S5 Ringing

[arj3090 2]

I have some bad ringing in my S5 that I can't seem to improve. I have tried different speeds, accelerations, and jerks. When I get the ringing to fade, my corners bulge.

 

I tried to take a picture to show the ringing and bulging corners, but the features are hard to see in the picture.

 

I recently replaced the head assembly after a bad crash. I did not notice the ringing prior, but it could have been there, and I did not scrutinize the prints as much as now.

 

Any advice on how to eliminate ringing while keeping sharp corners? I looked for an input shaping function but could not find anything saying the S5 supports it.

RingTest.3mf

[gr5 2,192]

That's the tradeoff.  The default S5 acceleration and jerk settings fix the ringing but then you get those bulging corners.  You can fix one or the other.  That's why there is the "engineering" profile which has lots of ringing but parts are much more accurate when measured with a micrometer.  And the regular profile which has nicer looking parts (not much ringing).

 

the one way to fix both issues is to print slower.  35mm/sec is pretty good but you can even go as slow as 25mm/sec.

 

Personally, for what I do (print mechanical parts for customers), no one cares about the ringing but accuracy is very important so I use the engineering profiles.

[gr5 2,192]

By the way, I think the way you remove ringing is really cool.  So UM printers have high accel.  I think S5 has 5000m/s/s.  So if printing speed is 150mm/sec it takes 30ms to get up to full speed.  Printing at 70mm/sec takes about half that or 14ms.  It's actually faster as the jerk allows you to start at around 14mm/sec before you even start accelerating.  At these accelerations you are 150mm/sec in 2.25mm of distance (distance = 1/2 * accel * t^2).

 

Whereas the ringing frequency is usually slower?  I think?  That is what people tell me.  This greatly simplifies the analysis.

 

So you first calculate the period of the ringing.  You can do that by measuring the distance between the bumps - do it where you know it is at full speed (2.25mm from the corner if at 5m/s/s accel and 150mm/sec printing speed - closer to the corner if printing slower - farther from corner if accel is slower).

 

Then if the period is say, 30ms, and accel lasting 14ms you can see that all the acceleration is going into the first half of the oscillating cycle.  Which creates the most possible ringing.  But you want the acceleration to last exactly 30ms if that is the ringing period.  So you pick an acceleration that means it will take 30ms to get up to full speed.  That will result in zero ringing.  speedup time is speed/accel.

[arj3090 2]

Thank you for the response and advice. I am experimenting with the technique you described and will report back.

[arj3090 2]

Update - I ran a part at 70 mm/s with 5000 mm/s/s acceleration. I then measured the spacing of the ringing to be 1.5mm

 

Based on my calculations, the period would be 21.4ms. At a speed of 70 mm/s, the acceleration to reach 70 mm/s in 21.4ms would be 3266 mm/s/s.

 

Using 70mm/s at 3266 mm/s/s the result is less pronounced ringing, but it is still there.

 

Just a little history of what lead to this attempt to go faster on the UM S5..... We recently aquired a Bambu X1 Carbon that now sits next to the S5. The Bambu runs parts in about 1/4 of the time with as good or better results. I decided to try to get the S5 to be a little faster in order to at least be useful.

[gr5 2,192]

What is the jerk speed?   I guess you have to lower it?

 

That initial jerk puts extra acceleration into the first half of the vibration - the first .75mm of travel.  So lower the jerk speed as well I guess?  I don't know how to deal with that - you want a lower accel in the first .75mm and then much higher accel in the second half so the speed increase is equal in both halves of a ring: 35mm/sec in the first .75mm and another 35mm/sec (to reach 70mm/sec) in the second half.  But that would require adding an extra point along the travel (where it increases acceleration .75mm along the path).  I don't know if Cura does that for you or it just lowers the jerk a lot (say to 10mm/sec?).  I don't really know.

 

I guess if cura did it for you they wouldn't have to lower the acceleration at all.  They'd just set the speed for the first .75mm to 35mm/sec and leave jerk and accel high.  Then the remaining travel would be to 70mm/sec.  That would force the total acceleration in the first and second .75mm to be equal which would cancel out the ringing.

 

Well the S5 has a big heavy print head with heavy 8mm rods passing through it which add more inertia.  And the belts are the same size as the smaller printers like the UM2.  So it just has a lot more ringing.  UM engineers were used to bowden printers with super light print heads but when they went dual nozzle they ended up with a heavier head and they did some kind of accel/jerk hack but I don't know how well that works as I don't think it breaks up the move into two segments.  I don't know what the heck they did.  Probably just lowered the jerk a lot and did the acceleration calculation that we just did.

 

I'm curious what accel and jerk settings are inserted into the gcode when you enable jerk and accel control.

 

[arj3090 2]

The jerk is set to the lowest setting of 20. From the G Code file:

 

M204 S4083
M205 X20 Y20

 

 

RingingTest.gcode