Hello Guest it is July 12, 2020, 04:36:56 PM

Author Topic: Optimizing Isolation Routing and Motor Tuning questions  (Read 1601 times)

0 Members and 1 Guest are viewing this topic.

Optimizing Isolation Routing and Motor Tuning questions
« on: May 25, 2018, 07:08:01 PM »
Background: My 20" x 40" DIY CNC table has a 14" x 30" machinable area and has been in hobbyist service for 7 years, always using Mach3 v3.043.066.

My PC is 64-bit. 3Ghz, 3GB RAM, Windows 7 Pro (had been XP years back), Mach3 Kernel speed has always been 25Khz, parallel port driven BOB.  B.O.B. is CNC4PC C1R10.
For XY axis: Keling KL-4030 Drivers, Keling KL23H286-20-8B 425 in/oz stepper motors, powered by KL 350-36, 36 volt, 9.6A Switching power supply.  XY gantries are timing belt driven, Z gantry is ACME Lead Screw driven. All axis use V-groove bearings on aluminum linear rails.  I recently modified the .75" x .75" aluminum rails by adding anodized V-Rails to them while maintaining their basic dimensions.  Recent "pull" tests of the X and Y gantries with timing belts removed showed the X-gantry required between 3-1/2 to 4 lbs of pull to move through full travel.  The Y-gantry required 2 to 2-1/2 lbs of pull to move through full travel.  The pull required prior to the V-rail upgrades are unknown but believed to have been a few pounds more for each axis.  The XY timing pulleys are 1" diameters, so the stepper motors have more than sufficient drive torque for their loads.  Stepper driver's heat sinks are only ever slightly warm.  With the V-rail additions, the XYZ axes travels are smooth with very low mechanical slop.

Calculated XY resolution is .0008".  As such, I am trying to determine why diagonal bit travels have visible stair-stepping and why some circular pads one straight circuit traces have their upper and lower sides slight offset from each other.  From what I've read in researching this forum, this could be due to motor tuning, velocity vs acceleration vs CV Control settings.  CV Control settings have always be defaults, with only two items checked: CV Dist Tolerance=180units, G100 AdaptiveNubsCV.
I'll post a reply with the XY Motor Tuning screenshot but this post includes back lit photos of recent isolation routed PCB showing areas of interest for feedback.

The isolation routing was done using a .005" dia. tip, 60 degree engraving bit, 20,000 RPM, Feed Rate 15 IPM, D.O.C. of .006", through 1 mil copper clad board 1/16" thick.

X-CircleTracking-1 photo shows upper and lower half of solder pads are slightly offset in the X-axis and this does also occur when trace and pad are aligned on the Y-axis.  Is this an acceleration issue?
X-CircleTracking-2 photo shows vertical and horizontal traces routed are nice, smooth and straight (expected behavior) with diagonal portions of these traces show stair-stepping.  Some stair-stepping is expected but on the order of .001", or so I think.  Is this also an acceleration issue?

P1000687-c photo, A board I did about 6 years ago, does not show the same degree of stair-stepping.

I'm wondering whether since reducing the XY axis friction with the V-rail modification is the a factor in the root-cause of what is now happening.

Thanks for your feedback and let me know if I can provide further information.





airnocker

Everything depends on everything else
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #1 on: May 25, 2018, 07:09:23 PM »
Here is my XY Motor Tuning screenshot that the forum wouldn't allow me to add to the original post.  Although it is only the X, the Y is identical.

airnocker

Everything depends on everything else
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #2 on: May 25, 2018, 07:48:55 PM »
Hi,
are you using microstepping?

Your XY steppers may have 425oz.in torque between full steps and likely to have very close to that between 1/2 steps but between microsteps very much less.

Lets say the rotor is currently on a full step position. The current in winding A is max while current in winding B is zero. If an 1/8 (8 microstepping assumed) step move is called the current in
A will reduce to about 7/8 of max and in B up to about 1/8 of max. The torque acting on the rotor trying to get it to move is about 1/8 of rated torque, ie only 53oz.in
If the drag or load is such that 53 oz.in is not enough to move the axis it will stay still. If another 1/8 step move is called  the current in A will reduce to 3/4 max and
in B increase to 1/4 of max and the torque will be about 1/4 of rated or 106oz.in. This will probably be enough to move the axis and so it will move two microsteps,
this one and the one it missed previously.

Microstepping is vey good a smoothing motion and suppressing vibrations but it is not a good method for increasing resolution. The above example is why. The torque between
adjacent microsteps is very poor and cannot be relied on. Torque between full steps can. In fact torque between half step can also usually be relied on to move.

May I suggest do the calculation 'what is the resolution of my machine if I can rely on half stepping only' It may well be much less resolution than you expected and may describe the discrepencies
you see.

Craig
My wife left with my best friend...
     and I miss him!
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #3 on: May 25, 2018, 08:41:22 PM »
Hi,
are you using microstepping?

Your XY steppers may have 425oz.in torque between full steps and likely to have very close to that between 1/2 steps but between microsteps very much less.

Lets say the rotor is currently on a full step position. The current in winding A is max while current in winding B is zero. If an 1/8 (8 microstepping assumed) step move is called the current in
A will reduce to about 7/8 of max and in B up to about 1/8 of max. The torque acting on the rotor trying to get it to move is about 1/8 of rated torque, ie only 53oz.in
If the drag or load is such that 53 oz.in is not enough to move the axis it will stay still. If another 1/8 step move is called  the current in A will reduce to 3/4 max and
in B increase to 1/4 of max and the torque will be about 1/4 of rated or 106oz.in. This will probably be enough to move the axis and so it will move two microsteps,
this one and the one it missed previously.

Microstepping is vey good a smoothing motion and suppressing vibrations but it is not a good method for increasing resolution. The above example is why. The torque between
adjacent microsteps is very poor and cannot be relied on. Torque between full steps can. In fact torque between half step can also usually be relied on to move.

May I suggest do the calculation 'what is the resolution of my machine if I can rely on half stepping only' It may well be much less resolution than you expected and may describe the discrepencies
you see.

Craig

Thanks for the fast response Craig.  Yes, I am using 1/16 micro-stepping, and forgot about that reducing the effecting torque.  So my resolution at 1/2 stepping would be 8 times worse or .0064" which is slightly less than what I am seeing in these photos, however, I've always used 1/16 micro-stepping and the pcb from 7 years ago does not show these artifacts of today, although it does show some.  That's what leads me to wonder if it is a change is frictional forces in X and Y rails.

To add more information and perspective, back in April I did two copies of the isolation routed PCB shown in my photos here, They showed little diagonal stair-stepping.  After doing these two copies, I made design changes and was preparing to do two more with the new changes.  I happened to checked the tension on my XY lower V-groove bearings and found I could easily turn them with very little friction, so I snugged them up a bit, then ran the new g-code, twice.  That's when I noticed some of the larger spiral drilled holes were oval, not round as before, noticeable stair-stepping and greater position errors.  It was then I confirmed the uneven wear-down of the top edges of the aluminum 90 degree angle by the v-groove bearings was a probable cause.  This is why I decided to add the v-rails and measure the XY axis travel forces afterwards.

I'm grateful to be reminded that micro-stepping is not a desirable way to gain travel resolution, too.

Thanks Craig
airnocker

Everything depends on everything else
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #4 on: May 25, 2018, 09:23:58 PM »
Quote
So my resolution at 1/2 stepping would be 8 times worse or .0064" which is slightly less than what I am seeing

Here are comparative resolution calculations:

Based on this I wouldn't be loosing much by changing to 1/8 microstepping just to compare results, although it is a tiny torque improvement on the worst-case side.
airnocker

Everything depends on everything else
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #5 on: May 26, 2018, 01:01:42 AM »
Hi,
my understanding, I wish I'd kept the article from which it came, is that no matter what microstepping regime you have in place your effective resolution
is that of half stepping. The only way to really increase resolution is to have gear reduction. Your machine is and has been fine with 16 microsteps, I suggest
leave it that way but realize the true (or reliable) resolution is that of 1/2 steps.

You have already established that a small difference in mechanical conditions can change the machine. If there were zero friction, zero load and zero inertia then you could get your
0.0008 inch resolution, more realistically you might get 0.0016 inch but can probably guarantee no better than 0.0064 inch.


I use Vexta 5 phase steppers on a low lash 10:1 planetary gearbox. The 5 phase steppers are natural 500 steps per revolution or 0.72 degree. With the gearbox
that becomes 0.072 degree per step, or 5000 steps per rev. I have 5mm pitch ground ballscrews and have a step/unit of 1000 steps per mm. I could if I used
1/2 stepping improve the resolution to 0.0005 mm or 2000 steps/mm. The planetary gearbox have a maximum of 3 arc minutes of backlash or the equivalent
of 0.0007 mm or 0.7um. It didn't seem sensible to me to try to have a resolution finer than the potential backlash.

I have good and reliable resolution but because of the gearing its quite slow. My stepper and drivers go 2400 rpm no trouble so with the gear reduction that's 240 rpm
of the ballscrew, at 5mm pitch that is 1200mm per minute. You don't get something for nothing.

Craig
My wife left with my best friend...
     and I miss him!

Offline RICH

*
  • *
  •  7,411 7,411
    • View Profile
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #6 on: May 26, 2018, 08:32:58 AM »
airnocker,
I have attached a motor curve for you your steppers.
1. Find the torque to just move the axis. The best way is to actualy measure the torque using a
     torque gauge. I personaly use an oz-in gauge. The intent is to adjust individual components of
     the axis and use the gauge so you know what the affect is of any adjustment. This includes
     adjusting for any non movement of the axis and also preload of bearings.When done you you
     now have a value for the total torque. Note that belt tension has an affect on not only torque but
     also accuracy of  movement.
   
NOTE: There are programs that will calculate required motor torque values based on a machines
             configuration.
   
2.Now look at the motor curve. For any given torque there is an associated pps. Just note the basis
   for the motor torque curve as it is 30V , 4.5 A / phase @ 1/2 step setting.
  I suggest you have a good read of the information available on the Gecko site about steppers.
  1Nm= 141 in-oz, so approx you have 300 in-oz at 1/2 step setting. For your 15 IPM see where
  you are on the motor curve. I would suggest a motor max pps of  1000 to 2000 as you want power.

Graig makes good points on theoretical as compared to actual resolution. You will read on the Gecko
site that micro stepping above 10 will gain you nothing in terms of resolution.  If you want true resolution
it comes from the mechanics of the system. Just another example of resolution.......not bragging......
my engraving machine only runs at a max feed rate of  3-5 ipm, steps per unit are approx 350,0000
so 1/350000= 0.000003" but true axis accuracy is 0.000040 ( not easy to measure a micro inch movement )!

FWIW,
Have fun.......

RICH
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #7 on: May 30, 2018, 05:04:31 AM »
Hi,
I am not a fan of encoder feedback stepper servos. The motors are still steppers and suffer the same reduction in torque with speed as open loop steppers.
The manufacturers play the marketing game very well and work with the buyers fear of 'losing steps'. As it turns out an open loop stepper is no more prone to losing steps
than a feedback stepper and the notion that feedback is going to be able to correct stepper lag under overload condition (where they are likely to lag) is fallacious.

What the manufacturers can claim is an marked improvement I resolution. Because of the low differential torque between adjacent microsteps on non-feedback steppers, motion from one
microstep position to the next can't be assured and therefore cannot really be counted as resolvable steps. The encoders used on Leadshine stepper servos are 1000 ppr or
4000 count. Because the encoder can measure between full steps and the stepper driver can adjust its output so that the rotor lands on exactly the right location between steps, ie microsteps
ARE resolvable. This means Leadshine stepper servos go close to a resolution of 4000 steps per revolution whereas a open loop stepper can achieve 400 steps per revolution at half step mode.
While this represents a considerable improvement over open loop steppers I still regard the premium paid for feedback steppers to be unjustifiable when for a small extra sum you could be into
genuine AC servos which kick stepper anus all day long!

Craig
My wife left with my best friend...
     and I miss him!

Offline Davek0974

*
  •  2,597 2,597
    • View Profile
    • DD Metal Products Ltd
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #8 on: May 30, 2018, 10:14:11 AM »

While this represents a considerable improvement over open loop steppers I still regard the premium paid for feedback steppers to be unjustifiable when for a small extra sum you could be into
genuine AC servos which kick stepper anus all day long!

Craig

Got that right :) I nearly got closed loop steppers for my last build, fell for the hype and sales blarney, luckily I was guided towards AC servos and got some imports from AliExpress - when testing one on the bench, found they were 100% amazing devices, no steps, practically strong enough to break any machine i could build and instant success, full torque from standstill to full pelt too.
Bridgeport Mill, Mach3 V062, CSMIO-IP/A controller, AC Servo Drives.
Plasma table, Mach3 V062, Step motors, C&CNC THC.

Offline ger21

*
  • *
  •  6,282 6,282
    • View Profile
    • The CNC Woodworker
Re: Optimizing Isolation Routing and Motor Tuning questions
« Reply #9 on: May 30, 2018, 10:29:58 AM »
As others have said, this is almost certainly due to the resolution being too low. The machine is belt driven, directly from the stepper? You prbably want it geared so that one revolution of the stepper gives you about 1/2" of travel. This is a decent balance of speed and resolution.

You also may be seeing some belt flex, which could be the culprit on the circles.
Gerry

2010 Screenset
http://www.thecncwoodworker.com/2010.html

JointCAM Dovetail and Box Joint software
http://www.g-forcecnc.com/jointcam.html