Screw drive for X and Y axis

I have a fairly stock Prusa Longboat with (split) belt drives for the X and Y axis.

By my reckoning this gives me a step size of about 0.25mm.

Then there is the stretch in the belts to consider.


So.

I was thinking of changing the X and Y axis to use a simple screw thread (like on the Z axis) to increase accuracy.

If I use M6 threaded rod which has a pitch of 1mm then my step size becomes 0.005mm


I am currently printing at 40mm/sec on perimeters and 60 on fill, so that means the motors need to run at 2400rpm for the perimeter. For the infill I typically just do 45 degree so that means each motor has to run at 2545rpm...

So the question is... will the motor handle these speeds in practice? It seems to be rated to a maximum of 3000, but will it actually do this OK? Will it accelerate in a reasonable time (assuming I keep the weight minimal)..?

Maybe I need to go up to say M10 x 1.5 (but then the weight/inertia becomes a problem maybe) or maybe gear the motor up or look for a different motor to use?

Any other thoughts?

George

I don't think your step size is 0.25. Check you configuration.h how many steps you have per mm. Most have 40 or 80 steps per mm, which is 0,025 mm or 0.0125mm per step.

If you are more at 40 steps/mm you are running the stepper driver with 1/8th microsteps. Most can do 1/16th by changing some bridges.

Your idea with rods does not work as good as you might expect. If you do full steps you need 200 Steps/rotation or mm with M6. Your speed will be limited to something around 50mm/s because the firmware can not handle higher frequencies. For higher frequencies they use double and quadstepping which do not work good with full step mode.

Also M6 rods are not made for this. They will wear off quite fast. If you do it you need the good stuff from cnc machines.

If you want something more precise look for the belt doubler from mad kite and adopt it to your printer, if no prusa version is on thingiverse. That gives you the double accuracy (while reducing speed to 50%).

---

Repetier-Software - the home of Repetier-Host (Windows, Linux and Mac OS X) and Repetier-Firmware.
Repetier-Server - the solution to control your printer from everywhere.
Visit us on Facebook and Twitter!

Use fast acme screw and reduce your stepping. You should be able to closely match the feedrates you currently employ.

I am working on a design with some surplus Berg lead screws with a 0.375 lead. (sorry, jumped out of the metric. that would be, what, 9.525 mm?) 200 steps per revolution, gives me a step size of about .048mm? That can't be quite right. I have been going to bounce this off someone in the group, should I start my own topic, or is this an acceptable spot? currently planning on using standard 1.8 deg steppers. can i get comparable speed and accuracy to my standard prusa belt driven system?

The 200 Steps per revolution is in full step mode. With 1/16th Microsteps you have 1600 Steps/revolution. So you can choose how accurate or fast you want to be. If you go down to 1/8th stepping you can use quad stepping, so frequencies of 40000Hz should be possible. At least if the motor has enough force for all this. Not sure how frictions may reduce this all.

---

Repetier-Software - the home of Repetier-Host (Windows, Linux and Mac OS X) and Repetier-Firmware.
Repetier-Server - the solution to control your printer from everywhere.
Visit us on Facebook and Twitter!

The three issues with lead screw are.
Speed, mentioned here, addressable if you use screw with low TPI, though they are usually expensive.
Backlash, nuts are less that ideal in this respect, the usual solution is to use either matched delrin nuts or two nuts with aspring between them to add a preload.
Efficiency, conventional screws are enormously inefficient compared to belts, in terms of work in to work out, and the preloaded nuts make it worse, ballscrews don't have this issue, but they are expensive, and don't generally come in small sizes.

Having said all of that I have a small Sherline mill I converted to CNC years ago and it's still using the stock ACME screws @20TPI, it'll run at 60IPM, though it's probably not good for wear on the brass nut they use. To do that the motor has to run at about 1200RPM, so you'd almost certainly need a higher supply voltage that the 12V most repraps run.
For point of reference, the mill above is running with 270inoz nema23 motors with a 48V supply driven from a PC running EMC2 via a gecko G540.

You might be able to pull it off with a 3/8"-5 multi-start acme rod as xiando suggests.

[www.mcmaster.com]

But at $90 (no preload - $150 with the extra nuts) for X&Y axes that is an expensive proposition...

AND

a full turn of that screw will move a nut 25.4mm. So if you're using 1.8 degree motors (200 step) and 1/16 micro-stepping the step resolution should be around 0.008mm/step

Compare that to a 36T GT2 setup on same steppers/settings and you get ~ 0.007mm/step. And it's a fraction of the cost.

But I would be hesitant to use 1/16 microstepping on that rod, especially at high speeds. A tiny NEMA 17 might skip steps trying to overcome the inertia of the rod... Maybe xiando can speak to this.

But if money is no object experiment away!

Also, is it a 3/8" - 8, 10 or 12 rod that you have?

Edited 1 time(s). Last edit at 09/26/2012 03:11PM by akhlut.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Thanks for the replies.

Wasn't aware of the micro-stepping, I was just going off the motor specs.

I feel like I am getting pretty poor resolution right now. Looking at Configuration H I have 80 steps per unit, so I guess that means 0.0125mm per step.

My problem is that when I try to print a fine zig-zag for a sort of spline, the corners are getting very round and the definition is being lost. So if this isnt down to the step size, I guess it must be something related to either the stiffness of the drive members or the behaviour of the filament in the small gap between the nozzle and the part.

George

Rounded at corners is normal to some extent the filament is stretched as it comes out of the extruder, so it "shortcuts" corners very slightly making them "rounded" rather than sharp.
You can reduce the effect with a smaller diameter nozzle, since there is less stretching for a given extrusion volume.

What nozzle size do you have? Changing from 0.5mm nozzle to 0.4mm will give sharper corners. Also slowing down the print, or at least perimeter, speed will also help more sharply define the shape. Finally, check your extruder calibration - I use this guide in the Slic3r documentation [github.com] - this gave me significantly more consistent vertical walls, and sharper corners too.

There are issues with reprap firmware and micro pauses that cause it to be unsuitable for threaded rod systems, and even for acme drive systems. If you are going to have a threaded rod system, then use a cnc driver, and also run it with mach 3, or emc2. They are far superior with timings, and speeds required.

typically x and y coordinates are not moving at same speed, you will be limited to max speed you can have the faster axis run from stop to start and from start to stop.

Screw drives are unlikely to gain you higher resolution at the same speed. You'll need to significantly reduce speed in order to see a positioning accuracy improvement.

It's physics. Consider that many users tune their RepRaps to operate fairly close to the maximum speed their machines can operate without losing steps. Losing steps occurs when the force required to accelerate the print head exceeds the torque the motor can provide. The weight of the print head isn't going to change, but screws are fairly inefficent compared to belts. They also add some rotational inertial to the drive shaft. This means the motor needs to exert more input force to achieve the same output force at the extruder.

Also note that maximum torque in a stepper motor decreases as the speed increases. This is because a stepper motor consists of magnetic coils...also known as inductors...which resist changes in the flow of current. Chopper drives and higher voltage power supplies can counteract some of this, but there's a limit: the torque will decrease as the current pulse shape approaches a sawtooth rather than square pulse or sine.

The result is that, yes, screws can be used for higher positional accuracy, but you must expect an equally dramatic decrease in speed unless you make a corresponding upgrade in motor and controller specs.

Edited 1 time(s). Last edit at 09/26/2012 11:41PM by macegr.

FWIW ahklut, I neither specified a screw diameter or a thread pitch. Please refrain from stating that I suggested a specific solution when I did not do so. I merely suggested that fast lead screw and reducing the stepping will result in similar speeds with similar precision.

jamesdanielv Wrote:
-------------------------------------------------------
> There are issues with reprap firmware and micro
> pauses that cause it to be unsuitable for threaded
> rod systems, and even for acme drive systems.

What micro pauses?

I have a CNC mill conversion (with threaded rods) running with reprap firmware and have not seen any issues like that.

I was responding to Idaholion - I didn't make that clear. AFAIK fast-start and multi-start are the same thing. Maybe I am wrong. I know that acme screws have various pitches, and that multi-start leadscrews are faster than single-start. I was simply providing an example of what you suggested. I was suggesting that you might comment on what I put forward, as you seem to have the engineering knowledge that I do not.

Sorry for that.

xiando Wrote:
-------------------------------------------------------
> FWIW ahklut, I neither specified a screw diameter
> or a thread pitch. Please refrain from stating
> that I suggested a specific solution when I did
> not do so. I merely suggested that fast lead screw
> and reducing the stepping will result in similar
> speeds with similar precision.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

pauses during acceleration, partially from narrow pulse width (10us or less) when opto-isolators respond at about 150us or greater. pauses during serial communication making look ahead algorithms unobtainable.

if you have big enough motors, you can get it to work decently, but you will have issues with acceleration. if you can prove differently then show a large printed part, printed at 25-45mm/s on reprap hardware. Then i will reference your specs as the bare minimum for threaded acme systems, and firmware.