1.8 deg/step vs 0.9 deg/step - noticable difference?

Is there a noticable difference in accuracy/performance when using 0.9 deg/step motors vs 1.8 deg/step?

Also, how much backlash exists in these belt-driven gantry setups? Would making a larger belt-driven gantry (say, a 48 inch belt per axis) introduce any more backlash?

Thanks!

No, None, and No.

For very long belts, they tend to vibrate unless you make them thicker and under more tension, or run them very slowly.

---

www.Fablicator.com

Nothing noticeable, but technically the .9 are better because of the nuances of micro-stepping.

A very quick google:
[www.divms.uiowa.edu]

... the inaccuracy of stepper motors is defined by something like some deg's of a step - so with double count of steps your mechanical accuracy should be doubled with 0.9deg-steppers compared to 1.8deg's.

But then you have the inaccuracy of your pulleys, belts and mechanical stage, what's dekades more than this, so not very interesting for common repraps ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

A .9deg stepper at 1/8 microstepping is has more accurate holding resolution and has more torque than a 1.8deg at 1/16 microstepping. The problem is that a .9deg stepper needs to count twice as many steps to move the same amount on 1/16 microstepping, this will quickly eat up your tiny micro controller's memory and introduce pauses in the print. So you are limited on running them with 1/8 microstepping until ARM micros come out and are more common.

As far as lash goes, you would not want to increase belt length without getting a heavier grade belt, as it will stretch and induce more lash.

Primary lash will depend on type of belt and pully used, in T5 and T2.5 you will see more lash than MXL or GT2 belting due to tooth profiles. Tensile cords will determine how much stretch per foot the belt will allow. T5 and T2.5 were made to synchronize shafts and not do dual direction linear motion and will have back lash as part of the design.

If you used a pulley system, that is the only concievable way to increase belt length without introducing more lash.

Quote

T5 and T2.5 were made to synchronize shafts and not do dual direction linear motion and will have back lash as part of the design.

I think this is a myth. You can get T5 and T2.5 pulleys that have no backlash or you can get ones with some clearance. I have bought from Farnell and Beltingonline and neither have any backlash. I got some on eBay from Hong Kong and they had loads.

---

[www.hydraraptor.blogspot.com]

Nophead, it's not a myth if it is listed in the engineering design specifications. Maybe you got some with different tolerances than typical T5 or T2.5. In the end though it is a trapezoidal profile tooth, and a round tooth will give less slip.

Well I have read somewhere that it depends on the pulley spec. I.e. you can get ones with clearance for high speed transmission and ones without for no backlash, and as I say none of the ones I have bought in the UK have backlash and neither do my printed ones.

MXL belt is also trapezoidal. I thought it was just the imperial equivalent of T5?

---

[www.hydraraptor.blogspot.com]

BTW,
Increasing microstep resolution does not affect torque, that definitely is a myth. With sine and cosine drive the torque and power is constant in all positions.

---

[www.hydraraptor.blogspot.com]

nophead, i respect you but everything i have read says otherwise.

First off T5 is 5mm pitch, MXL is a bit over 2mm pitch.

Second, the tooth profile maybe trapazoidal, but they are deeper with more rounded corners and really grab industrial pullies.

Third, they have been used by me in the past in various similar industrial machines, but for vision inspection equipment.

Quote about typical MXL belting design and construction:
"The MXL belt is a classical synchronous belt with a pitch of 0.08" (2.032 mm). It is recommended for applications where maximum synchronisation, small package and high speed are required. Space-saving and highly stable, this belt is the ideal solution to precision drives such as office machines and printers.

Construction

• Trapezoidal tooth form.
• Elastomeric backing and teeth combine durability and light weight.
• Nylon facing protects and reinforces the tooth surfaces.
• Fibreglass cords provide length stability and flexibility.

Advantages

• Power transmission of up to 0.8 kW and speeds of up to 20000 rpm.
• MXL belts allow small pulley diameters (from 6 mm diameter) with a
maximum number of teeth in mesh.
• Highly suitable for stepper motors.
• Accurate positioning.
• Very stable."

About microstepping please read:
READ HERE

To each their own on these subjects i guess, but it makes since you are basically oscillating the rotor between strong magnets at a funny angle and trying to hold it with a chopping of power.

Edited 1 time(s). Last edit at 05/26/2012 09:51AM by WildBill.

I have read that microstepping argument before but it is totally misleading. People keep quoting it to me, saying microstepping reduces torque, but it actually says INCREMENTAL torque, which is entirely different.

A stepper motor only provides torque when the rotor is displaced from where it wants to be, or stated another way applying torque causes the rotor to displace. If you have a sinusoidal drive, it makes no difference where the motor stops, on a 16th step, an 8th or even a full step, the torque displacement curve is always the same. Its holding torque is always the same, and the total power dissipation is the same because sin^2 + cos^2 = 1.

What the article states is the incremental torque caused by applying one 16th step is less than applying one 8th step, which is of course true. It doesn't mean the motor has any less ability to drive a load or that the rotor will be displaced any more by a given torque. Why would it? If you consider the full step position (two coils on 70%) and the position 1/16th further on, all that has happened is the currents have both changed slightly. Why would the torque suddenly be dramatically less and then when it goes another 1/16th so that it is on an 1/8th step position why would it get more torque again? Taken to its logical conclusion a pure analogue sinusoidal drive would have no torque at all.

The only valid point the article does make is that if you try to move a very small amount stiction may prevent it moving at all. You could have a situation where 1/8th step would move but 1/16th step would stick and stay put. But when you put in the second 1/16th step it would move as it is now on the 8th step position. So you haven't lost any torque, or accuracy but you may not have gained the resolution you expected to gain. The article is wrong when it says you lose accuracy. That would mean some of the 1/16th step positions where not in between the adjacent 1/8th step positions, which would mean the rotor sometimes move the wrong direction.

On belts all I know is my personal observation that my T5 and T2.5 belts and pulleys do not have backlash because there is no clearance between the teeth.

Edited 1 time(s). Last edit at 05/26/2012 11:40AM by nophead.

---

[www.hydraraptor.blogspot.com]

Wild Bill, I use 0.9 steppers and 14 tooth XML pulleys on X and Y and E on Huxley seedling using Sprinter or Marlin firmware on a standard 16mhz CPU.

I'm intrigued as to why I would need to use an ARM CPU as both versions of Firmware running @16mhz on Huxley seedling easily exceed the maximum step rate for the stepper motors at 16x stepping at fast positioning rates jog & fast moves as extruding is much slower anyway.

I moved to XML belt after noticing that Lazer cutters use XML belt.
The Laser cutters move exceedingly fast with high accuracy they also use 0.9 steppers I've not seen an ARM CPU yet in a Laser cutter either.

As to XML over T5 or T2.5 I find the T5 slightly stiff so it requires a higher tension on the belt and presumably more torque to move over XML. I also have not seen any backlash on T5 belt as my Z axis on Huxley seedling is driven using T5 belt.

I have used both T5 and XML the profile is identical on both as I laser cut BIQ belt pegs for both.

Wildbill may be getting confused with HTD belts that do have a round profile.

Edited to add z Axis drive belt being 4mm wide T5 belt.

Edited 1 time(s). Last edit at 06/15/2012 07:54PM by BodgeIt.

---

Bodge It [reprap.org]
=======================================

---

BIQ Sanguinololu SD LCD board	BIQ Stepcon	BIQ Opto Endstop
BIQ Heater Block PCB	BIQ Extruder Peek clamp replacement	BIQ Huxley Seedling
BIQ Sanguinololu mounting	BIQ standalone Sanguinololu or Ramps mounting	Print It Stick It Cut it

My rep strap: [repstrapbertha.blogspot.com]

Buy the bits from B&Q pipestrap [diyrepstrap.blogspot.com]
How to Build a Darwin without any Rep Rap Parts [repstrapdarwin.blogspot.com]
Web Site [www.takeaway3dtech.com]

While I buy Nophead's argument about torque, I do have one question that no one has seemed to answer. If you combine 0.9 degree drives, with 16x micro-stepping, and small pulleys ( e.g. T2, 16 tooth) is it possible that you can outstrip the ability of Marlin (or Sprinter or your favorite code) to generate drive pulses? Seems that we are between 20K and 50K pulses/second, and I am not sure what the max is for the software. Of course one can slow down feed rates, but then that is self-defeating. I would see this as a reason to reduce to 1/8 step from 1/16, or am I wrong?

Bennett

I don't think there is an issue with pulse frequency. The problem is you will overflow the memory with too many steps on long moves.

I am running 0.9 degree motors with 1/8th stepping on my Prusa and it works just fine. I think on the Z-axis I'm actually using 1/4 steps because the lead on the all-thread already provides quite a bit of mechanical gearing.

Because of the limit on the counter memory size, you really end up having to compare these two options:
1/8 stepping + 0.9 degree motors vs. 1/16 stepping on 1.8 degree motors

I like the 0.9 degree motors, but they're slightly more expensive, harder to find and I'm not sure there's really any resulting difference at all between them when it comes to print quality.

I just looked up what type variables hold the number of steps to move in marlin firmware in the file stepper.h.
unless i'm terribly mistaken, it would take a very big printer to overflow the size on a long datatype
back of the envelope calc for size:
using only the positive values of long(2,147,483,647)
steps per MM for typical printed gear ~80 x2 because of .9 degree stepper instead of 1.8 degree= 160 steps per MM
2,147,483,647 maximum steps before overflow / ~160 steps per mm get you around 6.7 kilometers in the X/Y axis.
since my printer uses 630 steps per MM for Z...
2,147,483,647/(630*2)/1000(to get meters/1000 to get kilmeters -- that's about 1.7 kilometers

unless I bleeped up the math, I think I'll have no problems using a .9 degree stepper with 1/16th microstepping.

Olestra Wrote:
-------------------------------------------------------
> unless I bleeped up the math, I think I'll have no
> problems using a .9 degree stepper with 1/16th
> microstepping.

There's another factor and that is the maximum step rate. The step pulses are generated by the microcontroller one by one and while I'm not sure about the maximum rate, I think its something like 350 mm/s for 1.8 degree stepper with 1/16 microstepping (with the usual gearing). Using a .9 degree stepper halves the maximum speed while, in return, gaining some completely useless theoretic resolution.

While microstepping does not reduce torque, many of its claimed advantages are lies.

Using 1/16 microstepping will not give you 16 times more accuracy. You can maintain reasonably accuracy with half stepping, but anything beyond that is very unreliable. The only material advantage of microstepping is improved smoothness (note that DSP/advanced drivers will already perform smooth commutation so smoothness is only an advantage for the super cheap drivers).

I would much, much rather use 0.9 degree motors at 1/4 stepping than a 1.8 degree motor at 1/8.

Edited 1 time(s). Last edit at 11/03/2012 08:51PM by 691175002.

i have to differ, i built a two axis laser pointer with two nema 8 stepper and a worm gear reduction and there's a big difference between 1/2 and 1/16 microstepping, at least under little load.

I think the memory problem is diffent than you think. I believe that for a move from point A to point B in all four axis requires two bits for each axis or one 8 bit word per step. This implies that a circular move say 6" in diameter with 1/16 microstep needs 16 X 200 steps x 20 pitch x 3.14 x 6 = 1,205,760 bytes for that move.