#Broke my unicorn on X! How do I fix it? (Resolved)

90 messages · Page 1 of 1 (latest)

slate verge
#

So I’ve recently done some upgrades to the printer, mostly “bling” but also to increase stiffness.

Setup:

  1. V2.4 350
  2. Chaotic labs cnc parts on the gantry throughout
  3. Stealthburner in ASA CF
  4. G2E, both in ASA CF
  5. Filametrix
  6. Cartographer + cartographer CNC mount
  7. ERCF + filamentalist enclosure sitting on top of printer
  8. Measurements taken with the SB EBB2209 board accelerometer in all cases
  9. Belt tension at 180hz with double shear. 120 without, but the issue is present in either 180 or 120hz setup

Changes done:

  1. Bearings changed to hybrid ceramic from west 3D as the old ones were getting noisy at speed
  2. Replaced bottom deck panel with ACM instead of ABS
  3. Installed double shear mod and ramped up belt tensions. This was installed in “third time” and the issue with X was present before this.
IMG_5019.jpg IMG_5020.jpg
#

While Y has improved (especially following the increase in belt tension from the double shear mod), allowing for roughly 1k higher accels with the same vibration magnitude from the double shear mod, the X axis appears to have deteriorated.

However upon closer inspection I’m noticing that the X amplitude is significantly lower, maybe as a result of the increased X stiffness from the better bearings (the old ones were not great)? And probably that is exposing vibrations that would otherwise not be visible? 🤔

Or did I mess up the install somewhere?

Please note that the X chart was almost identical with double shear installed or removed.

I’m at a loss as I can’t explain why would a better set of bearings or the acm bottom deck panel cause this?

I’ve triple checked the gantry and there is no belt rubbing or binding at the bearings anywhere - it’s physically impossible to bind the bearings with the CL CNC kit as the bearing spacers enforce a set distance and no matter how much you torque them they cannot bind.

#

Below is without the double shear mod. Same issue on X. also tried with the filamentalist unmounted from the top of the printer - same results.

IMG_5025.png
#

Please note that the usual suspects are probably not to blame - the filamentalist was there in the previous tests, as was the Bowden routing and umbilical routing.

dusk spire
dusk spire
dusk spire
slate verge
dusk spire
slate verge
#

Please note there is a stealthmax in the rear enforcing the Bowden tube to come in through the vent like this.

IMG_5029.jpg IMG_5027.jpg
dusk spire
slate verge
dusk spire
slate verge
dusk spire
slate verge
#

Cartographer measurements

IMG_5031.png IMG_5032.png IMG_5030.png
slate verge
#

@chilly willow sorry to ping directly but I’d love your expertise here 🙂 busting my head with the above today and no matter what I try results are the same 🙏🏻 this also means my unicorn status badge is revoked 😂

chilly willow
#

lol

#

did you excite at freq for peak 2 in x?

#

Also did you try to bring you tension back down to spec?

slate verge
slate verge
chilly willow
#

did you have the cnc xy joints before or are these new

slate verge
#

Toolhead is tight and quite rigid left to right. No movement whatsoever. Front to back there is a bit of flex due to the cartographer CNC mount hence the “nodding” when measuring with the cartographer adxl.

chilly willow
#

have you checked belt paht?

slate verge
# chilly willow did you have the cnc xy joints before or are these new

My test journey this week:

  1. Baseline - ASA CF parts for toolhead and XY joints. Everything was fine.
  2. Replace bearings and bottom deck panel in one sitting - chart above emerged with the second low amplitude peak on X.
  3. Replaced ASA CF XY joints with the CL aluminum joints - no change.
  4. Installed double shear mod and cranked tension - Y became better. X unchanged
slate verge
chilly willow
#

looks like it is a bearing issue then

#

which bearings did you replace? all of them

slate verge
#

These were some expensive bearings …. 🤦🏻‍♂️

chilly willow
#

did you also change out your toothed idelers

slate verge
#

Nope these are unchanged. They were gates genuine from E3D. Wanted to replace them but they were out of stock

#

Maybe worth swapping those out too you recon?

chilly willow
#

the bearings in the toothed idlers are more fragile than the bearings in the bearings.
if they were installed at the same time, I would think they would also be going bad

slate verge
#

Mmmm you’ve got a point…

#

I’ll try and get hold of some thank you!!

chilly willow
#

If you have spare regular bearings what you can do is replace the toothed bearings with a bearing stack (for testing purposes) and see if it improves

#

likely to get vfa in prints.

#

I haven't done this myself, so I do not know if the teeth will throw off the measurement

slate verge
#

Funny thing is print quality is excellent. I see absolutely nothing there. Maybe I’m chasing my tail too, but it’s just puzzling. What I am noticing is a significant reduction in X amplitude which maybe points to something binding ever so slightly?

IMG_4853.png IMG_5025.png IMG_5022.png
#

Before - After with stock tension - After with increased tension

chilly willow
#

based upon what you said you did and where the problem arose I woudl be be looking to

#

the xy joints
the toolhead
belt routing

#

this is because all three of these things you would have had to adjust to replace the bearings

#

potentially also racking

slate verge
slate verge
#

Alright, parts ordered. Belts + new toothed idlers. Should be here in a couple of days. I'll feedback when installed 🙂 thank you!!

slate verge
#

@chilly willow a bit better now. Took apart the toolhead. Re tightened all extruder screws and reseated the belts on the toolhead mount. I think it’s definitely coming from the toolhead as taking that part apart and putting it together again resulted in some improvement?

I think I need new belts too. They’ve had quite a few hours on them (1.5-2k or so) so they may start getting compromised rigidity wise.

IMG_5043.png IMG_5044.png
#

Before and after

IMG_5025.png IMG_5043.png
chilly willow
#

I agree still something going on.

#

Ping me when you installed the new toothed idlers.

slate verge
#

@chilly willow unicorn is back and with a vengance (the higher tensions have materially helped with the Y accelerations - up 1-1.5k from before). The second minor X peak is from the toolhead, as the G2E is heavy and vibrates ever so slightly at that frequency. But it thankfully doesnt spoil the charts now.

inputshaper_20241205_123447_X.png inputshaper_20241205_123447_Y.png
#

So what did I do...

#

1. Replaced belt and toothed idlers on the gantry last night. I was kind of hoping this would have fixed it, but nope! Absolutely 0 change. Maybe even ever so slightly worse due to the belts not being run in yet. You can imagine my face when i saw those results...

2. Then last night and until 2 in the morning I went down a rabbit hole - Removed the extruder, replaced its X carriage mount with a new one just in case the old one was worn and was vibrating. Replaced the revo mount with a new one too. Again, absolutely no change whatsoever.

#

3. This morning I decided to re-inspect the motion system. I was battling slightly off belt tension with the new belts and thought I had not cut them to exact same size. But that was not the case. I took apart the panels, re-squared the gantry and validated with my machinists square. Then when I was inspecting the Z pins for alignment I noticed that the bottom belt clamp was not tight on any of the 4 corners.

#

That metal piece clamps the belt, the front belt tensioner and also holds the screw that drops in the G5C mounts underneath it. So apparently the gantry was not properly coupled to the frame as that bolt could (theoretically) move ever so slightly.

#

So re-aligned everything, tightened everything and the above is the result! So now we know - always always double check your screws!

#

This printer has no right to produce such graphs 😄 Its loaded with 6kg of filament on top of it, has a mounted ERCF on plastic extrusion mounts on its rear and 1kg of carbon in the stealthmax hanging behind it 😄

1.jpeg 6.jpeg 5.jpeg 4.jpeg 3.jpeg 2.jpeg 1.jpeg
chilly willow
#

@slate verge thank you for the detailed solution.
I would not have gone there, and figured out it was the z screws.
I have updated my resonance testing pdf on github

#

Can you print a voron cube please and post the results

slate verge
slate verge
#

Pretty clean! It’s the most revealing filament I have - it shows everything 😂

#

Running a full one at 5.4k too

chilly willow
#

slight ringing starting to happen on "x", could be PA and pressure advance smooth time

#

silk does show everything. It is the TPU in the mixture

slate verge
#

@chilly willow Also stepper noise as it’s accelerating through its resonance zone (70-120mm/sec). printed at 140mm/sec so right after corners you’ll get very slight artefacting I’d guess.

#

Print below at 5.4k with ei on both axis. My PA is too high for that acceleration for sure

IMG_5078.jpg IMG_5077.jpg IMG_5076.jpg IMG_5075.jpg
#

Pretty happy with this - the worst case scenario and I need to use just the right light to see it.

chilly willow
#

Looks good.

slate verge
#

Thank you for the help :)!!

chilly willow
slate verge
#

@chilly willow promised I'll post some graphs with the new input shaper routine from klipper. Please see below. First picture is the previous version, second picture is the latest version with the sweeping motion, third picture is S&T graphs (on the old version)

#

Y Axis

#

In my case the changes I've noticed:

#
  1. The graphs have shifted ever so slightly to a lower frequency, but this could also be run to run variance.
#
  1. The Y axis graph has the secondary peak from the toolhead/umbilical wobbling ever so slightly more pronounced (but the overall amplitude is slightly less too)
#

So if your printer is mechanically sound, I cant see a real difference

chilly willow
#

Slightly lower accel in y, but nothing g yo write home about. I agree not much of a difference

dusk spire
slate verge