loose momentum and drop down vs cycling around and giving it another change to escape.

Just a curiosity only one extraction pipe would work? why do we need two extractions , how would that apply in terms of efficiency ?

a single long three stage, with three separate bins, would be as good for capacity as two two stage separators. maybe single stage after a small cyclone would do the job, some tools are gonna spew out large chips that i'd worry would clog this thing.

When hunting for more efficiency in lower percentages it is always the little tweeks that get the gains. For testing I'd be using more dust material to get better averages. Obviously the 3d printed parts could be smoother and may get some gains, but that may be better for final stages due to the hassle and time needed. Is it possible to add a small vacuum port to each bucket to ensure negative pressure? Maybe a baffle or long tube to reduce the slow moving dust from being pulled back into the system. I'd try tapping into the main line for the buckets at a point with the highest velocity, which may be before the cyclone units.

I wonder if it would perform differently if you plumbed before the blower rather than after

I think you would probably see good gains figuring out how to slow down the air velocity (not the cfm, just the speed) in the fines collection section. You mention that they use an 8" instead of 6" body - that's a huge difference in air speed at the same flow rate. Maybe a 6" x 8" transition immediately after the impeller / as part of the impeller so the fines collection side is at a lower air speed? A third unit also in parallel would drop the air speed by a third. May also be worth playing with the geometry of the outlet. Could mess with length, diameter, shape, adding baffles to counter the centrifugal flow.

The "shaft" of the part with the fins should be a cone shape as well, so it's not only introducing the spin, but give the particles an initial direction towards the side wall, that way you're not relying on the centrifugal force alone. Also, glue a small lip on one side of the collection outlet inside the tube, so the particles are not just keep spinning in circles but caught on this lip and fall into the bin. I think the series idea was good if you gave it a proper chance. As you concluded in the first video, airspeed makes a big difference. With two in series, you essentially doubled the airspeed, that's why did it preform worse than in parallel. So the correct test would've been a single unit vs two in series. Obviously the more chance you give the particles to fall out the better, so the sweet spot is probably four units total. Two in series and those two in parallel. And if you don't insist on acrylic tubing (which is the main bulk of the cost), it could be made cheap (PVC pipes).

Print in ABS and acetone vapor smooth

As previously indicated, the main issue is that your exit port is not tangential to the outside of the tube. That discontinuity destroys laminar flow, and the dust does not cleanly exit into the tub. It remains suspended, and can exit with the air. You only have centrifugal force, which diminishes, not gravity (as in a cyclone) which does not. BTW, dust canister pressurization is NOT an indication of airflow- once established, the pressure will not fluctuate and no flow will happen (where would it go?).

Do the stl files use the more aggressive stator blades or original?

Saw the thumbnail and thought, "cool, but it's probably a bit overkill, no?" Then I saw that you make guitars. Yup, all the dust collection and separating you can manage is probably still not enough.

I would have to spend more that 500 bc I don't have a 3d printer....great video!

You can't easily ground non conductive materials to counter static electricity, you could however try a conductive paint or coating. I've DIY'd such a coating from raphite powder and varnish. It was very cheap and worked reasonably well! Would definitely be enough for ESD

Huge miss on the series system testing. As you identified your setup allowed short-circuiting. Need individual containers, with an inline mounting system. It will likely perform much better than parallel (this is what we do in mineral processing to get the metals out of the ore…).

I am new to 3d printing. What infill percentages/patterns are you using? I've heard that 3d prints aren't air tight. Are you coating this with poly to seal them up?

maybe you can try to mimic how air compressor in turbines work, you the impeller with X fins, that collect heavy particles, but in the second impeller make it with more fins for finer particles

Most cyclones have the suction at the end of the system, creating a negative pressure in the barrel (my barrel implodes every once in a while).It feels like that would actually suck up the dust into the barrels more, where as now you're pushing the material through and the barrels might be so full of pressured air that smaller dust particles don't see it as the path of least resistance? So try vacuum instead of pressure, is what I'm suggesting. But I'm not a fluid dynamics expert at all.

Using ribbed flex tube slows air flow and adds turbulence. Bends and every fitting also impedes air movement. Smooth wall tube is best. Series would only work if you had 4 collection cans, not very practical. Sticking with my Thien baffle for now.

Where did you get the container from?

how much saw dust was on the inside of the duct tape on that Y joint?

Another fantastic video, how far are you planning on going with these experiments? - Larger Tube? Different designs such as the G800? As with everything, there's a trade-off and their newer designs (Especially the G1000) appear to use more of a traditional cyclone design, so I think the main point of the G700 design is to reduce overall height rather than to improve filtering. It may well be that the original G700 design was already on the edge of the performance limit before you changed it to use a smaller tube diameter so that might be why you're seeing such a dramatic difference with the smaller tube.

You need to measure the amount of material comes out the exit and then subtract that from the initial volume. This is because material is still trapped in the whole system.

Yeah, I didn't expect the series to win, but thank for entertaining the thoughts of us armchair experts.

The issue is the walls of the pipe and the exit hole. The exit hole should be as large as possible, and go from the sides of the cylinder down into the hole. That'll sort it.

I think the new impellers are still too shallow - the fine dust is staying in the middle chamber for too long.

I notice in your last video and this video, you always measure what is captured and through subtraction determine the efficiency. But this doesn't account for any dust trapped in your hoses, in the separators, etc. Instead of that, you should try measuring the weight of the final filter before and after filtration to determine what the filter is trapping that the separator is not trapping.

The design seems weird, you have a lot of dead space at the end there where dust gets trapped til it eventually gets large enough to be pushed into the hole. Maybe you could close that deadspace off a bit to where its flush with the edge of the find particle hole?

Erm, what on Earth happened to my comment? Did it get scrubbed because I had a link in it? Anyway, it was about changing the shape/size of the drops going into the bins. The response was positive, although the modification was too extensive to fit into these quick alterations. I do agree that it's a fairly extensive bit of surgery, but I hope that the response suggests that we' might get a third video in this series?

Maybe the impeller needs to go the opposite direction. Instead of being more aggressive it should be less aggressive. Further once you go beyond the tuning of the separator your efficiency will get worse again. Your threading the needle......here!

If you want, you can try with multicyclone. Like this one.kzfaq.info/get/bejne/peCpZ8R5y7CymHU.htmlsi=CAp-hwHnWBWXTzWX This is for water but you get an idea. Nice and very little 3d printing necessary. I would prefer vertical multicyclone cause gravity helps there even at low velocity. And you dont need to overcome it as in horizontal ones. At this level you can read some engineering handbook with calculations and optimize system to your own needs. I wish you good luck!

When building a test/prototype design, I find it's best to use fasteners wherever possible .. glue is for the final assembly/small parts. ;) Also, one thing I haven't seen someone do is upscale a Dyson design. With the Dyson vacuums, they use a bunch of small cones (in parallel) to create multiple vortexes that spin against a holed cone, it's very effective on tiny particles, be curious how well it'd work for workshops. Dyson's latest: kzfaq.info/get/bejne/hZNojNaSmZq8ooU.html

Looks like your blower is too powerful for 2 tubes. The fact that 2.5" hose got you better result means each tube need to get a lower air speed. Maybe it's time to add a 3rd tube.

I think you feed too much material at the same time (is not how it will work once attached to a machine)