Stepper Drive Rotary Table Results
Bill Chennault
Posts: 1,198
All—
·
I have finished the first round of real testing of the stepper motor driven rotary table. Here is the setup (I am writing this as if no one is familiar with what I have posted before) . . .
·
I have a Grizzly·H5685 rotary table designed for a mill. I drive it with a NEMA 17 Jameco 155459 stepper motor via a 3:1 timing belt arrangement from SPD/SI. The stepper motor receives instructions from a Peter Norberg Consulting BS0710-USB stepper controller board. In turn, that board is controlled by a BS2px. (I started off using a BS2, but switched to the faster Parallax processor because I thought there was a baud rate problem; there was not. I could have easily stayed with the BS2.)
·
I built a mount for the rotary table and the stepper motor out of 0.25” aluminum. (Pictures, below.)
·
I knew nothing about stepper motors or controllers and one of my goals was to find out something about them. Now, I know a little more. That is good.
·
Another of my goals is to design a highly precise robot shoulder at a price a hobbyist MIGHT afford. This test was done to determine if a stepper motor would give me the repeatability necessary to meet the definition of “highly precise.”
·
I experimented driving the little Jameco stepper motor with the Peter Norberg Consulting stepper board until I found the parameters necessary to achieve what looked like the best for repeatability and reasonable speed. I discovered that using the BS0710-USB in “mode 0” (half power, full step, single winding) gave me the best results at the time. However, during initial testing, I came to the conclusion that all of my equipment is brand new. And, like a new car, needed to be broken in. This included the stepper, the timing belt and pulleys, and the rotary table. Therefore, I will retest the configuration using the other modes . . . half step, alternate single/double windings on – non constant torque; full step, double windings on, full power, full steps; and microstep, eight microsteps per full step, constant torque.
·
I ran the rotary table configuration as described above in mode 0 for well over 48 hours. At the end of this period, I used the magnetic clamp dial indicator that I had previously affixed to the rotary table to take 100 readings. The results given by the dial indicator, which is marked in 0.0001” (1/10000) follow.
·
n = 100
Mean = 0.0026”
Median = 0.0026”
Mode = 0.0025”
·
In other words, the setup is highly precise.
·
Further items of interest . . . the table rotated 110 degrees each half cycle for a total of 220 degrees from the beginning point to the measurement point and then back to the beginning point. I paused one second in PBasic prior to each reversal. (The Peter Norberg Consulting firmware takes care of all ramp-up and ramp-down necessities; you just tell it how quickly to do it. I made the ramp speed changes rather slow.) Each complete cycle of the table took approximately 48 seconds. The extreme variation over the entire set of 100 samples tested was 0.001” (from 0.0021” to 0.0031”, but this only happened a single time during the sampling period).
·
The Jameco 155459 stepper motor never even got warm. The SN754410NE Half-H driver (OK. I had to look I up! So, sue me!) got only slightly warm as measured by the super-precise digital Mark I fingertip probe.
·
Cons: The Jameco 155459 stepper does not have enough torque. I am thinking of trying it as a direct drive stepper without the timing belt and 3:1 reduction. I am sure it will turn the table, I am just not sure if it will handle much of a load as a direct drive device.
·
IN THE MEANTIME, for future testing, I am contemplating the Peter Norberg Consulting UCC3030, a three amp stepper motor controller and either the SherlineIPD-supplied (it’s an NMB product) 136 oz-in stepper·(scroll to the bottom). By way of comparison, the little Jameco 155459 is a 29 oz-in stepper motor.
·
Remember, my goal is a hobbyist-priced, highly precise robot shoulder. I wonder if a 136 oz-in stepper motor in direct drive has enough torque to do what most hobbyists would like to do? (What is that, anyway?)
·
I would appreciate any suggestions and explanations about what I might have REALLY done and I will TRY to answer any questions.
·
The hardware . . .
A link to a very exciting 8Mb movie of the above action! (You even get to HEAR the stepper ramp up and ramp down! Plus, you can watch the indicator dial and the timing belt move! Could anything be more exciting?)
—Bill
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
You are what you write.
Post Edited (Bill Chennault) : 6/13/2007 12:57:33 AM GMT
·
I have finished the first round of real testing of the stepper motor driven rotary table. Here is the setup (I am writing this as if no one is familiar with what I have posted before) . . .
·
I have a Grizzly·H5685 rotary table designed for a mill. I drive it with a NEMA 17 Jameco 155459 stepper motor via a 3:1 timing belt arrangement from SPD/SI. The stepper motor receives instructions from a Peter Norberg Consulting BS0710-USB stepper controller board. In turn, that board is controlled by a BS2px. (I started off using a BS2, but switched to the faster Parallax processor because I thought there was a baud rate problem; there was not. I could have easily stayed with the BS2.)
·
I built a mount for the rotary table and the stepper motor out of 0.25” aluminum. (Pictures, below.)
·
I knew nothing about stepper motors or controllers and one of my goals was to find out something about them. Now, I know a little more. That is good.
·
Another of my goals is to design a highly precise robot shoulder at a price a hobbyist MIGHT afford. This test was done to determine if a stepper motor would give me the repeatability necessary to meet the definition of “highly precise.”
·
I experimented driving the little Jameco stepper motor with the Peter Norberg Consulting stepper board until I found the parameters necessary to achieve what looked like the best for repeatability and reasonable speed. I discovered that using the BS0710-USB in “mode 0” (half power, full step, single winding) gave me the best results at the time. However, during initial testing, I came to the conclusion that all of my equipment is brand new. And, like a new car, needed to be broken in. This included the stepper, the timing belt and pulleys, and the rotary table. Therefore, I will retest the configuration using the other modes . . . half step, alternate single/double windings on – non constant torque; full step, double windings on, full power, full steps; and microstep, eight microsteps per full step, constant torque.
·
I ran the rotary table configuration as described above in mode 0 for well over 48 hours. At the end of this period, I used the magnetic clamp dial indicator that I had previously affixed to the rotary table to take 100 readings. The results given by the dial indicator, which is marked in 0.0001” (1/10000) follow.
·
n = 100
Mean = 0.0026”
Median = 0.0026”
Mode = 0.0025”
·
In other words, the setup is highly precise.
·
Further items of interest . . . the table rotated 110 degrees each half cycle for a total of 220 degrees from the beginning point to the measurement point and then back to the beginning point. I paused one second in PBasic prior to each reversal. (The Peter Norberg Consulting firmware takes care of all ramp-up and ramp-down necessities; you just tell it how quickly to do it. I made the ramp speed changes rather slow.) Each complete cycle of the table took approximately 48 seconds. The extreme variation over the entire set of 100 samples tested was 0.001” (from 0.0021” to 0.0031”, but this only happened a single time during the sampling period).
·
The Jameco 155459 stepper motor never even got warm. The SN754410NE Half-H driver (OK. I had to look I up! So, sue me!) got only slightly warm as measured by the super-precise digital Mark I fingertip probe.
·
Cons: The Jameco 155459 stepper does not have enough torque. I am thinking of trying it as a direct drive stepper without the timing belt and 3:1 reduction. I am sure it will turn the table, I am just not sure if it will handle much of a load as a direct drive device.
·
IN THE MEANTIME, for future testing, I am contemplating the Peter Norberg Consulting UCC3030, a three amp stepper motor controller and either the SherlineIPD-supplied (it’s an NMB product) 136 oz-in stepper·(scroll to the bottom). By way of comparison, the little Jameco 155459 is a 29 oz-in stepper motor.
·
Remember, my goal is a hobbyist-priced, highly precise robot shoulder. I wonder if a 136 oz-in stepper motor in direct drive has enough torque to do what most hobbyists would like to do? (What is that, anyway?)
·
I would appreciate any suggestions and explanations about what I might have REALLY done and I will TRY to answer any questions.
·
The hardware . . .
A link to a very exciting 8Mb movie of the above action! (You even get to HEAR the stepper ramp up and ramp down! Plus, you can watch the indicator dial and the timing belt move! Could anything be more exciting?)
—Bill
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
You are what you write.
Post Edited (Bill Chennault) : 6/13/2007 12:57:33 AM GMT
Comments
Nice to see some real experimentation going on! I think you've got the repeatability to hobbiest levels, what do you think about the speed?
I did some work with a lead-screw positioner this past winter and had found reasonable repeatability with the stepper. But, for sheer speed and accuracy, using an external high-speed counter to read the encoder and a big DC motor on an H-bridge to power it really improved things.
In your setup, you could encode on the rotating table which would take care of any gear slop (which I imagine is quite small in any case).
Strictly speaking, shouldn't the plunger of the dial indicator be alligned with the small knot on the wooden box to minimize error due to the softer wood deforming after many reps? Just kidding!!
Cheers,
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tom Sisk
http://www.siskconsult.com
·
Gee. How in the world could I have missed the knot hole situation? Now, I have to question all my results in light of indicating on the softer wood surrounding the knot hole! WHEN WILL I EVER LEARN? [noparse]:)[/noparse]
I am not in love with steppers. I am just learning. What encoder CPR did you have in your experiments? HomeShopCNC is offering a 500 CPR encoder on their 350 oz-in and 850 oz-in brush motors. For a reason I do not·understand, they call those motors "servos." Of course, I learned the definition of "servos" here.
I have been thinking of trying out one of Banebot's high-end gear-motors with a Banebot installed and calibrated encoder on the rotory table thingy. However, the Banebot encoder is only 128 CPR. Still, with some coding that does good ramping, you could get very, very close and if you went over you could back up. And, all of this would cost much less than a stepper setup. Of course, the Banebot high-end stuff produces A LOT of torque. If I loaded the rotary table down sufficiently, the Banebot gear-motor would simply twist the table shaft off, I imagine. Still, they are cheap enough--compared to a stepper setup--to experiment with.
Hmmm. I just saw where Banebot added encoders on their gear-motors down to 632 oz-in stall torque. I don't know how stall torque relates to holding torque, though.
The big Banebot gear-motor is rated at 18,061 oz-in of stall torque. Sounds like a lot.
If you can remember, I would sure like to know the CPR you had on your encoder on that lead screw.
Oh! The speed of my rotary table setup. I consider it fast enough for a robot shoulder in either dimension. I am very wary of how much power that little Jameco stepper will produce, however (29 oz-in (or 87 oz-in with the reduction], but I have no real-world experience). But, since they are only $21, I imagine I will get another one and another rotary table and find out. I can "write off" the twenty one bucks and I would like to have another rotary table.
--Bill
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
You are what you write.
Post Edited (Bill Chennault) : 6/13/2007 2:39:38 AM GMT
I read the indicator wrong by a factor of ten. This edit corrects that in the text below. The conclusion remains the same, however. The configuration is highly repeatible.
--Bill
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
You are what you write.