Half step vs quarter steep (microstep)

[slavkok]

Hello...
Can someone tell me how big diference is in quarter and half step drive. I know that quarter should minimize resonance problem but in other side they require double the pulse speed rate. I use the L6207 H bridge and AVR controller for phase sequence decoder and don't know what sequence to get...

Thanks.

[jeffs555]

Quarter step will be smoother because the steps are smaller, and also because the steps normally have equal torque. With the basic half step mode the torque is not the same on each step and this can lead to vibration or resonance problems. There are ways of doing half step that have equal torque on each step and if your hardware will handle micro-stepping you should be able to do one of these. http://techref.massmind.org/techref/...p/halfstep.htm

Since results will vary depending on the steppers and the machine, the only way to know would be to try both half and quarter.

[slavkok]

Hmm Maybe that is the catch. I'm already tested on the bench (just motor with with different flywhells as load) and can't see the differences. I'm already do that "torque compensated half steep. As L6207 have poor response if phase are switched off I shift all switch points for 1/2 (u)step to the right so I newer get the phase with zero current. So seems that I should finish my machine and do the live test what is better. Anyway it's only matter of changing software in AVR micro and adding two resistors.

There as attached phase sequence's tried at me. The quarter steep seems to not significant improve performance, but double steep pulse rate. As TurboCNC on 486/100 laptop can go to some 11000 steps/second the quarter step is beyond this system (at 15000 doesn't work but other computer go to 18000steps)

So I have more job to do...

[pminmo]

There is half step and then there is half step...:-) Not all half steps are created equal, and you are probably seeing that. Half step that utilizes the pole points, ie one phase on one phase off and both phases on 100% will be different than half step that utilizes 70.7% for both phases on. Then you have halfstep that doesn't use pole points like your 100% phase - 41% phase which will give you different results yet.

[Mariss Freimanis]

I'd like to add a couple of comments:

Let's assume a modern, square motor is being used, a motor whose torque is the vector sum of its winding currents. The goal is to have torque be independent of the angular location of the motor. Only the sine and cosine function satisfies this requirement by having a constant vector sum.

From that viewpoint, a 2 microstep sequence is 0,1 .707,.707 and 1,0 which gives a vector sum of '1' in each instance and a torque ripple of 0%. A half-step sequence is 0,1 and 1,1 having vector sums of '1' and '1.414' respectively and a torque ripple of 41%.

Unloaded, both motors will behave the same but apply a constant torque load and the half-stepped motor will be measurably worse in step placement accuracy. The 2-microstep motor will keep its even 0.9 degree step spacing while the half-stepped motor will have a large-step / small-step response. There is only a quantitative difference between a 2-microstep drive and a higher resolution microstep drive.

There are two reasons for microstepping: Low-speed resonance suppression and increasing the resolution of the motor.

1) Low-speed resonance: A step motor is a mass-spring system. As such, it has a mechanical resonant frequency and it behaves like a low-pass filter. A playground swing is a good analogy; it too has a resonant frequency and behaves like a low-pass filter. You can push the swing every cycle, every other cycle or even every 10th cycle and it will swing (resonate) but you sure can't push it more often because then the push 'frequency' would be above the low-pass cutoff frequency. It just would swing.

A step motor's resonant frequency is between 30Hz and 120Hz depending on the motor's size. If it were say 50Hz, (200 full steps per sec or 1RPS), the motor would resonate at 200 steps/sec, 100 step/s and 66.6steps/sec because it was "pumped" every cycle, every other cycle or every third cycle respectively. It would also become smooth for all speeds above 200 steps/sec because those frequencies would be above the low-pass filter cutoff frequency.

Microstepping helps in one way because the microstep "pushes" come at a more rapid rate for the same speed. A 10-microstep drive wouldn't resonate the motor above a 20 full-step rate; any speed above that would be above the motor's cutoff frequency. The second way microstepping helps is less dynamic energy invested per step. When you full-step, the motor takes that 1.8 degree step in a few milliseconds and then 'rings' like a bell around the new step location. It built up a dynamic velocity on its way to the new location that now must be dissipated. Using 10 microsteps, the distance to travel (0.18 degrees) is 1/10th as far and 1/10th the dynamic velocity is generated getting there. Energy invested is mass times velocity squared (E = (mv^2)/2) so only 1/100th of the energy is present to resonate the motor compared to full-stepped motor. That's 99% of the way there to eliminating resonance using 10 microsteps. It takes an infinite microstep drive to eliminate the remaining 1%.:-) These two effects (frequency above low-pass filter cutoff and low residual energy), join to make low speed resonance a complete non-issue by the time a 10-microstep drive is used. Anything above that just chases the remaining 1% at the considerable of cost requiring higher step pulse frequencies for the same speed. Caveat: For this to work, it requires a motor with good linearity (vector sum of the currents truly is equal to torque) and a drive that produces very accurate sine and cosine phase currents.

2) Improving resolution: Most step motors sport a "+/-5% non-accumulative tolerance" specification. What this means is you take a full step of 1.8 degrees, it will be be that give or take 0.09 degrees and this error doesn't accumulate; you will be exactly back to where you started from after 1 full revolution. That is the definition of an eccentric error like the wobble of an out of round shaft.

+/-0.09 degrees has an error uncertainty range of 0.18 degrees. 2,000 of those fit in a circle so you can say a step motor has an ultimate accuracy of 1/2,000th of a full revolution. Yet when you full-step a motor, you can only stop at 200 radial locations. What's wrong with this picture? Your accuracy is 10-times better than your resolution so you are getting only 10% of what the motor has to offer. A 10-microstep drive harmonizes accuracy and resolution; beyond 10 microsteps you only get empty resolution because it exceeds accuracy and there is little to be gained with that.

This pretty much sums it up. Microstepping brings two advantages to the table as compensation for what it costs, resonance suppression and increased resolution. If you are going to design a microstep drive instead of a half-step drive, don't stop at 2 microsteps. That is a poor utilization of resources once you are committed to microstepping. Go with at least 10 microsteps.

Sorry for the long post.

Mariss

[slavkok]

Why sorry for long post?

If post have some real things they are newer to long.
As I have TurboCNC as Gcode interpreter I'm limited with steep rate. The steep resolution is small even at full steep so I doing that only to reduce nasty resonance.

Slavko.

[pminmo]

Sorry for the long post.

Mariss

Teaching, based on facts and science is always a good thing!