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

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

Página 16 / 53
  RSS
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

sounds like you have the bases covered. 

Respondido : 22/06/2019 2:38 am
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I had to go back to the prusa pulleys (it may be all in my head, but at least it's working with one less unknown right now).  After a failed benchy with about 8 back and forth Y layer shifts I couldn't get it to home Y without bottoming out my tensioner.  

What firmware variables would I edit to reduce the hit strength?  Not the speed, but the force it hits the end stop with?  

In listening to the machine sounds during homing operations, it sounds as though it's simply impacting the stop with too much force. On the X I can hear it on the test to check the X axis distance.  It will hit so hard sometimes it will register as a loose pulley.  Same goes for the Y.  

How does the homing threshold variable work?  Is it the amount of current it must generate to trigger as a hit?  Meaning, the higher the value the more solid of a hit/push back is has to register?  

The reason I ask is that I'm definitely having to tension the belts much much higher than I normally would right now to get it to properly trigger the homing.  So I'm going the wrong direction. 

Respondido : 22/06/2019 4:18 am
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Yes, the more I hear of your efforts, the more alarming the direction. You're seeing all sorts of what sound like mechanical binding effects that I would associate with too high belt tension. You can't set hit hardness directly, only velocity. I think you've managed to go too far up in speed.

I would revert all the way back to my original values in the firmware, probably easiest by re-downloading and editing just the config.h language setting and define which are 0.9 motors in Configuration_prusa.h

Set machine to stealth. 

Set belt mechanical tension to 6 lbs (pretty loose). Do so by pinching the x-idler into x-end with just one finger and thumb. Set your bolts to hold that. Feel the x-belt and replicate that feel on y.

We're going to intentionally have to be too loose. Fail the homing. Power off / increase belt tension slightly / power on / try homing again. 

Do small increases of 1/4 bolt turn until you hit the right range for X. I'm going to be it is way looser than you current have it. I would expect 1/2 lb to deflect your belts at least 1/4 inch.

A rough approximation -- make a fist. See that little tendon in the middle of your wrist? Press down on your tendon. The belts should not tenser than what your tendon feels like.

Once x is working, match y to x by feel, (freaking ignore the support numbers because they are extra meaningless now that the feedback is totally different from 0.9 motors. Those belt numbers are useless.

 

Esta publicación ha sido modificada el hace 5 years por Bunny Science
Respondido : 22/06/2019 5:47 am
AnatomicFlack me gusta
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
Twisted pairs

There is one little detail that is different in my setup that I haven't mentioned because I'm not sure it makes enough difference. I split apart my Yotino motor cable wires and made them twisted pairs instead of a flat ribbon. That was done to minimize EM noise, but I don't know if it actually makes an difference. Could that reduce signal noise enough to make stall detection more reliable? I don't think so, but it might be worth trying.

Red & blue are one pair

Green & black are other pair.

I released the connection pins from one housing, split the wires apart, and used a drill and paper clip to twist each pair. The popped them back into their connector housing. You did write down the pin positions or keep one for reference.

This pict shows what I did but these are of course not the Yotino cable harness. You can still see each pair is twisted.

(not shown) After each pair is twisted, I then loosely twisted the pairs together to form a nice cable.

Esta publicación ha sido modificada el hace 5 years 2 veces por Bunny Science
Respondido : 22/06/2019 11:07 am
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I went back to your default values a while ago.  I was trying to eliminate any extra variables.  

Last night I was lowering the homing threshold to 2 on both x and y and had some limited success. On your defaults my tension still needs to be tighter by feel than I'm used too but the selftest numbers are reporting about 255-265. Which leads me to agree with you that those numbers are largely relative to nothing but changing parts and can't really be compared to differing part configs. 

I was watching the console output last night when running the left test to get a better idea of what's going on to cause the various failures. On thing I was seeing is that my Y axis length is reporting as slightly larger than what must be stored in from the initial XYZ calibration I ran after the new setup. The front stop may be having trouble stopping the bed. I'm trying to determine if that's due to stepper force or a failing part.  

The 0.9s just seem to operate with more force than the 1.8s. The hits are just far more solid sounding than I'm used to at any belt tension. 

I'm going to run through the steps you outlined today. I've done similar a few times already but with also changing parts in the mix.  

I'm not giving up, I just need more feedback from the machine to know which direction to head.  What is the value output to the console during the homing tests? mscnt I think it's called? 

As far as the motor cables... That's exactly what I did. I split all four wires apart and twisted paired the black/green and red/blue.  Then sleeved the whole thing. I did shorten my cables by half the original length in the process because I also put on the proper molex ends for the EINSY side and didn't need all that extra length. 

Esta publicación ha sido modificada el hace 5 years por AnatomicFlack
Respondido : 22/06/2019 2:24 pm
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Sorry you are having such a time getting things sorted.

Maybe you should also test with one of the OMC's. They are pretty cheap, but you have to grind the shaft flat and wire in a connector. That would give you a second known to work (here) motor model.

Respondido : 22/06/2019 2:38 pm
AnatomicFlack me gusta
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: guy.k2

Sorry you are having such a time getting things sorted.

Maybe you should also test with one of the OMC's. They are pretty cheap, but you have to grind the shaft flat and wire in a connector. That would give you a second known to work (here) motor model.

Please don't apologize, you are doing more than anyone else in the community to even make using 0.9s possible. I fully expected to face some issues and I'm too stubborn to give up that easily. 🙂 I'll be honest, if I can't get the Moons to cooperate with the RAMBo I may seriously start researching using a Duet.  I didn't want to make such a major switch in hardware or diverge from the PRUSA firmware (especially because I haven't even looked into how that'll impact the MMU2), but we may be pushing the limits of the current board to manage higher precision hardware.  

I might pick up a couple of OMCs to test eventually.  Because my machine has always been touchy printing in stealth and very susceptible to layer shifts my suspicion still lies with my RAMBo being a big part of the problem.  At least we'll be able to rule that possibility out shortly.  I've always had to use normal mode whenever my machine acts up and begins layer shifting without good reason.  

For the moment, I'm able to print in normal mode and my homing seems pretty reliable so far with the homing thresholds on X and Y set to 2. I'm printing a benchy in normal mode now, I'll try one in stealth next to see if I get more layer shifts.   

 

Respondido : 22/06/2019 3:43 pm
CybrSage
(@cybrsage)
Honorable Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Agreed, he is a HUGE help!

Respondido : 22/06/2019 4:18 pm
AnatomicFlack me gusta
nikolaistolstoy@gmail.com
(@nikolaistolstoygmail-com)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: guy.k2

nikolaistolstoy are you printing in normal or stealth mode? Also if in normal is crash detection on?

Trying to understand whether your layer shifts are simply missed steps or a detected crash followed by a failed rehoming/repositioning.

My machine is always in stealth and (obviously) crash detection is off.

BTW, the system switches temporarily out of stealth during homing. Otherwise homing would never work. 

 

Normal mode. Just looked and X crashes is up to 13 and Y 0, definitely way more than before. Do you think switching to stealth will eliminate these possibly false positive crash detections? 

Respondido : 22/06/2019 5:10 pm
nikolaistolstoy@gmail.com
(@nikolaistolstoygmail-com)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: AnatomicFlack

I had to go back to the prusa pulleys (it may be all in my head, but at least it's working with one less unknown right now).  After a failed benchy with about 8 back and forth Y layer shifts I couldn't get it to home Y without bottoming out my tensioner.  

What firmware variables would I edit to reduce the hit strength?  Not the speed, but the force it hits the end stop with?  

In listening to the machine sounds during homing operations, it sounds as though it's simply impacting the stop with too much force. On the X I can hear it on the test to check the X axis distance.  It will hit so hard sometimes it will register as a loose pulley.  Same goes for the Y.  

How does the homing threshold variable work?  Is it the amount of current it must generate to trigger as a hit?  Meaning, the higher the value the more solid of a hit/push back is has to register?  

The reason I ask is that I'm definitely having to tension the belts much much higher than I normally would right now to get it to properly trigger the homing.  So I'm going the wrong direction. 

I also had to go back to Prusa pulleys after trying both derblue and ballinstein. I just put the pulley in reverse on Y and didn’t heave to grind anything. Problem was those pulleys were not as well made as Prusa and the effective circumference was changing on the pulley causing resistance on the axis to go up and down and I would get the motors stalling in middle of travel. The smaller grub screws on those also were smaller than Prusa and so I was getting movement on motor shaft and wavy walls.

Respondido : 22/06/2019 5:14 pm
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Heh... Guess which one was printed in stealth mode? 😉

Yesterday's one had 6 Y shifts before I killed the job. 

I need to try some longer prints in normal... But stealth hates life right now using the same exact tensions that work fine in normal mode.  

My printer at the moment hates all things stealth. 

Respondido : 22/06/2019 6:25 pm
Bunny Science
(@bunny-science)
Noble Member
Topic starter answered:
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Nikolai, yes, Stealth mode will almost certainly stop the false positive crash detections. Stealth mode also turns off crash detection.

anatomicflack, a printer that prints in standard mode, but not stealth has mechanical resistance problem. Something is creating excess resistance to motion. Higher torque of standard mode is enough to move the axes, but the weaker stealth mode cannot overcome the resistance. You're an experienced 3D printer user. So I hesitate to post the dx sequence, but it is pretty obvious that you have a binding or friction issue.

Completely disconnect and remove belts from both axes.

Move carriages over entire range. There should be practically zero resistance or binding through the entire range. Basically a flick of a finger should make y axis fly across its entire movement range. The x-axis won't fly across its entire range (due to the wire umbilical), but it should fly across most of its range. If there is any resistance, sort out the bearing condition, lubrication, or rod alignment problem. You may need to adjust clamping tension of the carriage to achieve free motion.

One easily lurking problem - check that no wires in the x-carriage are clamped against top of lower bearing. It is easy to let a motor wire drop lower than the wire channel and impinge with the bearing. That pushes the lower bearing out of alignment as one tightens the bottom x-cover. Make sure those wires stay UP while the u-shaped lower x-cover is being attached.

Finally, one cryptic issue - idler pulley friction. Thread belt through just the idler and through nothing else. Hold the two ends of belt to approximate belt path. Move belt ends back and forth feel in a flossing motion to feel whether the idler freely spins. I recommend testing the idlers this way because this problem bit me but was undetected by conventional checks. The idler felt like it could move ok when felt with my finger, but "flossing" a belt back and forth revealed it was actually dragging.

There is something creating extra resistance and making stealth mode not work. You will find it.

If you find absolutely nothing creating friction or binding - then yes, something stinks in that EINSY.

 

 

Respondido : 22/06/2019 8:25 pm
AnatomicFlack me gusta
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: guy.k2

Nikolai, yes, Stealth mode will almost certainly stop the false positive crash detections. Stealth mode also turns off crash detection.

anatomicflack, a printer that prints in standard mode, but not stealth has mechanical resistance problem. Something is creating excess resistance to motion. Higher torque of standard mode is enough to move the axes, but the weaker stealth mode cannot overcome the resistance. You're an experienced 3D printer user. So I hesitate to post the dx sequence, but it is pretty obvious that you have a binding or friction issue.

Completely disconnect and remove belts from both axes.

Move carriages over entire range. There should be practically zero resistance or binding through the entire range. Basically a flick of a finger should make y axis fly across its entire movement range. The x-axis won't fly across its entire range (due to the wire umbilical), but it should fly across most of its range. If there is any resistance, sort out the bearing condition, lubrication, or rod alignment problem. You may need to adjust clamping tension of the carriage to achieve free motion.

One easily lurking problem - check that no wires in the x-carriage are clamped against top of lower bearing. It is easy to let a motor wire drop lower than the wire channel and impinge with the bearing. That pushes the lower bearing out of alignment as one tightens the bottom x-cover. Make sure those wires stay UP while the u-shaped lower x-cover is being attached.

Finally, one cryptic issue - idler pulley friction. Thread belt through just the idler and through nothing else. Hold the two ends of belt to approximate belt path. Move belt ends back and forth feel in a flossing motion to feel whether the idler freely spins. I recommend testing the idlers this way because this problem bit me but was undetected by conventional checks. The idler felt like it could move ok when felt with my finger, but "flossing" a belt back and forth revealed it was actually dragging.

There is something creating extra resistance and making stealth mode not work. You will find it.

If you find absolutely nothing creating friction or binding - then yes, something stinks in that EINSY.

 

 

Only arrogant people would take offense to good solid advice. I have plenty of experience, but we can all use reminding once in a while of common sense things to check.  Often we look too far past the problem because the mental checklist gets shorter as we get older and/or more comfortable. 😉

I just rebuilt pretty much everything but the Y axis when I updated to your extruder, carriage and x-ends.  So I did a full rod/bearing tear down, clean and lube on X and Z.  The Y I waited until last night to deal with because I had done the Nyloc bed leveling, which is now undone as of last night and I'm back to the stock bed spacers.  the 7x7 leveling seems to have eliminated the need for this anyway.  

The X and Y without any belts or load move really freely.  The flick test would pass with flying colors. I'm quite anal about keeping as much pressure as possible off my wire looms as well (moving to silicone heater wire on Y helped that quite a bit). I will be cleaning and lubing the Y in the next few days.  I already have Misumi bearings with the G6 rods, but I'm changing them out for H5 rods and either a new set of Misumi bearings or the SKF bearings I bought quite some time ago, but couldn't find proper pressure fit holders (and didn't feel like designing any) so I went back to the Misumi.  I also just swapped out the idlers on both X and Y because a good supply on non disposable garbage ones in normal dimensions is difficult to come by.  I'm almost ready to switch to the LDO or E3D idlers, but the larger dimensions have me holding off.  

When I tear down the Y and replace the carriage with the new version I just printed I will do a full check on everything to see where I may have any resistance. 

I've also been looking through some of the printable belt tension gauges on Thingiverse.  They aren't measurement gauges, but comparative through the use of springs or weights.  This one, or the original it's remixed from, seems like the most promising: https://www.thingiverse.com/thing:2589577

At the very least, it will give me solid comparison tensions I can use for a sanity check. 

On the plus side, the new plated carriage printed in normal mode without incident and looks fantastic.  I'm printing a new set of bear bearing holders right now.  

Respondido : 23/06/2019 4:31 am
Evan me gusta
Dreide
(@dreide)
Trusted Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: AnatomicFlack

I've also been looking through some of the printable belt tension gauges on Thingiverse.  They aren't measurement gauges, but comparative through the use of springs or weights.  This one, or the original it's remixed from, seems like the most promising: https://www.thingiverse.com/thing:2589577

Somewhat off-topic, but remember that you already have a precise way to measure belt deflection - your Z axis. So all you need is an appropriate scale and a thingy to connect the scale to the belt. See, for example, how-exactly-to-increase-tension-on-x-and-y-belts/#post-118431.

Or use the much faster "pluck" method. See for example, use-free-span-frequency-hz-to-check-and-set-belt-tension/#post-73398.

Respondido : 24/06/2019 12:22 pm
AnatomicFlack me gusta
--
 --
(@)
Illustrious Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Belt tension is very easy if you have a good trigger pull gauge, or better quality fish/luggage scale.  You need fair accuracy at 8 ounces (205g).  Pull the belt up with 8 ounces (205g), look for 1/4" deflection (6mm).  This deflection yields about 8 pounds-force belt tension (35 N). Gates specs 2 lbf minimum, motor shaft axial load max is 11 lbf (1/2" from bearings on a "typical" stepper).

Respondido : 24/06/2019 4:58 pm
AnatomicFlack me gusta
--
 --
(@)
Illustrious Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

lol- in that last post of mine change axial to radial ... brain fade.

Respondido : 24/06/2019 5:09 pm
nikolaistolstoy@gmail.com
(@nikolaistolstoygmail-com)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

I feel like an idiot now. After chasing down the X axis homing issue for a long time I finally realized I left the zip tie fastener parts on the left side when I re-did my cable tie with the carriage all the way left. So it was hitting the EINSY and causing problems.

 

Now homing just works. No need to tweak homing parameters. I still have it set to a known good location after printing but I could even probably turn that off.  So for anyone else having problems definitely re-do the cable. It’s really a pretty quick job.

This was the only little thing about the mod that was a bit bothersome so now it’s jsut great. Thanks again guy-k2.

Esta publicación ha sido modificada el hace 5 years por [email protected]
Respondido : 24/06/2019 5:56 pm
CybrSage me gusta
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)
Posted by: Dreide
Posted by: AnatomicFlack

I've also been looking through some of the printable belt tension gauges on Thingiverse.  They aren't measurement gauges, but comparative through the use of springs or weights.  This one, or the original it's remixed from, seems like the most promising: https://www.thingiverse.com/thing:2589577

Somewhat off-topic, but remember that you already have a precise way to measure belt deflection - your Z axis. So all you need is an appropriate scale and a thingy to connect the scale to the belt. See, for example, how-exactly-to-increase-tension-on-x-and-y-belts/#post-118431.

Or use the much faster "pluck" method. See for example, use-free-span-frequency-hz-to-check-and-set-belt-tension/#post-73398.

Very interesting... I like the idea of the faster pluck method.  I found an app for my iPhone and shockingly the belt pluck freq is far more accurate and repeatable than I'd have expected from the phone's mic.  

I also happen to have a trigger gauge.  🙂  So I definitely have some ways to test now. 

The idea behind that one printed gauge is to eliminate the variable of loose belt length being a factor.  Using a fixed 150mm length means your tension values for each belt should be directly comparable.  It also eliminates some of the eyeballing for accurate deflection distance.  

I ended up rebuilding the Y axis last night.  Using a new tensioner system that uses an E3d idler and included a motor mount with bearing support (I used the half-way option - https://www.thingiverse.com/thing:3122625 ).  It, unfortunately, doesn't have integrated front or back stops (which I didn't love), so it's using the bearings once again, but I'm using the official bear bearing holders printed almost solid, so they aren't going anywhere.  I adjusted everything to make the Y friction as light as possible with as little bed wiggle possible.  While at it I also backed off the X carriage bearing covers, just a bit.  The overall tensions are a bit lower for sure, but I haven't put it through its paces yet printing.  If I can get a reliable print out in normal and stealth I will try to change the homing threshold back to 3 to see where I have to adjust the tension from the setting of 2.  

I have a new E3D 20T idler (LDO from PrintedSolid actually) on its way to throw in the modified X-end tensioner ( https://www.thingiverse.com/thing:3308226 ).  I found no smoking gun that I could easily point at causing unessasary friction on either axis though.  So my money is still on my EINSY being suspect.

Respondido : 24/06/2019 9:01 pm
AnatomicFlack
(@anatomicflack)
Eminent Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

Y is definitely improved.  (left is now, right was before rebuilding Y)

The other E3D idler for X can't get here soon enough.  

Respondido : 25/06/2019 2:21 am
CybrSage me gusta
CybrSage
(@cybrsage)
Honorable Member
RE: Stepper Motor Upgrades to Eliminate VFA's (Vertical Fine Artifacts)

OK Guys, wish me luck! I just replaced my X and Y with Moon 0.9 motors (and replaced the last bearings with properly greased ones as I disassembled the Y - did X and Y weeks ago).

This will be my first attempt at compiling my own firmware, using the very well written instructions provided earlier ( https://github.com/guykuo/Prusa-Firmware/blob/0.9-Degree-Stepper-Support/README.md ).

I have the two Moon steppers and the upgraded Bondtech extruder, as well as the MMU2s and MK3s upgrades.

(Could be delayed if work picks up suddenly.)

Respondido : 25/06/2019 3:47 pm
Página 16 / 53
Compartir: