Hello Guest it is March 29, 2024, 11:01:50 AM

Author Topic: Hybrid Stepper Setup in M4  (Read 5202 times)

0 Members and 1 Guest are viewing this topic.

Hybrid Stepper Setup in M4
« on: December 01, 2018, 03:20:17 PM »
I'm using some Schneider LMD drives which are Ethernet enabled closed loop steppers and they do some very odd things in M4 with a Pokeys57CNC. They do odd things differently in one direction than another and I can't get the speed anywhere near the way these things are capable of.
As an example if I send them a direct command via a terminal app they move liked greased lightning the EXACT relative distance I command but if I take the math I use to M4 then nothing is as precise and crisp. They even move oddly less in one direction than another...every cycle I lose a few thou off each side...it's absolutely bizarre.
I can literally set them up in 256*1.8/200 MS which is the default spec (51200MS) write some Schneider based MCODE for doing circles and the machine fairly flys around (almost scary) on little dovetail ways with .0002-3 run-out...I can't get anything near that through M4
I'm hoping someone else or MACH staff has dealt with these drives before.
Thanks!
Re: Hybrid Stepper Setup in M4
« Reply #1 on: December 01, 2018, 04:18:43 PM »
Hi,
I think your microstepping regime is wrong.

You have left it as default at 256 microsteps per full step or 51200 microstepps per revolution. If you want your stepper to do 1200rpm
or 20 revs per second then the pulse rate would be 20 X 51200 =1.024MHz. That's radio frequency stuff.

This is from the advertising material of the 57CNC:

Quote
•high performance 8-axis 125 kHz pulse engine with dedicated motor connectors

Thus you are demanding a pulse rate from the 57CNC ten times its capability.

Microstepping with steppers is not, contrary to the name about increasing resolution, microstepping beyond half steps seldom works,
although is closer to true representation with closed loop drives. The real purpose of microstepping is smoothness of motion.

In the early days astronomers used steppers to drive there telescope drives but suffered from vibration. They came up with the idea
of microstepping, no doubt hoping to achieve greater resolution but found that dream illusory but did end up with much smoother
motion.

May I suggest you drop the ridiculous 256 microstepps per full step......that's 25 arc sec resolution.....just pie in the sky stuff.
Try 8 microstepps per full step. That's resolution of 13.5 arc min, which is very acceptable AND take advantage of the smooth motion.
That would reduce the pulse per rev to 8 X 200 = 1600 per rev. At 20 revs per sec it would require a pulse rate of 32kHz, well
within the 57CNC ability.

Craig
'I enjoy sex at 73.....I live at 71 so its not too far to walk.'
Re: Hybrid Stepper Setup in M4
« Reply #2 on: December 01, 2018, 06:02:33 PM »
Yea I did that math too but was hopeful something might be possible but that does in fact make perfect sense...it's just sad I guess since the internal capabilities of the drive itself are nothing short of amazing...

Is there perhaps another controller to drive this level of motor more effectively...this is a toy for me in truth and it's the exercise itself that intrigues me the most...;)

Thanks Craig!
Re: Hybrid Stepper Setup in M4
« Reply #3 on: December 01, 2018, 06:12:45 PM »
While I agree with Craig that anything finer than 8 micro steps per full step
is wishful thinking, it is possible to run at higher step rates using a device
with step pulse generators in dedicated hardware. The SmoothStepper
is one example. It can produce steps at up to 4 MHz.
Steve Stallings
www.PMDX.com
Re: Hybrid Stepper Setup in M4
« Reply #4 on: December 01, 2018, 06:23:18 PM »
There we go....Steve and Craig...do you like the ESS better perhaps than a Pokeys...in all honesty I'm only half-hearted about the 57 in general and have a second project after this...Thanks Guys!
Re: Hybrid Stepper Setup in M4
« Reply #5 on: December 01, 2018, 06:34:35 PM »
Hi,

Quote
Is there perhaps another controller to drive this level of motor more effectively...this is a toy for me in truth and it's the exercise itself that intrigues me the most

There is, the Ehternet SmoothStepper (ESS) by Warp9 TD is good to 4MHz and the Hicon by Vital Systems is goo to 8MHz.
You don't need either.....the only reason you are having difficulty is because you wish to retain totally impractical and unrealistic resolution.
Its good for bragging......but it has no substance in reality.

With your current setup:
51200 pulse per rev connected to a 5mm pitch ballscrew suggests a linear resolution of 0.1um....or 100nm. If your machine is that good that you
can expect 100nm accuracy then what are you doing pissing around with steppers (of any description), a 57CNC and Mach4?

With my recommendation of 8 micro steps per full step:
1600 pulse per rev connected to a 5mm pitch ballscrew is a linear resolution of 3.1um. That's impressive.....not just for a hobby machine but would
be respectable is heavy industrial machines worth hundreds of thousands of dollars. You will also be able to take full advantage of the speed of the
steppers.

Steppers, closed loop  or otherwise lose torque the faster you go. Even good (low inductance) steppers are likely to have less that 25% of their holding
torque at 1000rpm. If you believe the advertising material that suggests closed loop steppers avoid that problem then....YOU"VE BEEN HAD by slick
advertising. I would argue that expecting a stepper to go much faster than 1000rpm is not realistic.....that sort of speed it where servos come into
their own.

In truth (I have just read your comment about the ESS) is that you don't need a better controller but be realistic about what you expect from it.
I like the ESS, I have one and love it. I have a servo as a spindle motor and to drive it at full resolution (8000 counts per rev) at full speed (3500rpm)
requires a pulse stream of 466kHz. The servo drive can handle up to 500kHz with a differential drive. So I had to make one for my ESS as my BoB
couldn't manage that being single ended. If you think its a simple matter to drive a signal in the low MHz range you're dreaming.

Craig
'I enjoy sex at 73.....I live at 71 so its not too far to walk.'
Re: Hybrid Stepper Setup in M4
« Reply #6 on: December 01, 2018, 06:45:23 PM »
Hi,
this is from the Schnieder website for your 23 size stepper:

Quote
Encoder (3) Line count 1000 lines / 4000 edges per rev

So what's the point in trying to achieve a micro step regime that has finer resolution than the encoder can support?
The encoder has 1000 lines or therefore 4000 counts per rev. Trying to exceed that is pointless.

Craig
'I enjoy sex at 73.....I live at 71 so its not too far to walk.'
Re: Hybrid Stepper Setup in M4
« Reply #7 on: December 01, 2018, 06:56:40 PM »
ahhhhh...Craig you take the fun all out of "why not"...lol...I will reset them...I promise...I will rerun my test setup and go from there...but damn...

Thanks for the reality check...;)
J
Re: Hybrid Stepper Setup in M4
« Reply #8 on: December 01, 2018, 07:12:14 PM »
I just had X,Y buzzing around to the table limits for a few minutes straight flat out on an internal program and then on esc back  to my dial indicator and .0002...the sound alone is music...lol...but I'll dial them back like I promised.
Re: Hybrid Stepper Setup in M4
« Reply #9 on: December 01, 2018, 07:40:27 PM »
Hi,
even I you had a controller that could deliver the pulses and the electronics to drive a signal over the cable so you could use
256 micro steps OR 8 micro steps I doubt you could tell the difference. The resolution at even 8 micro steps is likely to exceed the
rigidity and accuracy of a hobby machine so the finish and accuracy of the parts you make are going to be determined by the
mechanics of the machine....not the resolution of your steppers at all.

I see your comment about whizzing the machine around but from the internal program. The internal program is speed limited only by
the TCP/IP speed, ie very high indeed.

If you had Gcode to copy that program and run it in Mach4 through your 57CNC controller provided you kept within the pulse speed limitation of
the 57CNC it would perform perfectly. But that would mean slowing it down to approx. 10% of its current speed. Alternately you can, as I have
recommended, reduce the resolution and therefore can increase the speed. There is a tradeoff and balance that you can experiment with.

In my previous post I told you about me making my own line driver to drive my servo at 466kHz. I needed that frequency only because I wanted
full resolution, ie 8000 counts per rev or 2.7 arc min......for a spindle.....why? The reason is because I was being a jerk....I don't need anything like
that resolution on a spindle, 30 arc min would be entirely adequate for rigid tapping. That would have reduced the required pulse speed to 42kHz
which my single ended BoB could have done easily!

Craig
'I enjoy sex at 73.....I live at 71 so its not too far to walk.'