Whats Wrong with the Prusa i3?

I am in the early stages of designing my own RepRap. What problems are you having with the Prusa? What features would you like to see?

I have some initial ideas, but I want to let this post soak a bit before I begin to taint your ideas with my own. I will say now, that it will be a Prusa like design, it will also be entirly free and open source.

If I had to say as a beginner, the spool of filament being straight above instead of left and above would help I believe.

The mounting point of the y-belt in the center is bad chosen for a three-bearing design.
Leave it in the center when you use four bearings or move it to the dual bearing side, when using three bearings.

Use a three point bed leveling design. Make the y-bed design as light as possible. ( y-frog made of Dibond (R) or carbon )

Use a geared direct drive design for the extruder with a NEMA11 stepper. ( light but without Bowden tube )

Good luck with your build
-Olaf

For stability and speed a corexy design seems to be better than the prusa i3 design, though I've managed to work out most of the many issues. There are for me a few key improvements:
- a rock solid print bed and y carriage, easy to square or rather self squaring with the frame, it should be so level you don't need to manually level it. No springs etc. 4 large linear bearings instead of smaller 3. Harder to setup bed much more solid.
- auto bed leveling for the last couple of 0.1mm
- single z motor design, no more synchronizing two motors
- thermo couple and full metal hot end, e3d v6
- solid belt tensioning system, I've broken more than a few thingiverse tensioners before I designed my own.
- 24V heated bed, hotend optional. Fast heated bed heating and reaching 110 degrees easily
- bowden system, light compact hotend design for fast operation
- include design for an enclosure for abs, hips and nylon printing.
- use of high quality components like belts electronics and bearings for long life use, low mainantance.
- z rods using acme rods, or use anti wobble with anti backlash.


Considerations:
- 2020 slot extrusions for easy and parametric assembly, allowing user to easily scale the build volume
- easy dual extruder conversion
- easy bowden/direct drive conversion
- modular tool head design with general purpose connector, easy to swap hotend, dual hot end, pen/pencil and spindle. Maybe other options like a syringe tool head etc.
- I like ramps 1.4 as it's cheap, easy to replace and expand. Ymmv, it's not very tidy connector and cable wise and could use a few extras. A ramps 1.5 with redesigned connector positioning, more mosfets and stepsticks, including 2 thermo couple converters so that a thermo couple can be connected directly. On my dual extruder i3 with part cooling fan I now use 3 external little boards.
- the use of rc models silicone flexible high current wiring, for heated bed.

Looking forward to your ideas!

Edited 1 time(s). Last edit at 10/26/2015 04:53AM by imqqmi.

Hi,

I'm a Prusa i3 owner for two years.

Which prusa i3 ? There's so many versions and builds of that initial design !
Everything's wrong with most of the cheap kits, from acrylic frame to budget extruders.
Even if Jo Prusa himself does not consider most of rework parts,
some improvements was great improvements to the original design :
- ACME 8mm screws (or preferabily threaded nema17 shafts) on the Z axis
- Belt tensioning system on X and Y axis
- Auto bed shifting
- Adjustable endstops mounted on the frame or plastic parts instead of on the rods
- 10mm smooth guides or Vslots (Markerfarm's I3V)
- Plates to replace Ychassis small threaded roads
- No Z wobble parts
that's all for the major improvements.

What I still dislike about the Prusa I3.

First of all, the two motor Z axis. Ok, two independant screws helps for the setup.
You can turn one screw to aling to the other. But most of the time that creates missalignment,
because for any reason, one motor can shift from the other.
The initial embeded nut and threaded rods is not a proper transmission system,
you got an important wear at the start, growing and growing.
A geared belt system is a durable transmission with no possible missaligment, that uses a single motor.

Seccond, threaded rods for the Y chassis. It's a wrong good idea.
The threaded rods are not stiff enough, and it's too easy to get out of square.
Through, that two points attachement to the frame makes a pivot point on the Y direction.
There's many ways to make a clip-in or tubing chassis to get a squared strong chassis.
Otherhand, if you need to change anything attached to that threaded rod chassis,
you need to take it apparts entirely and spend a lot of time trying to square it up.

The frame itself could be better. The legs lay on the back side only,
counting on the Y chassis attachement to keep the machine square on the front side.
The opening in the main frame does not allow to use the both sides of the Xcarriage,
which can be used to attach fans, motors or a connexion board for example,
and balance weight instead to have a huge overhang on one side.

The Y carriage support the bed with 4 screw.
But 3 leveling points are better, and you cannot modify the genuine carriage for this.
For a 200x200mm bed with a glass plate, weight is not a factor,
but actualy the best print surface is a stiff aluminium plate.
The 8mm guides are maybe not enough for that extra weight.
Through, if you want a bigger printer, the moving bed will be a factor,
for the weight as for the acceleration. A 300x300mm bed is over 2 times the weight.

Beside some holes in the frame, there's no consideration about cable management.
Some rework versions uses cable carriers which are not an interesting solution IMHO.
I don't agree with all the choices in the Mendel90 (DB9 connectors), but for cable management
it's a good start point to think about it.
Maybe something is to consider about IDE flat cables and connectors, or ATX molex ?

The I3 is engineered for direct drive single extruder. You don't know wherer to fix you bowden(s).
There's no extra room to use any dual hotends without loosing a part of your print area.
And dual hotends on the Y direction will cause quality issues becaus of the large overhang.

Overall, the Prusa I3 is meant to be fast and easy to build with parts you can find at the local store.
But that choices makes the I3 a hard to maintain experience. Move it from one side of your desk to the other,
and you will need some time to square it up again.
The time spent to endlessly improve or square-up the printer is truely wasted time.
The huge number of rework parts talks. And that's not almost because of the I3 success story.
There's a lot of things you want to rework on a Prusa I3.

++JM

Edit : oops, I forget an important point : printed idler pulleys (or bearings instead of idler pulleys).
I get better results and improvement of print quality while using a regular pulley on a shaft within two 625 2RS bearings.

Edited 1 time(s). Last edit at 10/26/2015 10:41AM by J-Max.

You guys are awesome. Keep the complaints coming!

My i3 frame can be moved around with no change in alignment.

Hi guys,

True RRuser, but don't forget to say it's not a genuine I3 but an i3 PSIQUE.
It don't have the threaded rod chassis, but one made out of steel plates.
That's not exactly the same stuff.

++JM

you should design one with built in belt tensioners and if not a proximity sensor for the z-stop, maybe put the z-stop on the top, easier calibration, imo.

The biggest fault IMHO of the design is the lack of ACCURATE alignment of the two polished rods on which the bearings ride. The lack of understanding this importance, I think, has led to many problems with bearing life, overheated motors, and binding in general. Most builders just ASSume that the two printed rod holders for the X axis are exactly the same or that the location of the Y rods is perfect. Using 4 bearings instead of three makes the misalignment even worse. You need a way to either accurately align the rods parallel or allow one set of bearings to float in one axis to ride along with any misalignment.

Sorry for the really late response, but here's pretty much what im hearing from you.

1. Spool holder
2. consider mounting point of Y belt on sled
3. 3 point bed leveling
4. geared drive with light stepper
5. inherently square/level geometry
6. auto-leveling for the last .1mm
7. single Z motor
8. full metal hot-end
9. solid belt tensioner
10. 24V system
11. lighter extruder (bowden)
12. full enclosure for printing engineering plastics
13. use acme rods, and backlash compensation
14. easy to scale build volume
15. Easy extruder conversion
16. built in cooling fan
17. use RAMPS
18. better electrical connectors
19. stiff frame
20. better end-stop system
21. better Z movement system
22. leave room for accessories on the other side of the X carriage.
23. Make sure that Y axis is stiff enough
24. design in wire-management
25. include solder-less wire connectors


Like I said earlier in the thread, I was planning on designing my own prusa spin off (eventually), here's what i'm thinking:

1. its still going to look, feel, and act like a prusa i3
2. a laser cut steel frame, that is inherently square, and level.
3. built in cable guides
4. All solder-less connectors, with different types for different applications in order to seperate them better.
5. M10 guide rods for Y axis, M8 for X and Z. (note, this might change, I haven't run any calculations yet to see if these will be stiff enough)
6. replace linear bearing with tight fitting bushings
7. 200mmx400mm print bed, two of the standard 200mmx200mm plates with a 200x400 piece of glass.
8. replace the X and Y belts with M8 with 8mm lead acme screws
9. replace the Z axis threaded rods with M8 with 2mm lead acme screws
10. some sort geared extruder


My intent is to make a printer that is inherently more reliable and accurate than the traditional Prusa i3. I am aware that the larger print-bed will greatly slow down print-speed, but to me the trade-off is worth it.

Soldering will always be more reliable than connectors.

Quote
AudaciousTuddle
Sorry for the really late response, but here's pretty much what im hearing from you.

1. Spool holder
2. consider mounting point of Y belt on sled
3. 3 point bed leveling
4. geared drive with light stepper
5. inherently square/level geometry
6. auto-leveling for the last .1mm
7. single Z motor
8. full metal hot-end
9. solid belt tensioner
10. 24V system
11. lighter extruder (bowden)
12. full enclosure for printing engineering plastics
13. use acme rods, and backlash compensation
14. easy to scale build volume
15. Easy extruder conversion
16. built in cooling fan
17. use RAMPS
18. better electrical connectors
19. stiff frame
20. better end-stop system
21. better Z movement system
22. leave room for accessories on the other side of the X carriage.
23. Make sure that Y axis is stiff enough
24. design in wire-management
25. include solder-less wire connectors


Like I said earlier in the thread, I was planning on designing my own prusa spin off (eventually), here's what i'm thinking:

1. its still going to look, feel, and act like a prusa i3
2. a laser cut steel frame, that is inherently square, and level.
3. built in cable guides
4. All solder-less connectors, with different types for different applications in order to seperate them better.
5. M10 guide rods for Y axis, M8 for X and Z. (note, this might change, I haven't run any calculations yet to see if these will be stiff enough)
6. replace linear bearing with tight fitting bushings
7. 200mmx400mm print bed, two of the standard 200mmx200mm plates with a 200x400 piece of glass.
8. replace the X and Y belts with M8 with 8mm lead acme screws
9. replace the Z axis threaded rods with M8 with 2mm lead acme screws
10. some sort geared extruder


My intent is to make a printer that is inherently more reliable and accurate than the traditional Prusa i3. I am aware that the larger print-bed will greatly slow down print-speed, but to me the trade-off is worth it.

If you are going to have Acme screws for X and Y axes, then you better have a backlash removal mechanism that works.

And don't forget alignment options for the X and Y rods and/or bearing holders.

@ RRuser: i agree that soldered connections are more electrically efficent, but not having them has given me major headaches when im doing maintence and repair, and has even caused me to damage my machine (IE breaking thermisitors).

@tjnamtiw: I wouldn't dream of neglecting backlash dampening/removal, though i will still need to do some research/experimentation into the best way of doing so. Traditionally you use a nut-spring-nut sandwhich, but there might be a way that suits my needs better. And to be really honest i haven't done the math yet on what it will do to my resolution, so the lead-screw thing might change. I'm waiting until after exams until i get too deep into the design process.

I was hoping that I could get away with no alignment mechinisms at all, but have the alignment be inherent in the frame, but again i only have a rough plan, the heavier design will start coming in acouple weeks, and will probably take me months and months.

Interesting thread as most of the upgrades/changes mentioned I've done them or will do them soon.

So far I've moved to M8 leadscrews for the Z and it made quite a change, not only in precision as well as noise.
And changed all linear bearings to iGus bushings.

The next is X and Y, I got also M8 leadscrews and have got the nuts with built-in anti-backlash system.
Now just need to design the parts.

Hi guys,

To me leadscrews on the X and Y axis is not a good idea.
You will get a very slow machine with low acceleration.
Through, you won't get any significant precision.

Skarab designed the Lautr3k several years ago. It's a full leadscrew machine.
I heard someone these days who told me Skarab was not realy happy with his leadscrews system.
Next time I can talk with him I will ask about the exact reasons.

Belts are quick, precise, and allows fast accelerations and direction changes.
That's all that we want on a 3D printer.
It seems more interesting to me to work on better idler pulleys and tensioning system.

++JM

That is a really good point on the leadscrews. Typical belt steps/mm run around 80 steps/mm ish. Assuming 1/16th microstepping, a motor with 200 steps/mm, and a leadscrew with a lead of 8mm, you get 400 steps/mm. Assuming that the inertia of the rod, and the difference in mechanical advantage doesn't make a difference, you end up with a printer that is potentially 1/5th the speed. (I hadn't run these numbers until just now.) Assuming that the lead screws are perfect and that there isn't any backlash, the lead screw version will be 5x more precise. So you're probably right there.

I do have a reason that i want to test them still; they always work. I haven't had any problems with the belts in my printer yet, but from other applications I know that they stretch, and break and wobble and get loose. Acme rods are solid, and stay solid. As I have said before my goals for this project is to make a printer that just works, not a fast one. The other reason I want to try them is because I want to use a really large print bed, and that 5x mechanical advantage sounds like a good thing to me.

Edit: I looked up the machine you talked about, it has some similar ideas and some different. The biggest difference is that he used 25mm lead screws and i want to use the significantly cheaper 8mm leads. His would be faster, mine would be would be more precise. I haven't done any math yet but I assume that he would have torque issues, as the eficency of leadscrews decreases as the lead increases.

Edit #2: I just sent Scarab a PM, hopefully he will get back to me with some good info about his experience.

Edited 2 time(s). Last edit at 12/08/2015 05:59PM by AudaciousTuddle.