A continuous problem – Conveyor belts

One of my eventual goals is to set up some sort of conveyor system for the Mendel i’m building to allow continuous unattended printing. Wanting to learn more I took a look at the reprap.org wiki page but was surprised to see there is no references made on there to conveyor belt systems.

So I decided to put together my own summary of other peoples ideas and progress towards the goal of unattended continuous printing.

The traditional conveyor belt system

This is the most common and therefore well developed idea for obvious reasons. In short, the idea is to use a looped belt that passes over the printer bed. Once a print job is complete the belt moves the objects off the bed, peeling as it goes, until the object falls into a collection area.

From what I can tell Charles Pax has the most advanced working prototype so far.

His makerbot appears to be using a heated print bed in combination with a motorised mylar conveyor belt. This system looks great for print small or rounded objects on the relatively petite print bed used by the makerbot cupcake CNC.

However, as nophead pointed out repeatedly on the reprap forums, when you try to scale this up to printing large objects on the Mendel’s much larger build area the problem of warping becomes huge. Unlike a traditional print bed, a conveyor belt print is only held flat by the tension supplied at either end. As such, no amount of tension will keep a tissue box sized object from lifting the mylar print belt. This would then lead to the object warping.

One possible solution to this problem is to use atmospheric pressure to pin the print belt to the bed by creating a vacuum on the underside. This idea was explored by nophead on his blog some time ago. This can be see in his time lapsed video below of a part cooling down after a print.

In theory* a vacuum bed could supply a force of around 1kg per centimeter. I have it on good authority that this is not enough enough to stop the belt from lifting with a warping object. There is also another problem, it seems from anecdotal evidence that kapton tape starts to lose its adhesive properties after a number of uses. This would mean the constant replacement of the belt material and wouldnt allow for true continues printing. I wonder if the mylar tape used by Charles Pax has the same problem?

The reel to reel conveyor belt

This system would involve a large spool of kapton or mylar tape being mechanically unspooled at one end, drawn across the bed and then respooled at the other. If a vacuum system was used this could possibly solve any lifting problems and reel to reel conveyor belt also wouldnt suffer from degradation as its only used a few times. It would also remove the problem of how to reliably join kapton and mylar tape to make belts.

For obvious reasons this is not the most economical or environmental approach. As far as I can tell it hasn’t been tried with a reprap type printer.

The segmented conveyor belt

This is a system that uses a conveyor belt much like the tracks on a tank. As each track is a solid segment that can be stiff enough to resist the warping of the plastic it could do away with the need for a vacuum pump.

However it is likely this would be a very complicated and possibly expensive build build that could be prone to mechanical break down. Some commercial products have been suggested such as the one shown below.


Also a problem with this idea is that the tracks would leave marks on any printed objects where each segment meets. As far as I can tell no one has tried this approach.

The ‘Lazy Susan‘ rotating print bed

This idea is simple enough, have two or more printer bed surfaces that rotate into and out of the printing area. There are a lot of problems with this idea, not least that you still then require a way to physically remove the part from the bed autonomously and that the Mendel would have to be redesigned to accommodate the path of rotation.

As far as I can tell no one has tried this approach.

The changeable plate based print bed

In this approach a number of printer bed ‘plates’ are used. One approach could be have a stack of plates at the rear of the Mendel and one by one they are fed into the printer. One a print is complete the plate is then moved off the bed from the front of the printer and into some sort of collection area.

Here is a video of a Hass pallet pool machine, it gives you the general idea. It uses multiple ‘pallet’ to store objects to be worked on. It then collects the pallet, undertakes a task and returns the pallet with the completed object. Thanks Dale for pointing this example out.

This has many advantages in that there is no need for a vacuum solution or any requirement for a flexible conveyor material. Unfortunately it would also be difficult to construct and not a true ‘continuous’ printing system. As far as I can tell no one has tried this approach either.


So as you can see there is no shortage of different approaches and ideas for how to achieve continuous, reliable and unattended printing. I personally feel that with enough tinkering any of these different approaches could be made to work. However they all seem to fall down in one or more ways:

  • They reduce the total percentage of parts that can be self replicated by a Mendel.
  • They add considerably to the build cost of a reprap.
  • They add considerable complexity to DIY.
  • They require the replacements of perishables. Eg: caypton tap.

I have a few idea of my own about how I may the tackle the problem, but that’s for another day.


* Just how much force can be supplied by atmospheric pressure at sea level using a  realistically priced vacuum pump?

Lets assume you go all out and buy a $300 USD vacuum pump that supplies a 15 micron vacuum level. Lets also assuming your bed isnt a perfect seal and you can only reach 100micron of vacume. This converts to 13 pascals of pressure being maintained under the bed, pushing upwards.

Its worth noting here just what a pascal is. A pascal is a measure of force per unit area of 1N per meter squared.  Force is equal to mass by acceleration (F=ma). So an object weighing 1kg thats accelerated down by gravity at 9.8ms^-2 exerts a force of 9.8N.

Now assuming your ambient air temperature is 20 degrees C and your at sea level (STP) , the ambient air pressure is 100,000 pascals (100kPa). This is equal to just under 10,000kg spread over one square meter. To put this in perspective, thats equal to the same force that 10,000 1kg bags of flour would exert on a 1m by 1m table. So why are we not all compressed into the floor by the 10 tons of atmosphere stacked on our heads? Because we are mostly made of water which is incompressible and our body’s are at equilibrium with the pressure around us.

Anyway, so thats just under 10,000 kg per meter squared (Kg.m-2)  of atmospheric pressure pushing down on the printer bed with only 13Kg.m^-2 pushing up from below (near vacuum). This gives a product of 9,987 kg.m^-2 pushing down, or just under 1kg per square cm.

Thanks nophead for pointing out a mistake I made.

About Richard

I am a Materials Engineering working in the field of Magnetic Materials in Melbourne, Australia. This blog covers my personal interest in all things CNC.
This entry was posted in A Reprap Project and tagged , , , , , . Bookmark the permalink.

8 Responses to A continuous problem – Conveyor belts

  1. nophead says:

    You are out by a factor of 10. Pascals are force per unit area but force is measured in Newton, not Kg.

    I have found that atmospheric pressure is not enough to combat warping force near the corners. A thicker sheet, rather than a film, would be needed to average of the force. This is the how how rafts work and indeed perhaps extruding a raft onto a conveyor would work. But the downside of rafts outweighs the advantage of a conveyor for me. That is why I abandoned the idea.

  2. Dale Dunn says:

    Food for thought. Perhaps conveyors are not the way to go.

    Conventional manufacturing uses conveyors for conveying things from one place to another. For work done in some particular station, interchangeable pallets are used. The machines can automatically exchange pallets, which the workers load and unload outside the work zone. This is called a pallet pool. The machine keeps working. The more pallets you have, the longer the machine can run unattended. Search for “Haas pallet pool” to see examples from a popular machine tool builder.

    If some particular kind of heated bed produces optimal results, then build those heated beds as pallets. Build enough of them for the machine to run as long as you want. Unload them and prepare them for use later, or while the machine is still running. As few as two may be enough to keep a machine running continuously, if you can be there to unload and prep one pallet while the other is in the machine.

    Since a palletized system by definition involves easily replaced work surfaces, this would also allow for tailoring the pallets to the material being printed, size of the object being printed, whatever.


  3. stevew says:

    See: Hexapod Robot CNC, an expensive toy with 18(?) servos for the idea of a fixed work surface and a travelling tool. No help for a heated bed, but the idea that the precision tool move itself along a surface. Imagine a RepRap on wheels or treads with no Y-axis bed only the surface beneath over which it moved. Give it two ranges of Y-axis movement one fine and one coarse. Print say 200 mm in fine Y control then move forward 200 mm and repeat. Limited by power, filament, and smooth floor. A garden snail trail not of slime but RP parts. I think the Mori Seki and Haas pallet systems hold promise as RepRap pallets should not require fixturing that CNC machine do. I’m approaching a heated metal bed after A. Bowyer’s Dibond bed and plan on removable glass build surfaces for my Mendel. 1950’s Juke Box tech pick and place as nophead has done. . .

  4. Richard says:

    Like most things it seems, this has already been done! At least for a CNC robot.

    Take a look at the link below, its a very fine example of what a ‘hobbyist’ can come up with. Although I’m tipping he has a bit more experience than the average Joe.

    It will be an interesting day when someone makes a plastic extruding version. Imagine a spider like 6 legged robot with the extruder in its tail printing objects on what ever surface it pleases..

    I look forward to seeing how you go with your pallet style glass bed build.

  5. io says:

    Why you don’t use a robot arm that grabs the object and moves it away?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s