Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
 
Notifications
Clear all

Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)  

Page 32 / 53
  RSS
--
 --
(@)
Illustrious Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: Omnissiah

Nope, powering off and on did not fix it either 🙁 X works somewhat reliably now, but Y still struggles, regardless of how much belt tension there is.

What do you mean with "6 lbs tension"? How do I measure that?

Ideally - use a spring tension gauge to lift the mid point of the lower belt sections.  At 8 ounces, 250 grams, the X belt should deflect about 15/64 inch or 6 mm. The Y belt should deflect 12/64, or about 5 mm.   This sets a tension of just under 8 lbf, or 35 Newtons, a good starting point for new belts.   

Gates GT2 belts are spec'd for a minimum 2 and maximum of 25 lbf.  The NEMA17 motor bearings are rated for about 11 lbf at 1/2" radially.  The belts will stretch a bit and loosen to 6 lbs over a few months.

My math - not foolproof, but I've been using it for a while now and find the numbers work pretty well.

GT2-Belt Tension

This post was modified 5 years ago 2 times by --
Posted : 03/08/2019 11:01 pm
--
 --
(@)
Illustrious Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Uggh - it just clicked I've been using 8 mm shaft specs, not 5 mm shaft specs ...  I've been running and recommending belts a bit tight for the motors.  The LDO NEMA17 specs for a 5mm shaft are 28 N at 20 mm, about 6.3 lbf ... so change the deflection weight to 7 ounces or 200 grams.

This post was modified 5 years ago by --
Posted : 04/08/2019 12:20 am
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

How do the #define TMC2130_CURRENTS_R translate into mA? I'd like to try boosting mine a little via M907 to help diagnose the crashes. I ran a print with crash detection off, and there seem to be legit crashes causing layer shifts and no apparent reason. X/Y move smoothly and I don't have a shot bearing as I'm running Vesconite on X and Y. (I did not have crashes, ever, on the stock PR motors, this started after changing to the OMC 0.9s).

And I did actually use a spring scale on my belts; they are spot on deflection-wise at 200g.

Posted : 04/08/2019 1:59 am
Evan
 Evan
(@evan-2)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

@vintagepc,

Try here. 😎 

Posted : 04/08/2019 3:16 am
--
 --
(@)
Illustrious Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

@vintagepc - any chance your part has flat infill that is curling?  The only crashes I've ever had were on Y and from curl.  It is hard to see and I only discovered it because of the unusual noise the printer was making as it scuffed across the layer... next morning the crash and shift was obvious.  On reprint, I watched more closely and found some areas were just staring to curl when the noise started. And got worse to the point I stopped the print and changed speeds and cooling to reduce the curl.

This is from a subsequent print where I had already made changes and the curl was manageable, but still showing.

This post was modified 5 years ago by --
Posted : 04/08/2019 4:47 am
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: evan38109

@vintagepc,

Try here. 😎 

Thanks, I knew I read it somewhere but search is terrible...

 

Tim:

No, this is just mid-part with no unusual features. All that's printing are perimiters and normal infill. Lifting/warping parts is usually one of the first things I suspect.

Posted : 04/08/2019 11:16 am
Omnissiah
(@omnissiah)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
I'm still having problems with Y homing. For a while it worked reliably after I tightened the belt, but apparently that was temporary. It started failing again pretty much all the time. I've probably spent hours trying different belt tensions. I'm starting to suspect that the steppers/drivers warming up skew the thresholds too much.
 
Once I'm homed, I can run this thing at ridiculous speeds and accelerations (300mm/s and over 6000mm/s²) with a bit more juice on the drivers and crash detection disabled. But getting it homed in the first place is a massive challenge...
 
Kuo, how does TMC2130_SG_THRS_*_HOME work? To increase sensitivity, do I increase or decrease the value? I've tried everything from 1 to 9 and nothing seems to make a difference.

Posted by: Tim

Ideally - use a spring tension gauge to lift the mid point of the lower belt sections.  At 8 ounces, 250 grams, the X belt should deflect about 15/64 inch or 6 mm. The Y belt should deflect 12/64, or about 5 mm.   This sets a tension of just under 8 lbf, or 35 Newtons, a good starting point for new belts.   

Gates GT2 belts are spec'd for a minimum 2 and maximum of 25 lbf.  The NEMA17 motor bearings are rated for about 11 lbf at 1/2" radially.  The belts will stretch a bit and loosen to 6 lbs over a few months.

My math - not foolproof, but I've been using it for a while now and find the numbers work pretty well.

GT2-Belt Tension

Thanks, this is interesting. Unfortunately I don't have any suitable gauges.

Ideally I'd like to measure tension via frequency when plucked. There's a nice Spectrum Analyzer for Android with a waterfall plot that helps determine the frequency ( https://play.google.com/store/apps/details?id=org.intoorbit.spectrum ). Now I just need some reference values. What frequencies do you get when plucking the longest span on the Y axis?

Once that is determined, we could recommend the perfect belt tension this way and hopefully remove one potential problem factor.

This post was modified 5 years ago 5 times by Omnissiah
Posted : 04/08/2019 11:36 am
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

My phone says ~80 Hz on Y and ~60 on X but the X value may be off as it's hitting the carriage and it's hard to get a good pluck.

Also, FWIW on my crash issue, I'm getting both X and Y crashes at about the same rate which suggests it's probably not mechanical. 

Posted : 04/08/2019 11:43 am
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

No luck so far; I've tried bumping the current by 25 mA each time I get a crash on an axis during a print; I've gone up to M907 Y500 X375, still getting Y crashes and don't think I want to push too much further as it doesn't seem to be improving things. I'm kind of at a loss now, any suggestions?

Posted : 04/08/2019 2:21 pm
Omnissiah
(@omnissiah)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I've just disabled crash detection. In the past, 75% of all detected crashes weren't, and the remaining 25% never recovered without layer shift. My verdict is that crash detection is pointless.

Posted : 04/08/2019 3:11 pm
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: Omnissiah

I've just disabled crash detection. In the past, 75% of all detected crashes weren't, and the remaining 25% never recovered without layer shift. My verdict is that crash detection is pointless.

My experience prior to the 0.9 degree steppers begs to differ. I had zero issues with false crashes.

Food for thought; if the driver is self-calibrated during initial homing, based on "normal" running parameters of the axis... what happens if you have a very smooth/low load axis? Does it automatically result in lower running currents and a more sensitive crash detection?

Regardless, I completely tore down and rebuilt my Y axis. We'll see if that made a difference. But I did notice the same behaviour reported before when my printer tried to home at a high Z height - it would hit X=0, bump, and then continue running into the endstop, eventually resetting the printer. Running a G80 X C seems to have fixed that as it recalibrated - it bumps the axis endstop 10-20 times taking measurements.

 

Posted : 04/08/2019 5:17 pm
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I will attempt to summarize SG-threshold tuning. It's arduous and time consuming due to the many test runs needed. I specify values in my firmware that will work for most, but you may need to tune your own.

========

On our Prusas, tuning Stallguard2 comes down to varying two settings.
1. Velocity of motion
2. Stallguard threshold.

Stallguard2 signal is measured by the TMC2130 based on Back-EMF from the motor. We use this signal to detect end stop and print crashes. Of these two uses, end stops are easier to tune because the signal magnitude is much larger than print crashes.

Going to 0.9 degree motors alters the amount of back-EMF, motor resonance frequency and total available torque.

Don't confabulate adjustments of motor current (to alter total torque) with the settings at alter Stallguard2 measurement. Motor current has near negligible effect on stallguard signal. You will just chase your tail changing motor current. Adjust motor current only to overcome low torque causing motion failure. Also, remember that your direct changes to motor current settings apply only to non-stealth (aka standard, spread-cycle) mode.

Motor speed cannot be super slow or there is simply insufficient back-EMF to measure. On the other hand super fast motion makes baseline, back-EMF so high that it swamps out any changes in back-EMF due to stalling. Somewhere between too fast and too slow is also a speed at which the motor hits mechanical resonance. That is also a bad speed to be detect stallguard because resonance signals swamp out stall signal.

Belt tension changes resonance and baseline motion resistance (back EMF). Get your belt tension perfected before tuning velocity and SG threshold. If you alter belt tension, you will shift the usable values. Keeping belt tension in consistent range is important. Printing in a hot chamber will also alter belt tension. That is another variable to keep in mind.

For homing x and y axes, the functional velocity range I found is somewhere between 1500 to 3000 mm/sec. I have set a speed in my firmware that works well for my OMC and Moon's 0.9 degree motors. If your motors behave similar to the ones I have, my values should work. However, it is possible any particular specimen of motor differ from the ones I have tested. Yes, I have tested several specimens of each model to get a more widely applicable value, but there is variation in motors being delivered.

For any velocity you select, you must also find the stallguard threshold that is optimal for that velocity & motor. For exhaustive optimization, one must do test series at various velocities. That will give you a speed x SG threshold table from which to select your optimal values.

Trinamic documentation strongly recommends SG threshold values in the -10 to 10 range.
Lower threshold means more sensitive. As you test a motor, you will find a range of SG_threshold values that work. Some values will be more reliable than others. One must select among those values to find one that is best for your motor.

My testing found ideal SG_thresholds for x and y at 3 or 4

With regard to crash detection. I don't use it because I'm always in Stealth mode on my printers. Based on what I am reading from you users on forum, crash detection is probably not reliable with the reduced SG signal magnitudes of 0.9 motors.

============== Caution ==================
--- advanced option that opens up many ways to screw up your settings ---
You can enable service codes in Configuration_prusa.h by uncommenting

//#define TMC2130_SERVICE_CODES_M910_M918

Then one can temporarily (because this is not stored in EEPROM) set SG_thresholds
Just specify the axis and value. For instance...

M916 X4

This lets you test various values without recompiling and flashing firmware. You must set again after every printer reset or power cycle. Once you find your best values, you would set the values in my firmware. Compile and flash to make permanent.

 

Posted : 04/08/2019 5:40 pm
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Huh. I just finished a 2h print with zero crashes and default currents. Clearly I have unknowingly performed some sort of voodoo ritual.

Posted : 04/08/2019 8:13 pm
Omnissiah
(@omnissiah)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Voodoo shenanigans seem to follow along with 0.9° steppers 😉 But what do you mean with "the driver is self-calibrated during initial homing"? I've heard about something like that before, but I can't find more info from a cursory search in this thread. It might also be what's amplifying my homing troubles...

And thanks a lot Kuo, that's quite the informative. I really do appreciate your persistent willingness to help me and others in this thread.

Posted : 04/08/2019 11:00 pm
vintagepc
(@vintagepc)
Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

"the driver is self-calibrated during initial homing"

I'm not going to be able to find a source for that as it's one of those things I recall reading somewhere (possibly this thread?) but from the vague bits I recall it was something along the lines of the 2130 doing some magic voodoo during its first homing to give it a measure of "normal" for the axis and have a baseline for conditions like stalls or skipped steps. 

If you're having homing troubles, perhaps try G28 X Y C. The "C" seems to invoke the calibration routine explicitly according to the prusa firmware source. N=1 but it solved the failure to home I was seeing with it bumping once and then proceeding to grind into the endstop. 

As far as I can trace it is tmc2130_home_calibrate() in tmc2130.cpp, but I'm too tired to try to make sense of that right now. 

Posted : 05/08/2019 1:44 am
Eric Mathison
(@eric-mathison)
Active Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Just for the record, I'm experiencing the same issues as others with homing crashes and occasionally layer shifts during long prints. These issues only started when I switched to 0.9 steppers and firmware.

My hardware:

Moons' MS17HA2P4100 0.9 degree X & Y
Gates belts & Gates 16T pulley & Gates 20T idler

I reduced Y homing velocity to 1800. But that didn't help.

 

https://ericmathison.com/blog
Posted : 05/08/2019 10:58 am
d_ave
(@d_ave)
New Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
  1. I've been struggling to get my Moons 0.9 working on my Bear properly.  strangely enough all was working fine for the first 3 weeks after I built it (from scratch)  but after that - nothing but trouble.  Finally I got it working with a modified stock 3.7.2 with the only notification being steps 100/8, height 204 (due to bmg extruder)  and homing speed 2200. I noticed an interesting though.  Plug in the printer to a pronterface and run a self test, than a xyz calibration - you'll notice some of the "magic" the firmware is doing while calibrating. Also every time you flash a new firmware I recommend you to create tour eeprom from the arduino - some values are not removed after a factory reset and they mess everything up 
Posted : 05/08/2019 12:03 pm
Eric Mathison
(@eric-mathison)
Active Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

What would be a good belt tension gauge to use?

Would something like this work: https://www.amazon.com/Gates-7401-0076-Pencil-Tension-Deflection/dp/B00CJEGXK6

 
 
https://ericmathison.com/blog
Posted : 05/08/2019 1:12 pm
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I measured some belt deflections with my Lyman digital trigger gauge on two machines that both home reliably.

https://www.amazon.com/gp/product/B00162QGLS

Both have 0.9 Moons on X and 0.9 OMC on Y. X-axis is "short ears" on both units.

Getting consistent amount of deflection is fairly difficult. So I had the gauge average about 10 measurements. Pulled bottom belts at their mid free SPAN upward by 5 mm. 

Bear: X - 166 gm, Y = 128 gm at 5 mm deflection

Prusa: X = 234 gm, Y = 147 gm at 5 mm deflection

 

Unfortunately, it was near impossible to accurately hit 5 mm each time. So, the above 5 mm pull measurements are fairly unreliable figures. Pulling up to just touch the upper span was much easier to do consistently.

Bear: X - 560 gm, Y = 682 gm up to touch upper belt

Prusa: X = 682 gm, Y = 772 gm up to touch upper belt.

My frames are very well damped on sorbothane feet and have silicone drum dampers on metal frame parts. Strumming the Y belt is too quiet to pick up with microphone for frequency measurement. X I could get clean enough sound if I strummed TOP belt. Strumming bottom just hits x-carriage too much.

Prusa X top 129 Hz

Bear X top 172 Hz

If you are paying attention, you'll notice that the Prusa X (which is at higher tension by trigger gauge) vibrates at a LOWER frequency. It's the same Gates GT2 belt and span lengths are both 25 mm!

My attempt to get some more objective data to help others replicate my setup seems to yield confusing data.

I think I'm staying with my 2 mm deflection at palpable increase in rate of tension change method.

=======

Thinking of things that might differ between my machines and ones having homing difficulty

1. My frames are damped, but I don't think that will affect motor resonance enough to mess up homing.

2. My motor cables were all undone and retwisted to be twisted pairs instead of flat cable. That could possible make them more EM noise resistant.

3. My motors are simply smoother specimens. The 0.9 motors (with belt and cables detached) should rotate in your fingers much more smoothly than the original 1.8 degree motors. Detentes should be finely spaced and soft. No clunk clunk as you feel the shaft rotations.

4. I have done full factory resets with all data cleared.

5. The printers are afraid of my tearing them back apart.

 

 

This post was modified 5 years ago 2 times by Bunny Science
Posted : 05/08/2019 7:05 pm
Eric Mathison
(@eric-mathison)
Active Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I'd like to thank you for all of your efforts and information that you've been providing. I just ordered the Lyman digital trigger gauge. I'll report back in a week or so.

https://ericmathison.com/blog
Posted : 05/08/2019 10:42 pm
Page 32 / 53
Share: