A 3D printed, raspberry-brained ROV project

[gwenaskell]

Hello everyone,

Greetings from France. As some reported virus still locks us there, I took advantage of the last few months to immerse myself into a project that had been haunting my mind for a while. I have indeed been sailing across the atlantic coast since my young age, but since I'm a better engineer than a scuba diver, I wanted to build myself a nice-looking drone to be able to explore the depths of these beautiful areas.

I decided to make quite a few unusual design choices regarding the hull and the propulsion system, motivated by the opportunity to broaden the technical aspects of the project and obtain a nice hydrodynamic design in the end. You can judge for yourself. Here is a snapshot of the unfinished CAD assembly, with still a lot of work to be done (I actually did not expect my software to struggle that much on the hull drawing!):

disposition7.jpg (34.07 KiB) Viewed 15627 times

The overall length is about 50cm. As you may see, one big specificity is these two front thrusters mounted on rotating axes which can perform a 150° rotation in both direction (forward/backward, from the vertical position). These are connected to stepper motors that adjust the position based on the feedback of the IMU. The control program is handled by a raspberry Pi, linked to the surface through a CAT6 ethernet cable (which is one of the last items I haven't purchased yet).
This will provide 5 degrees of freedom, which it enough for a natural pilot control.

apart from that, the main specifications are :
- top speed of 4-5 knots (the front brushless motors can provide 2 kg of thrust each, it should be enough)
- rated depth: 30-50 meters (deepest values I could find in the areas I could explore)
- video stream with low latency
- slightly positive buoyancy (as most of ROVs)
- keep price as low as possible

Another specific aspect of this ROV is the hull structure. I am going to use my 3D printer to print the outer skin of the hull (2 mm thick) to replicate the CAD model accurately. This skin will be reinforced and waterproofed on the inner side by a fiber glass + epoxy composite (2-3 mm thick). The hull will eventually be constrained by a few perpendicular frames to provide more rigidity. I hope this will be enough to handle the pressure of 3-5 bars.

For now, I am still working on the hull design (especially splitting it in a lower and and a removable upper part), but the main blocking points I see coming are:

- a waterproofing solution for the thruster axes. I tested simple O-rings surrounded by a 3d printed shaft but it provided too much friction. I am thinking of a "twistable" hose that would be pressed on the hull on one side and the axis on the other side, but it requires a thin, flexible tube and I am not sure if it could handle the pressure

- the printing of the outer hull, that produced retractions on a few tests I made with ASA (similar to ABS in its printing properties) and inaccurate dimensions on the XY plane (of the printing. The Z axis of printing is the forward axis of the ROV). It is gonna be a problem when I subdivide each side of the hull because I can't print each side at once. (because of overhangs)

- the forming of the front glass, which is made of a 4 mm layer of plexiglas that I will try to bend in an oven. I hope this won't lead to high optical distorsions. I don't know either how to do the mold yet.

[fryslan76]

Nice looking design.

For the steppermotor, I would guess that they can handle just like brushless motors a bit of water. So place it in a wet compartment and only worry about the wire which need to come into the dry section. Other option is to replace the steppermotor with a servo, which can be bought waterproof or made waterproof.

[gwenaskell]

Hi,
Thanks! Yes indeed, it might be a good solution as well. It could even fill the stator with some kind of wax or mineral oil. It just won't be convenient to empty the water after the dive, but it should work anyway.

[asesorplaza1]

A very nice project, which you have fun building, a greeting from Spain.

[gwenaskell]

Thanks. I hope it won't look like a swimming pool after the first dive as the separation line between the hull components has a significant length. The split section is made up of two joins to reduce the risk of leak :

hull-joint-section.png (211.01 KiB) Viewed 15584 times

Bolts will be inserted in between at regular intervals.

I made a little sample to test waterproofness, but I don't have a compression chamber to verify this in real conditions. Don't know if anyone ever tried to make a DIY chamber using a car compressor and a pressure cooker, basically all the stuff I have in my hands

In the meantime, water pressure could maybe help tightening the area around the joints, but I would not rely on this hypothesis...

[asesorplaza1]

Good afternoon from Spain.
If you have access to a dive club, maybe they have a hyperbaric chamber, which can help you do the pressure tests, but it's hard to be let to use it for your project.
With a car compressor what you can do is by attaching the air outlet to one of the dry compartment wires, "blowing" air inside the ROV, you're doing the same thing, but conversely, instead of being the pressure outside the ROV, you're putting it inside, if you can hold the pressure like it's a car wheel , the dry compartment is waterproofed, if air leaks occur, it means that you have some pore through which the air escapes and in your case it will be where the water enters the dry compartment.
You can also buy in the Far East, an empty pump for 30'00 euros, to do the tests, if you keep the pressure negative or empty, your ROV will be perfect, if you do not keep the empty, the loss occurs by the pores that may have its dry compartment.
Greetings.

https://es.aliexpress.com/item/40007047 ... 63c0ndLSOG

[Oddmar]

Bear with me...been up all night trying to install Windblows 10 on a broken laptop and then debloat it.

I've noticed a trend where people with a new 3D printer try and print Everything, rather than just print things they've designed and BUY other common parts.

Like a guy trying to print a waterproof 4" tube, not as an experiment to see if it was possible to print it, but because he needed a 4" tube. The simple solution would be, go to the hardware store and purchase a piece of 4" PVC pipe.

So my advice to you is, knowing how difficult it is to print a watertight container...Don't.

Could you use a piece of pipe for your WTC and just print the shell, so you have a sleek hydrodynamic body over a traditional WTC? I would think that would be easier than trying to get the printed shell watertight. Or rather than 3D printing the outer shell, what I would do, if it were ME doing it, is shape the shell from foam, then coat with fiberglass cloth and epoxy resin, then dissolve the foam core with acetone or lacquer thinner.

How about using servos to tilt the front thrusters, rather than steppers? I doubt you need the precision of steppers for this task, and you could mount two mineral-oil-filled servos outside the WTC inside the 'hydroshell'. This would be easier than mounting them inside the WTC then dealing with seals and water ingress.

[gwenaskell]

Hi,
Thanks for your thoughtful answers.

Reversing the compressor airflow is an interesting idea, I'm not sure if this is possible to plug the hose to the air inlet on the one I have, I'll check.

I've noticed a trend where people with a new 3D printer try and print Everything, rather than just print things they've designed and BUY other common parts.

Like a guy trying to print a waterproof 4" tube, not as an experiment to see if it was possible to print it, but because he needed a 4" tube. The simple solution would be, go to the hardware store and purchase a piece of 4" PVC pipe.

So my advice to you is, knowing how difficult it is to print a watertight container...Don't.

I agree. I have seen quite a lot of guys on the internet 3D printing every single component they need when they could find an equivalent part on the internet or design it easily with traditional methods and get a stronger result.
I think there are two reasons for this : being intimidated with the knowledge, equipment and trial-and-error it may require to build a complex shape out of nothing using pieces of wood, foam, metal a/o fiberglass ; and freeing his mind of having to think one's design in accordance with the components that exist on the market. (and cost, as well. some small parts may be very expensive sometimes).

I can relate this but it is not exactly what motivates me here, though. First thing is that I don't want to use a PVC pipe. This is obviously a simpler solution, but giving it a creative look was important to me.
The idea of making a PVC WTC surrounded by an outer shell came to my mind first. It is surely a very good compromise. If the solution I eventually decided does not succeed at enduring the pressure, I'll move back to this first option.
One reason not to choose this was displacement. The important amount of water that takes place between the WTC and the shell is like a "wasted mass", as it moves with the whole thing, it gives it more inertia and slows it down. It may harm the piloting experience a little bit and I would like to avoid that if possible. Getting a more compact assembly also makes it more transportable.
Another reason was that... I wanted to fiberglass something more seriously, I haven't seen many attempts to combine 3D printing with laminate and I wanted to see how performing this combination could be. 3D printing a mold is rather common (though removing the mold can be a tricky operation), but I thought that combining the visual, mechanical and environmental properties of printed ASA with fiberglass could be interesting, instead of wasting the material.

So, to answer your worries regarding waterproofness... That's right, waterproofing a 3D printed frame is a pain in the *** and that's not what I'm thinking about. The main structural part of the hull is going to be the epoxy laminate and this should, I guess, make the whole watertight. My concerns are more about the resistance to pressure and the reliability of the splitting section.

The reasons for using steppers rather than servos are angle and speed. My searches for a servo with at least 320° rotation angle and the ability to perform a full rotation in a quarter of second were not very successful. I don't need much torque here, so I eventually thought that steppers could be more adapted. As @asesorplaza1 pointed out, these are kind of brushless motors, so they might handle the salt water. I'll connect them to the outside with a kind of crankcase with a hole to fill it up with mineral oil. Replacing air with a liquid might prevent the water for coming in for a few minutes of dive.

[gwenaskell]

Small update there :

After a few adjustments, I have been able to print the first parts of the hull without bad surprises. I used a 0.6 nozzle to speed up the print, but the joint section is a bit rounded. I might have to start again with a smaller nozzle... and be more patient ! (as I'm now confident in the printer's capability to print large parts with this material).

The top-front part in the (DIY) printing chamber:

20210425_204438.jpg (265.11 KiB) Viewed 14880 times

With the top-back and the tail section:

20210502_121707.jpg (299.22 KiB) Viewed 14880 times

That was the easy part. I also worked on shaping the front glass by thermoforming a 4mm layer of plexiglas. Unfortunately, it turned out to be harder than I hoped. I designed a plaster mold and a support with a air chamber valve to blow the plexiglas heated at 160°C against the mold with the compressor. I tried different supports designs but did not succeeded in preventing air from leaking during compression.

That's the best result I got :

20210501_222851.jpg (281.29 KiB) Viewed 14880 times

So I eventually fell back on manually pushing the plexiglas against the mold. I managed to give it something close to the desired shape, but as I feared, it slightly degraded the optical quality of the glass.

20210502_121814.jpg (265.33 KiB) Viewed 14880 times

The optical defects are noticeable from a human eye, but surprisingly, through a camera it's not that disturbing.

20210502_121930.jpg (197.8 KiB) Viewed 14880 times

I think I would get a better result by using a heat gun instead of an oven and by vacuuming instead of pressurizing, but I need to make a positive of the mold. To be continued...

[fryslan76]

Nice result with the plexiglass, even with the small imperfections.