Best flexible filament for pneumatic actuators

[ATS_DEMUC 0]

Hello. I'm looking for a flexible filament to print objects with empty chambers that can be inflated/deflated using pressured air (like a balloon). The closest to silicone properties (I know it's impossible) the better.
I also might print those filaments along with PVA structures - thought it would be relevant for adhesion properties.

I saw this guide but didn't help much. And I can't even find videos from the filament brands to visualize their appearance and compliance post extrusion.

Thanks in advance

[gr5 2,268]

Printing water tight is hard enough.  Air tight even more difficult.

 

I think you can probably do it but you will have to adjust temperature, speed, flow to get perfect extrusion.  Or at least no underextrusion.

 

You probably want ninjaflex.  But I'm not certain.

 

The property that you have a lot of control over (by choosing a filament) is the flexibility, or elasticity.  Which for these materials is almost always measured in "shore hardness".  Beware that there are 3 scales A,B,C and each scale gets softer.  

 

You mentioned silicone.  If you've ever done molding with silicone you would know that you can buy many different shore hardnesses.

 

So if you have a know material (for example some silicone) and you already know it's shore hardness you can try to decide if you want more or less flexible.  If you don't, then I recommend starting with ninjaflex.

 

Like you probably already saw me say: print slow, hot, >100% flow and add oil to the filament.

 

[ATS_DEMUC 0]

 

20 hours ago, gr5 said:

You probably want ninjaflex.  But I'm not certain.

Yes, there are some academic works in soft robots using ninjaflex, but I've printed it before, and it isn't great. Was looking for something better. I might give it another try, and even coat the print with silicone for example. 

 

20 hours ago, gr5 said:

So if you have a know material (for example some silicone) and you already know it's shore hardness you can try to decide if you want more or less flexible.

 

 

I'll look into this. I've used different silicones before, this is a great advice.

 

20 hours ago, gr5 said:

Like you probably already saw me say: print slow, hot, >100% flow and add oil to the filament.

 

Yes.. i've seen it. I gotta dig again and search for your ninjaflex advices. And look specifically for printing it in Ultimaker3

 

 

Thanks a lot. I know it won't be easy, and it's an unexplored field, but I'm looking for ways to speed up the manufacture of soft robots.

[gr5 2,268]

Note that all filaments have a spec sheet hidden somewhere.  I usually google the material and the word "modulus" and usually the answer you need is in a pdf.  You could try "shore".  The softer the material, the more difficult it is to push through the bowden.  Write down the shore A scale hardness e.g. 85A (and the modulus in pascals in case some filaments only publish one value or the other).  So you can compare filaments.  For both values, lower values are more flexible.  Anything not on the "A " scale is likely too stiff but there are translation tables to convert for example "D" scale to "A" scale.  You can't get much softer than ninjaflex (85A).  Cheetah is 95A and much much easier to print.

 

These materials are squished in the feeder so they under-feed.  Because the feeder is moving the surface of the filament through the feeder at some specific amount but the cross section of the filament is now squeezed smaller.  So it's less volume than the printer thinks it is.  So sometimes you need to crank up the flow but if you go to high the pressure continues to build forever or until something gives.  The thing that usually gives is you get spaghetti in the feeder.  Set the feeder pressure to the minimum.  There is a screw hole - use the same allen wrench for all other screws on the printer and loosen that (CCW) until the mark is all the way up.  That will squeeze the filament less.

 

Use the high end of the recommended temp range for the material.

 

Be prepared to print as slow as 10mm/sec.

 

Print a small cube (maybe 30mm) with 100% infill.  Adjust things in the tune menu until you can't see any gaps in the infill.  Don't overextrude - it will plow troughs with snowbanks if you are overextruding.  Settings to adjust: flow, speed, temp.  Keep good notes.

 

Unspool a meter or so of filament behind the printer.  It hangs like a string.  put one drop of oil per meter.  The bowden should always have lots of small droplets in it.  Use any mineral oil such as "3 in 1" or "sewing machine oil".  Even baby oil should work.  Set a timer and every 60 minutes (or whatever works) unspool another meter of filament until it almost touches the floor and add another drop of oil below the feeder to dribble down the filament.

 

 

[geert_2 558]

Silicones are water-tight because they repel water. But they are *not* gas-tight, and not oil-tight either. Gasses, water-vapours, oils, liquid wax, and solvents do leak through. Don't ask how I know...  🙂

 

This is why you need to impregnate silicone moulds with plenty of silicone oil, prior to casting solvent-containing materials like PU or PMMA in it. So it is totally saturated with silicone oil. Otherwise the solvents seep into the mould, cure in there, and destroy the mould.

 

So a silicone balloon won't hold air for very long.

 

If it has to be gas-tight, you would need the materials used in real cylinders and tools for compressed air. I am not sure which materials that are, maybe neoprene rubbers, or some self-lubricating synthetic rubbers?

 

And indeed, as said above, silicones come in a lot of different hardnesses, ranging from very soft shore A 0...10 (which is almost like flesh) to 90...100, which feels like plastics such as PE and PP.

 

To print water-tight, you have to print very slow, and in very thin layers. And thus at low temperatures, so the material does not decompose due to sitting in the nozzle for too long. With thick layers, you get water-jets spurting out of hundreds of tiny openings in the side walls. Here too, don't ask how I know...  🙂

 

 

[ATS_DEMUC 0]

On 3/11/2022 at 1:16 PM, geert_2 said:

Silicones are water-tight because they repel water. But they are *not* gas-tight, and not oil-tight either. Gasses, water-vapours, oils, liquid wax, and solvents do leak through. Don't ask how I know...  🙂

 

This is why you need to impregnate silicone moulds with plenty of silicone oil, prior to casting solvent-containing materials like PU or PMMA in it. So it is totally saturated with silicone oil. Otherwise the solvents seep into the mould, cure in there, and destroy the mould.

 

So a silicone balloon won't hold air for very long.

I'm in a research lab at a University and we develop Soft Robots through EcoFlex / SmoothSil casting. We have decent results. We don't actually need the air to be held forever, since they are continuously getting actuated (air gets pushed in and out). Also, I have a on-off control system that accounts for any pressure loss, and pressurizes it again, if I want to maintain a said pressure.

The purpose of this post was exactly to find alternatives for the silicone casting method, since it takes too much time, specially in the design phase - some molds need to be re-designed, re-printed, and they are made of several different pieces.

There are lots of different successful academic works using 3D printing techniques to manufacture soft robots. When it comes to FDM, people managed to create some grippers with NinjaFlex. We have made one in our lab a while ago, using an older printer, but the flexibility can't be compared to silicone, and the layers tend to separate when actuated. Maybe we should've printed slower. Maybe Ultimaker could provide us better results

 

[ATS_DEMUC 0]

On 3/9/2022 at 2:44 PM, gr5 said:

Note that all filaments have a spec sheet hidden somewhere.  I usually google the material and the word "modulus" and usually the answer you need is in a pdf.  You could try "shore".  The softer the material, the more difficult it is to push through the bowden.  Write down the shore A scale hardness e.g. 85A (and the modulus in pascals in case some filaments only publish one value or the other).  So you can compare filaments.  For both values, lower values are more flexible.  Anything not on the "A " scale is likely too stiff but there are translation tables to convert for example "D" scale to "A" scale.  You can't get much softer than ninjaflex (85A).  Cheetah is 95A and much much easier to print.

 

These materials are squished in the feeder so they under-feed.  Because the feeder is moving the surface of the filament through the feeder at some specific amount but the cross section of the filament is now squeezed smaller.  So it's less volume than the printer thinks it is.  So sometimes you need to crank up the flow but if you go to high the pressure continues to build forever or until something gives.  The thing that usually gives is you get spaghetti in the feeder.  Set the feeder pressure to the minimum.  There is a screw hole - use the same allen wrench for all other screws on the printer and loosen that (CCW) until the mark is all the way up.  That will squeeze the filament less.

 

Use the high end of the recommended temp range for the material.

 

Be prepared to print as slow as 10mm/sec.

 

Print a small cube (maybe 30mm) with 100% infill.  Adjust things in the tune menu until you can't see any gaps in the infill.  Don't overextrude - it will plow troughs with snowbanks if you are overextruding.  Settings to adjust: flow, speed, temp.  Keep good notes.

 

Unspool a meter or so of filament behind the printer.  It hangs like a string.  put one drop of oil per meter.  The bowden should always have lots of small droplets in it.  Use any mineral oil such as "3 in 1" or "sewing machine oil".  Even baby oil should work.  Set a timer and every 60 minutes (or whatever works) unspool another meter of filament until it almost touches the floor and add another drop of oil below the feeder to dribble down the filament.

 

 

I've read all of this, and I got 2 reactions out of it:
- WOW, you definitely know your stuff. 
- Now I'm scared.

It doesn't sound difficult, but it sounds complex. It kinda defeats the premise I'm going for, which is a new and faster manufacturing process for soft robots (as explained in my previous comment).
I'm not sure I have the means to be present every hour to unspool filament.

Without wanting to get off-topic, since this question is related to my research in soft robot manufacture, do you have any specific advice for printing settings/tips for PVA or BVOH filaments? I have the spreadsheets, but you might have some additional info. Plus, keep in mind that I want to use them as the main filament, not the support one.

And thanks for the great amount of info you provided

Edited March 14, 2022 by ATS_DEMUC
typos

[gr5 2,268]

I don't know BVOH but the trick with PVA is to keep it very dry.  It absorbs water quickly (similar to nylon filaments).  So even a 10 hour print might be doing worse at the end of the print.  Certainly a few days having the spool on the back of the printer is going to "ruin" it.  But you can easily dry PVA.  i do it by unspooling the amount I need and sitting it on the heated bed with the spool on top and put a towel over all of that (and usually a blanket also) and letting it bake for 4 hours around 70C.

 

I bought a liter of dessicant beads (color changing - they change color when they need recharging).  I put about 1/2 cup of the dessicant beads in a container with holes inside a 2 gallon zip lock and store the PVA in there.  Even with all that I often have to re-dry the PVA.

 

You can tell if PVA is wet because you can see steam coming out of the nozzle when you first load the filament and extrude a little.

[ATS_DEMUC 0]

18 hours ago, gr5 said:

the trick with PVA is to keep it very dry.

I'm was a bit hesitant to heat the filament. We have a vacuum chamber, maybe I should use that too before every print. I did find odd that while printing PVA I used to hear a bit of crackling/popping, like when you are near a campfire and hear the wood crack. 
Also, sometimes there are spots where the filament burns, there's two or three really small (like 1mm^2) brown patches on my prints. Maybe that's due do humidity aswell?

[geert_2 558]

Vacuum and dessicant will help to dry filament, but may not be enough: you might need a bit of heat to shake the moisture loose from the plastic. If you are not sure about heating, don't go above handwarm: as long as you can comfortably touch it, it should be okay.

 

Maybe a resin-printer could be suitable for your prints? Some resin-printers can print flexible, rubber-like materials, at least the expensive professional machines with polyjet systems. Ask around in your university if someone has such a machine? E.g. an Engineering-department, or Product Development Faculty, or so? Such machines can also do higher volumes and lots of parts at the same time.

 

I have no experience with printing flexible materials myself, but from all what I have read here in the past, you need to print extremely slow (10mm/s), increase flow rate quite a bit, oil the bowden-tube or use a PTFE-tube, and then still have good luck. Trying to push flexible material through a tube is like trying to push a rubber band into a keyhole...

 

Concerning casting silicones: I have fast-curing ones that cure in 20 minutes. There do exist even faster ones, but these cure so fast that you can't even mix and pour them out nicely (some cure in 10 seconds and need dedicated mixing pistols and mixing tips...). Printing a mould in PLA at normal speed, then smoothing that mould with dichloromethane or chloroforme, and casting silicone might go faster than trying to print it very slowly in a flexible material. So I would suggest that you also look into optimising your mould-design for easier printing, moulding and de-moulding. Youtube has lots of good tutorials on moulding and casting, mostly from the pre-3D-printing ages, but they do apply to 3D-printing moulds as well. I learned a lot from them.

 

[ATS_DEMUC 0]

21 hours ago, geert_2 said:

Concerning casting silicones: I have fast-curing ones that cure in 20 minutes. There do exist even faster ones, but these cure so fast that you can't even mix and pour them out nicely

 

 

Fast curing silicones present an issue for us, since we like to apply vacuum after mixing the silicone components, and after pouring the silicone in the mold. That removes almost all the air inside the mixture, resulting in a uniform material, that doesn't get bubbles when pressurized.

 

21 hours ago, geert_2 said:

Printing a mould in PLA at normal speed, then smoothing that mould with dichloromethane or chloroforme, and casting silicone might go faster than trying to print it very slowly in a flexible material.

 

 

That smoothing technique is new to me. We never smooth the moulds. We do have a special coating, XTC-3D, to basically do the same, but we don't even use it because the printing finish is acceptable (to be honest, I print the molds in a Prusa MK3 original + a Bondtech extruder). Thanks for the info.

And yes, I do waste a fair share of my time trying to design better moulds. But it's just a pain in the bum when you start to design more and more complex actuators 😆

Edited March 17, 2022 by ATS_DEMUC
typos

[gr5 2,268]

I've seen people use vacuum after mixing the silicone but skip the part about vacuum after pouring the silicone and they still seem to get pretty good results.  I don't have a vacuum chamber so I get some bubbles but not a problem for my applications.

[ATS_DEMUC 0]

22 hours ago, gr5 said:

I've seen people use vacuum after mixing the silicone but skip the part about vacuum after pouring the silicone and they still seem to get pretty good results.  I don't have a vacuum chamber so I get some bubbles but not a problem for my applications.

It also depends on the type of the mold, and type of silicone. Some silicones have super high viscosity and don't "slide" as fast as needed. That problem is augmented if the mold has such complex geometries that the silicone has trouble filling them with just gravity pulling it down.

[geert_2 558]

22 hours ago, gr5 said:

I've seen people use vacuum after mixing the silicone but skip the part about vacuum after pouring the silicone and they still seem to get pretty good results.  I don't have a vacuum chamber so I get some bubbles but not a problem for my applications.

 

A few things I learned from experience and from tutorial videos:

 

In very slow curing silicones (24-48h curing time) bubbles generally have time to rise to the surface and pop by themself. Unless the silicone is very thick indeed, then vacuum may be needed to get them out. It depends on the silicone, time, and temperature. Also in complex moulds with undercuts, vacuum is desired: then you can get the air out of the undercuts and pores, and have the silicone being sucked into the collapsing vacuum afterwards. A shaker (similar to those used to make gypsum and cement flow) helps bubbles rising.

 

Disadvantage of slow-curing silicones is that it takes lots of time, of course, and that the tiniest leak will make the whole mould leak empty: it leaks through microscopic gaps of one micron.

 

In faster curing silicones, you may not have enough time to vacuum. Then the way to pour the silicone into the mould is critical to avoid bubbles:

- make sure there is no wind in the room, close doors and windows, stand still,

- pour from as high as you can, and still get it in the mould (this is why it has to be windstill): 1 meter is good,

- pour in a thin stream, slower and thinner is better (but don't exceed curing time of course),

- sometimes you can spray the mould or model with stuff that reduces surface tension, to avoid bubbles on the surface,

- do not pour directly onto a model, but pour besides the model, and let the silicone rise around the model,

- with compressed air you can make bubbles on the surface pop, but blow very very gently, otherwise everything will be covered with liquid silicone...

 

The first items in the list make most of the bubbles pop while pouring, and the latter prevent air from being trapped on the model. Together, this removes all big bubbles and reduces the amount of tiny bubbles a lot.

 

Storing silicones in the fridge increases working time (for fast curing ones), but risks getting condensation.

 

Some fast-curing silicones are delivered in syringes with mixing tips, and a pistol: this allows injecting them with some force, so it fills the mould better. Similar to injection moulding. Such as dental silicones, but they are rather brittle, not very flexible, and not very suitable for seals and balloons.

 

Also, for really viscous silicones, sometimes you can design the mould in such a way that the silicone does not have to flow by itself: you pour it into a thick blob centrally in the mould, and then it is forced with brute force into all cavities when closing the mould. But this obviously does not work for every mould, and not for too liquid silicones.

 

Some photos:

 

Dental silicones (quite pasty, viscous) in 50ml mixing tips, and a pistol. Thicker tips indicate more viscous materials. Most of these cure in 5 minutes, but the light yellow in only a few seconds.

 

 

Viscous enough to stay in shape without leaking away.

 

 

Mould for pouring liquid silicones around a model (here curing time is ca 20min). The red plasticine was used to hang the model on.

 

 

A very old mould for pouring silicone around a model, and then casting composites into that silicone mould. the hard shell is to give the mould enough stability. This was before the age of 3D-printing. The silicone mould is cut in zig-zag to make bot halves align well: looks ugly, but works very well. The hard shell also has alignment keys.

 

 

This is a mould for pasty silicones, using brute force to make the paste flow into all cavities and corners. This requires carefull design of the seams and air channels, so the air does not get trapped anywhere, and the paste can reach all corners. In addition, I provided lots of holes to inject compressed air into the seams after curing, to break the connection between mould and silicone. And I provided slits to insert screwdrivers to pry apart both halves. Otherwise it would be impossible to separate both halves and get the model out, due to the complex design. It took me some iterations to optimise this. The red area is casted silicone. (But I can't show you the rest of the inside of the mould.)

 

 

Hope some of these techniques could help those of you who want to caste silicones.

 

[ATS_DEMUC 0]

28 minutes ago, geert_2 said:

In faster curing silicones, you may not have enough time to vacuum. Then the way to pour the silicone into the mould is critical to avoid bubbles:

- make sure there is no wind in the room, close doors and windows, stand still,

- pour from as high as you can, and still get it in the mould (this is why it has to be windstill): 1 meter is good,

- pour in a thin stream, slower and thinner is better (but don't exceed curing time of course),

- sometimes you can spray the mould or model with stuff that reduces surface tension, to avoid bubbles on the surface,

- do not pour directly onto a model, but pour besides the model, and let the silicone rise around the model,

- with compressed air you can make bubbles on the surface pop, but blow very very gently, otherwise everything will be covered with liquid silicone...

Very interesting read. We usually work with slow curing silicones because our main goal is to create perfectly balanced actuators when they are under pressure i.e. all their parts inflate equally, there should be no zones of higher stress, unless they are designed to be. For that reason, we rely a lot on the vacuum chamber, and that takes time.

 

We are now in a moment of experimentation. We'll try different silicones and test their performance. Maybe we could/should refine the fabrication techniques. Those advices, even if for faster curing ones, might come in handy.