If I do that it is to flatter a little my ego, but especially to show that anyone can do the same, I am no one.

I dare you, please: do better than me


In this article I will present you how I made the main part of a vacuum bot: The vacuum system.

Through this article I will detail the complete process of prototyping.

As I'm not an expert in cleaning vacuum tools, sometime I made choices based more on my intuition than on the result of a calculation.

A simple question to rule them all

The first question I choose to answer is: "Should I use a system with or without a dust bag" ?

Based on my experience and the fact that I do not have tools needed to make vacuum with bag, I chose to make a model without dust bag.

Meaning I will need a system to separate dust from the airflow.

There is a huge list of technology of dust collector on Wikipedia. After a little thinking I choose to use a simple cyclone dust separator.

The next question is: "How do I make this kind of thing?"
First reflex: go see if someone has not already done one.

I found a lot of "classic" dust separator using a conic shape, but at this time it seemed difficult  for me to model this shape (I was a beginner on Freecad), plus this kind of separator is high and I have height constraints.

Conical cyclonic dust separator

I finally found another kind of separator (simple shape, adjustable height, very adaptable, and very efficient): the cylindrical cyclonic dust separator.

Cylindrical cyclonic dust separator

From the idea, extract the prototype, if possible a working one

Ok, now I have a global idea of what I will use. However there are 2 main issues to be solved

  • Which fan to use? (size?, power?, etc.)
  • What dimensions to give to the dust storage?

The Fan

Of course I was inspired by the models available on the market. It seems that the biggest ventilation used is about 60 CFM (cubic foot per minute).

I managed to find a very suitable fan: The BL447-04W-B49 11028
- 12 volts.
- 2.0 amps.
- 11cm centrifugal fan blower.
- Able to flow 62.49 CFM.

A big fan for a vacuum bot !

The Dust Storage

At that moment, I did not want to print the dust container (too long). So I choose to use of a piece of PVC pipe of 10 cm diameter (very cheap).
There is 2 advantage to using this type of pipe.
- The shape is close of the fan shape which will simplify the assembly later
- The possibility of adding a hermetic screw cap that will allow easily empty the dust tank.

Prototype v0

For the first prototype I tried to keep things simple, just stack all components one on the other.

Little explanation:
At this time I had not planned to present my project, so I had to dig long to bring a photo usable.

The yellow triangles represent the flow of air and dust.
The salmon colored triangles represent the filtered (clean) air.
The stale air enters from the side of the system, passes through the separation cyclone (this is where magic happens) and exits through the central part of the system (where the fan should be).

This first prototype was functional, but not very efficient.
The main defects were:

  • Fine dust and very light objects did not stay in the dust bin and are forced out of the system.
  • Too high model (around 15cm).

Prototype v0.5

I finally manage the problem of light objects by modifying the cyclone dust separator and the adding a filter made from a teaspoon.

After a long dig into my old stuff, I even found a demonstration video of the prototype.

As you can see, my test lab is not as developed as Dyson or Boston Dynamics.
It's in my mind the art of DIY.
(At the end of the video you can see the teaspoon filter).

Due to my studies, I had to move, so I put the project on hold for a few months. After this period I took over the project but I focused on the mobile part.
I find much more fun to have a prototype that can move for real, (even if this imply to make changes later) than to simulate every detail on CAD software's for months.

I will detail the moving part (and autonomous displacement) in another post.
At this moment, the bot was remote controlled.

Around the period of December 2018, I finish to assemble my first functional prototype.

"Without further ado, with its post apocalypse look: My first prototype!!"

One of my friend made a funny short montage about it !

"When your roommate lazy to pass the broom..."
"So he makes a robot vacuum cleaner..."

NB: If you take a close look, you can see I've done some modifications between the first assembly and the the montage. I've had to change motors and layout of components, I will explain why in a next article.

Since I was able to test prototype in real conditions, several defects were highlighted:

  1. Some dust is not sucked because the suction is too weak or the dust is stuck to the ground by static electricity (usually textile fiber).
  2. The system is not suitable for changes in terrain heights.
  3. The system is too high.

It was time to imagine a new prototype.

Prototype v1

This upgraded version is a complete revision of the vacuum tool:

  • New design of the dust separator
  • Add of a lateral brush
  • Advanced emptying system

New design of the dust separator

In order to reduce the height of the system I was inspired by a handheld vacuum cleaner from the brand "Black & Decker". As you can see the dust collector is horizontal.

So I made my own version on Freecad.
NB: The blue part represent the PVC pipe (dust storage).

This design improve suction performance because of the short distance between the entry pipe and the ground.

Lateral Brush

"Some dust is not sucked because the suction is too weak or the dust is stuck to the ground by static electricity (usually textile fiber)".

I chose to solve this problem by adding a side brush. In addition to solving the problem it allows to enlarge the work surface of the robot.

I began by looking at available products in the market. Many of them use a nylon brush. The problem is that I didn't have this kind of brush and I didn't want to order expensive spare parts pack from Roomba. Furthermore I  only needed one side brush, not the complete collection...

The complete collection

Out of curiosity I tried to print some 3D model of brush. Unfortunately the only printable material I had at this time was PLA, so the brushes were too fragile to be used.

I did not get discouraged and I widened my search to find out that agricultural machinery could be a great inspiration because an modern hay rack did exactly what I needed.

This system (very similar to that of a cyclic helix pitch) is too complex to be used in an hobbyist project.

To simplify this system while keeping its advantages I first thought to make a forward inclined brush.

But the system was too complicated to adjust and when it was badly adjusted it spent its time bouncing. His bounds were an additional source of noise and drastically reduced the efficiency of the side brush.

So I have to look for a new idea. During my research I came across a machine called street sweeper.

It's something very similar that what I'm trying to do.

So I left a simpler idea: a flat brush.
I finally came up with the idea of using toothbrush heads to make a nylon brush.

Then I had to find a way to attach this brush and adapt it to changing terrain heights.

After some tests, I finally choose to use a 5-arm brush.

5-arm brush almost assembled

Advanced emptying system

If with the vertical prototype I needed only one attachment point at the base of the dust tank. On the horizontal prototype, I need an attachment point on each side.

The design of the part in charge of maintaining the vacuuming tools took time and asked me to create several prototypes because I wanted a way to empty the dust tank very easily and very quickly.

I finally "gave birth" to this system.

As you can see It's quite easy to empty the dust bag. Even better you can wash the filter of the bag easily !

So I explained the big steps in creating the vacuum system. I hope I was understandable.
You can find this version of this project on my twitter:

Used Tools

CAD software: Freecad
3D printer: Micro Delta Rework

(Website I use to get fresh ideas)
- Thingiverse
- Instructables
- Yeggi


The realization of this kind of project may seem simple, yet it requires a lot of energy and reflection. In this article I have shown only the most important of the many changes I made to the original concept.

When I started this project, I did not plan to present my design process, that's why I did not keep all the files related to this project.

3D files of the latest version of this project will be freely available, in FCSTD (Freecad) and STL format. Feel free to reuse it, modify it, etc.

And remember, I'm not a mechanical engineer, I do not have a course in engineering, all I know I've learned it alone. Meaning you can do it too.

Social stuff / Questions / Comments

Feel free to reach or tips me !

Mail: a_ghost_soul@protonmail.com
Twitter: @GhostAgs

If you appreciate my work please consider make a donation
Tipeee: https://fr.tipeee.com/ags-syndrome