PrismX Open-Source 3D Printer

The PrismX Open Source 3D Printer has arrived.

The PrismX is based on a variant of the original Prism Mendel that I call Prism AliceBlue. PrismAB includes Terawatt customizations such as our leadscrews and very heavily modified Kuehling's x-ends. AliceBlue exposed the potential for remarkable performance in the Prism Mendel design on its 4th print. I can't say overstate its reliability. It runs on-and-on racking up printing hours, sometimes when it shouldn't, with very little maintenance required. For the record, neither PrismAB nor PrismX are based on the MendelMax, nor any other Prism derivatives, they are both based purely on the Prism Mendel RepRap with customizations and upgrades we've developed independently (or through collaboration we sought).

Every axis on the PrismX has been completely revised. To make this happen properly, I've created a (nearly) complete PrismX SCAD model which can be animated for motion.

The machine's y-axis is light and compact. The Y-axis belt uses a 48T GT2 pulley to wrap around the 20mm extrusion - with about 4-5mm clearance top and bottom of the extrusion. The Universal plate clears the extrusion by ~3mm.

For the z-axis I tucked the leadscrew closer to the frame, and pulled the z-axis smooth rail ~3mm outside of the frame. This maximizes x-axis travel. The z-axis and leadscrews are deluxe here. They're the same precision leadscrews we use with our MendelMaxes, except they are center tapped on one side for an M4 screw to provide a flush mount for the GT2 pulley w/8mm bore. The axial transmission system is as smooth as butter. It's a stack (bottom to top) of: GT2 Pulley, thrust bearing, acetal bushing top/bottom of leadscrew support (which is 5mm taller), thrust bearing, and leadscrew nut. The belt-driven z-axis has a 1:1: gear ratio between motor and leadscrews. When PrismX goes to production it will be delivered with a top-constraint for the z-axis that uses an acetal bushing. No other open-source design beats this one. Yes, it would work with cheaper leadscrews, if somebody wants to help with variations that'd be great!

The 00str00der is equally awesome. The PrismX can work with other extruders, but it's considered standard issue for the PrismX. This thing never wears out and keeps up with anything you throw at it. I've got plans to make another upgrade to the thing. Lee Miller designed it originally with concepts we collaborated on, Terawatt performed first printing tests with it, and Lee has tested it in a Bowden configuration. It's unstoppable. I've designed an adapter plate so it works with most "standard" x-carriages.

Some of the other highlights in the machine's design include the Sheliak x-ends and x-carriage. I customized the x-ends for the new leadscrew and z-rail positions. They're not compliant with other Mendel designs, but they can be adjusted in OpenSCAD for backwards compatibility. The x-carriage is customized for the 00str00der, and to achieve ~210mm x-axis travel on the PrismX (w/std. 400mm length extrusion). Most of the machine is customized, a few parts remain from the original and wonderful Prism Mendel - which owns all bases, if you ask me. Another important upgrade are the press-fit LM8UU bearing retainers with steel bearing-retainer rings. I could write a separate post about them alone.

I'm a programmer so I'm fixated on parametrized and modular SCAD code. This means I made the Prism very-easy to scale. This also means we have an upgrade kit for MendelMax owners. With a few pieces of 420mm extrusion, MMax owners can keep things like the t-slot nuts, motors, and electronics; then add vitamins, plastic parts, and transmission systems to upgrade to PrismX.

One more thing about some of my speed claims. The PrismX is wicked fast, and it gets this from excellent frame alignment. More importantly, its print quality is starting to approach that of parts I've had made on SLA printers. It's also one of the easiest repraps to assemble, perhaps the original rostock is easier. The gantry is easy to align during assembly, as everybody who's built a Prism knows. Overall - on the plastic parts - I spent countless hours, days, weeks wrestling with alignment issues on the scale of 0.1-0.25mm; and getting things like consistent press-fit tolerances. 80-20 rule all the way, baby! The result is frame strength and parts alignment that allows me to crank up the firmware and slic3r settings and maintain quality with the PrismX. To make sure the parts alignment is reproducible, we're molding the plastic parts for PrismX. Anybody want to make an all-metal PrismX before we get to it? I'd love to help...

Finally, special shout-out and thank you to @Buback, @egenoup, @akhlut, @Wired1 for all of their contributions to the Prism lineage. When the PrismX goes into production we'll publish the source code on Github and provide support through a PrismX google group (and more forums - TBD).


http://reprap.org/wiki/Prismx
http://prismx.terawattindustries.com

-Free

very interesting, the extruder is great, having a belted extruder is a major upgrade over the original geared ones and still over the hearingbone ones. what acceleration are you running at? also whats the resolution on each axis, i myself am building a larger machine and this has got me thinking about using larger pulleys, currently with 15 tooth pulley i can get 750mm/s with 7000mm/s/s acceleration, i think ive reached the limitations of the arduino as the motor doesnt just stall it sort of studders for longer than it would take to make the move, although there is some friction somewhere in my system that i must hunt down. if i put a 48 tooth pulley, man i could get some serious speed, like around 2400mm/s.

Edited 1 time(s). Last edit at 05/21/2013 10:01AM by aduy.

Agreed, the 00str00der is a major highlight. I'm not sure which is better: being able to calibrate the extruder by measuring the hobbed bolt, or having a gear/belt drive system with expected lifetime on the order of 1000s of kilometers.

I think it's awesome you're able to use high speed values on your machine! We use acceleration around to 2200 - 2500 on this machine currently. In my opinion, when we see firmware with 'stock' or adjusted acceleration values above 5000mm/s^2 there's probably room for real-world adjustment. If we do the math using acceleration values like 9000mm/s^2, then we expect the robot to reach speeds like 100mm/s in well under 1ms, and 400mm/s in about 1-2ms. This is unreasonable for NEMA17s that are carrying a weighted load.

The jerk setting in grbl-based firmwares like Marlin, Repetier, and more constrain the top speeds reached during print paths. We can set acceleration to 9000mm/s^2, or other ridiculously high values, and use jerk settings like 1 or 0.1, and print paths will be seriously constrained and never hit top speeds. Finally, extruder motion and E_JERK constrains the overall print speed - so even if other axes are raring to go(!) the extruder speed is a primary constraint.

You're right with a 48T pulley like the PrismX Y-Axis you'd probably get higher speed, but the motor needs more torque as the pulley size grows, so there's a tradeoff. If there's already some binding on the axis, a larger pulley could expose more binding or skipping.

Speed is just the icing on the cake of overall performance.

My hot end can hold its own in the speed department, its the Europa, basically a water cooled version of the sg2 on ebay, currently one of the bearings on my machine is sort of makeshift until I print the part for it so thats why its binding. what hot end are you using and what drivers?

The press-fit bearing holders with steel retainer rings on the PrismX are worth checking out. They're really nice and they can be modd'ed to fit almost anything.

Right now it's rolling with a j-head as standard issue, but I was referring to firmware settings for the extruder motion (which can constrain motion on other axes) - irrespective of the hotend type. The stepper drivers are A4988 Pololus.

ah yes those are nice. im using open rail on my machine though. perhaps in a later revision you might switch, its quite smooth and fast, and you dont have to worry about bent rods.

Nice work on the PrismX! Life takes over, so I haven't been able to revisit the original Prism design in a long time, But I'm glad to see someone taking it and evolving it!

Good Luck!

Hey man I'm so glad you chimed in on this thread. You've been on my mind often although I don't know you. I think we share a common interest in Astronomy, for what it's worth.

Thank you so very much for establishing a solid foundation to iterate upon, and most of all for sharing your work and thus providing an opportunity for us to take part in its evolution. All bases are belong to the Prism!

This is quite awesome, aluminum extrusion is extremely nice to work with and its very rigid. I have a couple suggestions for you to take into consideration. Just so you know normally these things are not important, but since your machine is so precise and highly tuned it could make a difference, especially at the high speed that it prints at. Sorry if im pointing out little things that probably don't matter, im just intrigued by 3d printing and this is a whole new level.

One thing I noticed is that the pulleys on the x axis are not in alignment with the belt clamps. if you make the belt clamps lower so they are at the same height as the bottom of the pulley and idler you will have more even belt tension. basically the theory goes that the belt tension gets low in the middle of the x axis and higher at the edges towards the pulley, this causes the belts to flap, and that's when you get a little bit of backlash if any, and its a very easy fix.

Second suggestion, replacing the smooth idler for the gt2 belt with a toothed pulley makes the prints much smoother. for example in this photo you can see there are small surface ripples which is caused by the gt2 belt on the smooth pulley. basically what happens is as the belt moves across the smooth idler it doesn't remain perfectly level, each tooth causes it to move up and down, increasing and decreasing belt tension as it travels across the axis, at low speeds its not noticeable but at high speed it becomes a vibration, which shows through in the print as a ripple on the surface. the ripples are very slight but hopefully you can see what im talking about. I think this reduces friction slightly as well

You're right and I had grown to accept that misalignment on the belt tensioners - it's slight but noticeable. I like your suggestions. We're going to re-launch the machine on our site and I'll start testing these revisions for production.

good man!

I've been thinking about the pulleys as molds are being finalized. Doesn't a pulley which has a diameter that's a multiple of 1/2 the belt-pitch get the same effect? In other words, for a GT2 belt, a pulley diameter of - just making this up - 49mm cancels the effect you describe. Then, for a GT3 belt w/3mm pitch, the diameter would be a multiple of 1.5mm - so a 45mm-diameter pulley cancels that effect. Does that make sense? I might still like toothed pulleys better...

Edited 2 time(s). Last edit at 05/24/2013 11:20AM by tenacious.

I think you may be right, however back when t5 was the standard, people had great results flipping the belts over to the smooth side.

Very nice evolution of the Prism, good work.

You should maybe also post in the Prism subforum for more people to find it.
Never mind, just saw your post there


I'm interested in upgrading my old Prism X axis.
Did you post the PrismX 3D parts sources somewhere as you were planning to ?
[github.com] seems unpopulated for now.


Thank you for supporting open-source development !

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

You got it! SCADs are finally cleaned up and ready for the MendelMax upgrade kit. I'm making further preparations today to get the code pushed to github.

Alright - the SCADs and STLs are uploaded to our Github and Bitbucket repos. This link is the top-level repository which contains all the software Terawatt officially supports for the Prismx:
https://github.com/Terawatt-Industries/prismx

The scads are in the hardware/prismx folder. STLs are in dist/stl.

Peoples will need a printed-parts BOM and a vitamins BOM. We're going to create a printed-parts BOM and "plate" STLs sometime soon, after our first pre-assembled machines are shipped. You don't need to wait for us to publish a printed-parts BOM. The "scad/terawatt-prism.scad" and "dist/stl/terawatt-prism.stl" files are complete models of the machine and can be used to visually describe the required printed parts (=BOM) and for assembly reference. For vitamins, that will have to wait 3-4 weeks as we are planning to publish the vitamins BOM and assembly guides when kits are available.

Edited 1 time(s). Last edit at 06/22/2013 05:47PM by tenacious.

Thank you very much !

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

So, in comparison to the MendelMax, how stable is this machine?

I am still looking for a new machine to build and my Reprap Air 2 is pretty wobbly, so I really would want to build a machine that is solid. No wobbling.

I have already bought a 18mm thick piece of MDF to function as a baseplate for the new machine to-be.

I'm biased. I'd like to see others start to build the machine but until assembly instructions are available I suppose such would take a core reprapper.

First please let me speak to development stability. The code is very stable. I've committed revisions recently for an upper leadscrew guidler and z-endstop. The filament spool holder is likely to be revised also, actually it's in a separate repo. Several parts are in separate repos, including the 00str00der, and I'm finishing up the external submodule references and plate SCADs/STLs for the basic set are in there now.

Now for frame stability. As you requested let's compare the PrismX 3D Printer design to its cousin, the MendelMax 1.0/1.5. The Mendelmax is also similar to the Prusa frame design, so I'll include the original Prusa in this analysis. These machines have an open-triangle design at the top. In other words, the triangle is not closed (joined) at the top. The Prism Mendel and all its variants, including PrismX, have a closed-triangle design at the top. From a geometrical perspective, a triangle which is closed at the top, like the PrismX is stronger than one which is not closed, like the MendelMax and Prusa. Of course in this specific comparison there are plastic parts that make theory less applicable without a more complex analysis that would include the strength of the adjoining parts, or by crush testing some models perhaps. However, this is all moot - putting weight on top of a printer is not an accurate measurement of frame strength, and in most cases the machine doesn't experience this (except the few pounds of filament typically placed on top of the MMax).. I'll get to side-to-side strength soon, please read on.

On the bottom, the MendelMax and Prusa have a second horizontal layer to allow room for a y-carriage and second plate above for the heatbed. Effectively, they could be described as a pyramid on risers. The Prism contrasts this with a closed triangle, all the way around. There is no second layer on the bottom, just thick plastic feet. We actually eliminated the need for two y-axis plates on the MMax and Prusa with Terawatt's universal y-plate. With those machines we have to print tall spacers to get the plate above the bottom of the gantry.

On the PrismX z-axis, the smooth rails are moved outside of the frame and clamped snugly to the frame. This adds to the side-to-side strength of the Prism design. The concept is similar to the MMax, but there's much less plastic joining the structural metal parts, and the rails are ~4mm outside the frame. Again - I'm biased - but the friggin' PrismX gantry is like a rock! To put this in perspective, consider that my workbench is slightly warped and not perfectly level. When I crank the PrismX to very high print speeds the amount of flex within the frame itself is very slight, practically negligible. Instead of flexing within the gantry, the whole thing rocks, like a rocking pyramid. The Terawatt plastic parts set is being fulfilled with softer molded plastic feet to absorb vibration and reduce rocking on uneven workbenches at top speed.

One of the main highlights is the plastic parts design. The PrismX design philosophy assumes the following axiom: Monolithic parts are prone to monolithic failures. PrismX parts are more atomic and clamping. They're intended to be moldable, easier to print, and this makes it easier to get consistent quality and parts alignment. There are almost no complex bridges and most overhangs are slight. A given plastic part on the PrismX usually has a simple job, and if the job of the part is more complex than many smaller parts do the job. For example, the x-ends are clamping, and they also have a separate bearing holders and motor/idler mount. In the end there are more parts to assemble but each part can be manufactured with higher quality. Assembly should only performed once, or rarely, and in this case - if a part breaks we probably don't have to print another monolithic copy to repair the machine.

Let me take another minute to describe what it's like toting them around, which I've done with various models over the past year. Like any standard 3D printer it's not designed for portability, and it's rather large and weighs a couple/few-dozen pounds so this assumes a grown adult. The PrismX can be picked up by its top extrusion, basically using that as a handle, and it's solid to the feel when picked up. The power supply has a mount that's attached to the frame. Once unplugged it can be toted as a single unit, carried around by its inbuilt handle with perhaps a second hand used to guide it. Not that any of this matters, it's just a side-effect of how consolidated the design is even without an enclosure.

Thank you for your very complete reply!

I am really leaning towards this printer now. Even though I am not a core-reprapper, and I have only built one printer before, this might be something for me.
Though I have to admit, I have built my own Y-bed that works fine, and I have made a lot of improvements on my own printer that I will probably also try to incorporate on my printer to-be.

So, I understand your PrismX uses less extrusions (which are expensive in my country) and uses very modular printed parts, something that I enjoy seeing because a lot of times I see parts that are just too big for me to print in an evening. Sometimes I come home around 22:00 and then I go to bed around 24:00 and this keeps me from being able to print four to six hour printjobs on a daily basis.

Edit:
I have checked out the Github repository and found out that I am probably not on the level I need to be to know what and how to build this printer. I haven't found things like measurements for the extrusions and however I did discover the location of the STL's, it seems not all the STL's in there are needed to complete the build.

I understand there will be more development before this becomes more userfriendly?
It sounds like the machine for me!

Edited 1 time(s). Last edit at 06/28/2013 05:55PM by Ohmarinus.

Right now we fulfill this printer as a complete pre-assembled machine. Isn't that more userfriendly than a kit?

There is a complete visual model of the machine found in "terawatt-prismx.scad". In lieu of an assembly guide, this file is a complete model of the machine. Extrusion length is 400mm, just like the original prism. By the way, the PrismX works with any reasonable length of extrusion.

PrismX assembly is similar to the original Prism Mendel. If you have built a 3D printer before, can find an old Prism Mendel assembly guide, and can print these parts...you have the building blocks to build your own. I think the biggest gap right now is the vitamins BOM?

There are tons of STLs! I think you said it's too many? I think you just need a couple READMEs that point you to dist/stl/plates?

If you're ahead of us and want to get started building a PrismX, you should get started, then let me know where you need support.

EDIT: to clarify the design only saves 1 piece of extrusion - wish it was more. "Being ahead of us" means to build a PrismX on your own without an assembly guide before we've made kits available (with the guide). That makes you an early adopter from my perspective. We're literally finishing our first pre-assembled batches now. That means our full-color assembly guide is forthcoming, but still weeks away. If you're considering becoming an early adopter or have already gotten started please consider contacting me directly - either here or on our site. Also join the Prism mendel google group and let me know where you need help so we can share with everybody. The group is just getting started so your help is easily noticed and greatly appreciated. [groups.google.com]

Edited 1 time(s). Last edit at 06/28/2013 11:29PM by tenacious.

Hi, thanks for your reply.

Do you think I could build the PrismX with the same extrusions that I would use for a MendelMax? Because I have the option of buying a cheap frame kit with just the extrusions and a 100 t-slots for free with it.

Sounds like a good deal!

And yes, I would love to go into the woods and try to figure out the printer build on my own, though I don't have any idea what dist and plates are? Only heard about STL's.

Will start printing some STL's today


Edit:
Maybe it's a good idea for me to first build a MendelMax and later when more detailed information comes out, convert it to a PrismX? I'll also have a look into the Prism Mendel. Also, I haven't started printing the parts yet because I ordered a special translucent ABS variant that I want to build the frame parts out of Kinky!

Edited 1 time(s). Last edit at 06/30/2013 06:10AM by Ohmarinus.

Waaaaaant! Thanks a ton for helping me out with the belt the other day Free. I wont name the seller, but I have to be honest, the build quality of my prusa i3 all-inclusive kit is lacking. I thoroughly enjoyed building it though and i've loved getting it printing. I get pretty nice quality but it is s.l.o.w. and it rattles itself apart faster than I can buy new loctite.
The prismx will definitely be my next printer. I'm all in for being an early adopter. Perhaps I can come by the shop sometime soon and pick your brain and pick up a couple things for it if you're ok with that Free.
-Bill

Sure! It might be difficult to migrate the Prusa i3 without swapping significant hardware, but it'd be cool to analyze what you'd need.

Forkers are very welcome. R. Feynman once said, "What I cannot build, I cannot understand.". Therefore, get started building!

Well, I got a couple of the frame pieces printed before my el cheapo psu gave up the ghost. I think I'm going to build the gantry frame first and then work on the axes. Where do you recommend for picking up the extrusion? Is there any place better than just getting them from Misumi? Once I have the replacement psu I will finish printing the printed pieces. I hope to pick up an 00str00der pretty soon and an extra NEMA 17 (since my z axis runs on two smaller steppers). Then go from there.. I'm excited, this machine seems to have the best vitamins and axis support structure of any reprap I've seen (i.e. thrust bearings and high quality threadscrews)

In my country (The Netherlands) it's not that expensive to get a set of extrusions, the bigger cost will be the cutting cost, so for me it's cheaper to just order a few meters of extrusions and cut the parts myself. Though it is less precise and I have more metal scrapes on the edges when I do it myself.

One kit of extrusions from misumi for the mendelmax cost about 54 euros, and the extrusions that you buy per meter are not branded, but much cheaper, about 5 euros for a meter of 20x20.