PDA

View Full Version : microstep uniformity



leadinglights
27-04-2010, 11:40 AM
How uniform are the steps at various microstep sizes on a stepping motor? For that matter, what is the accuracy of the steps and microsteps?

The reason that I ask is that similar techniques used in ADCs for example, result in some steps disappearing at the smallest sizes, while others are over large - this is called differential non-linearity.

Thanks,

Mike.

irving2008
27-04-2010, 01:04 PM
On a typical stepper the linearity of the main 1.8deg steps is about 1.5%. Halfstepping is about the same. Once you get into microstepping below that then its down to the driver and the load. A typical 8bit driver will give a further 1 - 1.5% non-linearity on an unloaded motor but the load will determine the accuracy to a great extent as the magnetic forces act as a spring against the load. Microstepping (other than halfstep) should never be used to improve resolution only smoothness of motion.

leadinglights
27-04-2010, 09:56 PM
Thanks again Irving. As I willl be using ballscrews with a 5mm pitch, I think that halfstepping at 0.0125mm per half step is perhaps a bit lumpy so I am now looking at using a toothed belt drive for the ballscrews.

Mike.

Robin Hewitt
28-04-2010, 12:16 PM
5mm pitch, I think that halfstepping at 0.0125mm per half step is perhaps a bit lumpy so I am now looking at using a toothed belt drive for the ballscrews.

Hi Mike

I use a 4:5 tooth belt simply to drop the steppers below the top of the table and stop faffing about with motor alignment.

4:5 effectively reduces the 5mm pitch to a 4mm pitch so 400 steps/rev = 0.01mm. As the machine weighs less than a ton this is hardly achievable because it will bend under the tool pressure, but I still microstep to smooth it out a bit :naughty:

Don't skimp on the pulley size, I use 32:40XL

Robin

HiltonSteve
28-04-2010, 12:45 PM
Thanks again Irving. As I willl be using ballscrews with a 5mm pitch, I think that halfstepping at 0.0125mm per half step is perhaps a bit lumpy so I am now looking at using a toothed belt drive for the ballscrews.

Mike.

I have 16mm x 5mm pitch ballscrews on x,y and z on my machine running at 1/4 step, runs very nice and not noticed any lumpyness. Did try them at 1/8th and 1/16th step just to see how they ran but did not seem to make too much difference.

Mine are direct drive but belt driving them will give you more mounting options so definitely worth thinking about.

Just my two-penneth..!

leadinglights
28-04-2010, 01:55 PM
I was probably spoiled by the first CNC that I built - based on a Proxxon MF70, it worked it's little heart out to give almost unmeasurable errors, really smooth curves etc.. I would like to get a step resolution of 0.0025mm but would settle on 0.005mm. I am thinking of 2.5:1 (1000 half-steps per leadscrew revolution) with 16T to 40T Syncroflex 2.5mm pitch pulleys and 6mm wide belt - would that work with a 3Nm stepper motor?

HiltonSteve
28-04-2010, 02:13 PM
Probably don't have to tell you but remember belt driving at 2.5:1 will increase your resolution but will reduce your max feed rates.

My machine uses 3Nm steppers and I am driving them at 36V 5A, absolute maximum feed I can get without losing steps is 7500mm/min but have reduced it to 5,000mm/min so that I am not pushing it to the limit all the time and this is driving the screws direct without any belt reduction.

So 5,000mm/min divided by 2.5 will only give you 2,000. Does depend on voltage's, driver type etc but I would say that 2,000mm/min may not be enough.

leadinglights
28-04-2010, 02:54 PM
2000mm/min sounds like the speed of sound to me, my Proxxon manages about 160mm/min - going by Youtube videos of similar machines this is average to good for this type of conversion.

HiltonSteve
28-04-2010, 11:31 PM
Did I read that correctly....

Your maximum feedrate was 160mm/min????

That is seriously slow, are you sure about that?

John S
28-04-2010, 11:38 PM
Bloody hell you could eat a packet of Hob Nobs in that time..............

leadinglights
29-04-2010, 09:56 AM
How big a packet of Hob Nobs?

The Proxxon is limited to about 160mm by 40mm movement in X and Y. As the 1Nm steppers are only driven by 13.5V it could probably be sped up a bit, but a typical job might be a 25mm dia brass gear, or maybe a 80mm by 30mm PCB.

Robin Hewitt
29-04-2010, 11:37 AM
2000mm/min sounds like the speed of sound to me

That would be over 6rps on my lead screw handles :eek:

I can ramp it up to incredibe speeds but start to get uneasy at around 3rps.

Guess I'm not very brave :whistling:

leadinglights
05-05-2010, 10:26 PM
I have just been playing with the numbers on toothed belt drives and stepper motors but I don't know if the answer I have come up with are right.
Using the formulae given in the Transdev datasheet for Syncroflex toothed belt drives, I get a figure for tooth shear strength when using a 10mm wide belt and pulleys of 16T and 40T, 6 teeth engaged, of just a tad under 200 Newtons at the stepper motor. The Tension Member is not much better at about 300 Newtons.
Although I haven't settled on a stepper motor yet, the most likely candidate, the so-called 3Nm NEMA 23s that are available at the moment, put out about 2Nm at stall - about 160 newtons at the periphery of a 16 tooth pulley.
Is toothed belt really that wimpy, or have I lost a decimal place somewhere?

Hmm, a quick edit here. Looking at the datasheet for Zapps SY60STH88-3008BF, the maximum radial load is given as 75 Newtons. I assume that if you are applying a torque of 2Nm at a 12.5mm radius then you will get something like 160 Newtons of radial force reacting to it. I may have to think a bit more about the motors as well.

Robin Hewitt
06-05-2010, 10:43 AM
Hmmm

2Nm torque, 5:2 reduction, 5mm pitch screw gives a driving force of 0.63 tons.

Probably best if the belt fails before something else does :naughty:

Robin

How to calculate it...
! rev = 5/2.5 = 2mm
2mm/(2*pi)= 0.3183mm = 0.0003183m torque radius
2Nm/0.0003183m = 6283N = 0.63 tons

leadinglights
06-05-2010, 11:02 AM
Hmm, I had better apply myself to Irving's tudorial and motor calcs program. I had avoided that until I get some idea of the masses that I will be throwing around.
(Also, avoid getting finger trapped between table and frame at 0.63 tons)

Robin Hewitt
06-05-2010, 12:51 PM
f=ma

So 6283 Newtons would accelerate a mass of 18.5kg to Mach 1 in approx 1 second.

Robin :heehee:

leadinglights
06-05-2010, 02:16 PM
But I will be using Mach3, does that take 3 seconds?

John S
06-05-2010, 02:25 PM
1.85638543287408 Hob Nobs [approx]

.

Robin Hewitt
06-05-2010, 02:59 PM
But I will be using Mach3, does that take 3 seconds?

Indubitably :beer: