If you can't screw the accelerometer to the bed you could print something to screw it to and then screw the accelerometer to that (without removing the printed part from the bed, of course)
That's good thinking and I'm stealing the idea. My bed screws are all hidden beneath magnetic sheets that I am loathe to pierce. Should get more accurate readings with my actual print surface mounted too.
I just discovered your channel some days ago.. and from the first moment I heard you talking I knew that I should stay here to learn so much from you. I'm really thankful.. thanks!!
I used a little hobby hot glue gun to mount the accelerometer to the glass bed of my SidewinderX1 and it worked great. I hung the bottom of the pins off the edge of the bed to allow for some leverage, making it a little easier to remove. Be careful when you remove the accelerometer so as not to damage the bed or the unit itself.
Of course I like the smash button! Great tips! Perhaps there should be a test where we vary the acceleration for external perimeters as well to see how it affects the print.
Max acceleration test layer high 0.2-0.25 seems to be made for 0.4 nozzles. So i Thing i should use bigger layer high with 0.8 nozzles maybe 50% of nozzle size.
These videos have been great. I have been applying your tips to an old i3 plus, and it has changed this machine into a completely different beast. I am working on printing the parts for a Voron Legacy. I was building a 1.6 when the legacy dropped so I just switched over. Thanks for the amazing content.
I did this not long ago, I found out max accel for my ender 3 could be easily 6000 which is unbelievable for bedslinger, almost zero ringing and tiny gaps start to appear at 6500. Anyway I set max accel to 4000 since I don't care that much about speed. Klipper should be must for 3d printers
I bet your prints aren't as detailed with Input shaper and that acceleration. Try printing a calibration cube. Is it completely flat with square corners and no ringing ?
Год назад
My Ender 3 Pro is at 5000 m/s2 on X axis and 3500 m/s2 on Y, I get a little bit of remaining ringing but the edges are very good, I let it slow down for small details and overhangs... oh, but I don't use the stock steppers as they get too hot around 2000 m/s2...
I really wish you put this video out a week ago when I was struggling with resonance compensation! :D But now I know I need to go back and re-tune pressure advance :) Thanks and don't stop with these great, short Klipper tips & tricks!!!
Could you print a mount for the accelerometer, leave it on the bed, mount the accelerometer into it and then run the test? In my mind that would give you the most accurate results as it would be specifically recording the forces the print was subject to whilst being printed.
Very informative video, I have learn a lot from your video. Input shaping is a great technique to improve print quality without sacrificing time. I also have developed my own version of input shaper from scratch.
Great video, cant wait to play with frequency domain in fdm. best mount for accelerometers, is often strong magnets. Often used on industrial rotating machinery. Simple, rigid and doesn't unscrew.
A magnet is definetly not the best way to mount an accelerometer!!. This is only done for quick measurements to get an indication. For proper measuring it's rigidly mounted to a machined surface or glued with a special ceramic glue. With a magnet you filter out your high frequencies.
Great video thanks for the help. My only critique is that at 0:14 you ended your sentence with a preposition. “…you should be running your printer at.” Next time try, “…at which you should be running your 3D printer.”
On my printer the gap didn't widen at all. There was some other "slight" problem more visible. I got really bad layer shift on Y axis. The bed is too heavy for high acceleration.
Good way to attach accelerometer to the bed is to just print the mount and not remove it from it. You can then just screw the adxl to the printed mount.
Wouldn't you expect to see different resonance data when you mount the accelerometer at different locations on the bed? For the hard mounting, you mounted it at the corner while for the tape, you picked the front/center. Also, the way you had the cables running put a load on the bed. Wouldn't you want to be careful to arrange them to minimize load in the direction of movement? Finally, doesn't the mass of the print and height above the bed strongly influence the vibration of the bed/part? Seems that calibrating with an empty bed won't give good results when you have a few hundred grams of plastic on there. Looks like you really need to look at the movement of the other axies when looking at bed vibration at least for a bed flinger.
Excellent tutorial. Only error is the starting height being 1250. It’s actually 1500. Common error because the formula isn’t commonly shared. Here’s the formula though: value = start + factor * ((floor(z_height / band) + .5) * band)
lol, literally today he changed the formulas on the resonance compensation page to be something clearer for people to use: TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=ACCEL START=1500 STEP_DELTA=500 STEP_HEIGHT=5
What layer height should I use if using .6 nozzle should it be .3 when I printed the test the gap was never closed or barely changed thru whole print even at 7500 accel
Level of Information condensation is superb. Liked. Subscribed. Regarding resonances: do you see substantial differences with feet/base setups? E.g one shall calibrate in final printer Position?
i have a question. so i built myself a Voron 2.4 and it runs beautifully. my only problem is that no matter what speed i specify in my slicing software, when i preview. the speed is always brought back to 35mms. what am i missing?
External stepper motor driver TMC2225 with Input Shaping GH Smart Shaper Due to the shortage of electronic components (Raspberry boards are practically unavailable) and to the required skills, Klipper is not yet a solution within everyone’s reach. GH Enterprise has therefore found a solution that allows you to add Input Shaping to 3D printers in a simple and immediate way, avoiding the use of external host cards such as RPI4. Just replace the X and Y axis drivers with the GH Smart Shaper drivers, add some configuration commands to your g-code and you are ready to print with input shaping.
This sounds like an add? Also that method has downsides like it only works on machines with independent xy motors. So not on deltas or corexy machines and also it uses the print and measure method which is not as accurate as the accelerometer method
Which value is right on the Klipper Resonance Compensation page, the START=1250 in the command or the 1500 at the description? If the starting acceleration is 1250, and each band is printed at +500, than the last one is only 6750, right? The acceleration isn't changed mid bad, only at the start of a band, or am I reading it wrong?
The Klipper behavior with TUNING_TOWER when BAND parameter is active is confusing, unfortunately. START says 1250, but the actual start is 1500, as the docs correctly suggest. You can also see it in the console as the print progresses. So the correct accelerations in that tower are 1500, 2000, 2500, and so forth.
Kudos for the overview. I wonder if you can speak to the use of using smoothing limits on the more aggressive input shaping algorithms? Also , you said use the highest peak with mzv? Or was it the recommended value but with mzv? Thanks?
It really depends on your use case and printer, id rather run tad lower accels and MZV and get crisp corners vs more aggressive input shapers at higher accels As for MZV i use peak
Hi Nero! I wanna ask the afterburner mod, what you have on your switchwire. Can you share the stl files? I doesn't fint dual bowden afterburner on thingiverse. Thanks Man 🍻🤘😁🤘
I have no ideas why but my ringing tower doesn’t have the void that we are supposed to measure at. I’ve done it twice, all of the Cura settings have been set and verified multiple times. And the command prompts have been entered in order and where copied directly from the linked website. Does anyone know why that may be?
I printed the ringing tower twice with the acceleration tuning. But I am not getting any ringing and alle these half circles are closed… any ideas? Tried discord but the don’t seem to know
whant to build a voron buut im couple months in 3d printing. should i start whit a SW or all in a Voron 2? have no problem to take some time to tinker here and there
@@CanuckCreator Thanks for the information. I finally found the link github.com/VoronDesign/Voron-2/blob/Voron2.4/STLs/TEST_PRINTS/Voron_Design_Cube_v7.stl
I've got a question regarding the fact that we have to retune linear/pressure advance everytime we change input shaping settings. I've also read that after changing acceleration values the pressure advance calibration also has to be redone. Is it true? If so, how does that work if we vary the acceleration for inner/outer perimeter in one print, doesn't this mess things up?
When i print at speed of 100 print start worping from corner and go really bad. When i print at speed of 60 part didn't worp. Should i add powerfull pot colling or anything else. Please suggest
Thanks for this explanation. On my side, the Tower printing is not a success. If I have understood, the speed should increase regularly due to TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=ACCEL START=1500 STEP_DELTA=500 STEP_HEIGHT=5. My speed is 45mm/s and is not changing... :| An idea? Note: I'm using a Creality Ender 3 S1 Pro + Creality Sonic Pad
Here's a question. I have tuned my switchwire esteps flow input shaper and tuned acceleration. I have one problem? When printing the first layer, the perimeters and brim come out perfect. Then when it comes to the solid infill it looks good and smooth but it looks like it's starting/stopping too soon and late. I haven't had a chance to less with it lately, but do you think this is related to pressure advance? I just printed a hollow single wall test cube and there walls are within .01 so I doubt my issue is mechanical related. Thanks Nero!
You HAVE to mount the accelerometer with tha Z axis pointing UP Please take a look at the datasheet before mounting the Accelerometer. For best result it has to be oriented correctly. XY accelerometer plane has to be parallel to XY printer plane. Figure 51. Noise vs. Output Data Rate PAGE 33
Detecting ringing in the mechanism with a sensitive instrument is fine, but until you have worked out the relationship between detected ringing and visible ringing in the prints, the data is next to useless! And once you have that relationship worked out, you would have determined your 'best acceleration'! Seriously, is maximum acceleration so variable that it needs real-time control? "Printing PLA today...need to bump the acceleration on the X axis +0.03%"... If my printer needed to have its acceleration adjusted 'on the fly', I'd be looking for a new printer.
I am following the video to the T but my print doesn't make a gap at any point in the vertical grove that he shows to measure with calipers. Does anyone have an idea what I am doing wrong?
This channel is a benchmark on how to explain anything 3D printer related. A+++ work.
Yeah i totally agree, This man may be saving both my Ender 5 plus's as ive proper messed them up jumping in before learning the ins and outs first.
I "like that smash button" 9:21
His acceleration was obviously too high and his brain overtook his mouth.
Gotta love Canadians though!
That last name though
If you can't screw the accelerometer to the bed you could print something to screw it to and then screw the accelerometer to that (without removing the printed part from the bed, of course)
Maybe hot glue could work as well. Put masking tape on the bed, and glue the accelerometer to the tape. The tape flat on the bed shouldn’t “wiggle”?
That's good thinking and I'm stealing the idea. My bed screws are all hidden beneath magnetic sheets that I am loathe to pierce. Should get more accurate readings with my actual print surface mounted too.
I used tape
I just discovered your channel some days ago.. and from the first moment I heard you talking I knew that I should stay here to learn so much from you. I'm really thankful.. thanks!!
I used a little hobby hot glue gun to mount the accelerometer to the glass bed of my SidewinderX1 and it worked great. I hung the bottom of the pins off the edge of the bed to allow for some leverage, making it a little easier to remove. Be careful when you remove the accelerometer so as not to damage the bed or the unit itself.
Of course I like the smash button! Great tips! Perhaps there should be a test where we vary the acceleration for external perimeters as well to see how it affects the print.
watched the stream, but this is a great recap and totally worth watching.
Just got my switchwire up and running. Excited to get all thebfine tuning out the way and enjoy the faster better printing
Max acceleration test layer high 0.2-0.25 seems to be made for 0.4 nozzles. So i Thing i should use bigger layer high with 0.8 nozzles maybe 50% of nozzle size.
just print the accel mount for the bed and keep the bed hot so it doesnt come loose, then screw it down, and you have a good steady mount
These videos have been great. I have been applying your tips to an old i3 plus, and it has changed this machine into a completely different beast. I am working on printing the parts for a Voron Legacy. I was building a 1.6 when the legacy dropped so I just switched over. Thanks for the amazing content.
Great job Nero, really good flow and info in this video.
From start to finish, what is the correct order to tune?
We have:
Extruder steps/mm
Input shaper
Pressure advance
Acceleration
Pretty much
@@CanuckCreator You should make a full playlist of untuned -> tuned printer in order.
@@LanceThumping good idea
I did this not long ago, I found out max accel for my ender 3 could be easily 6000 which is unbelievable for bedslinger, almost zero ringing and tiny gaps start to appear at 6500. Anyway I set max accel to 4000 since I don't care that much about speed. Klipper should be must for 3d printers
I bet your prints aren't as detailed with Input shaper and that acceleration. Try printing a calibration cube. Is it completely flat with square corners and no ringing ?
My Ender 3 Pro is at 5000 m/s2 on X axis and 3500 m/s2 on Y, I get a little bit of remaining ringing but the edges are very good, I let it slow down for small details and overhangs... oh, but I don't use the stock steppers as they get too hot around 2000 m/s2...
Hello, what steppers are you using?
Great video. Can't wait to try it out with my sidewinder X1.
Edit: You are the first person I ever join. It's not much but I hope it helps
Great recap! Really appreciate your knowledge, presentations, and channel! Thanks!
It been 3 years, and Im printing for about 50-60 mm/s acc. Im gonna try Klipper for my old ender 3.
I really wish you put this video out a week ago when I was struggling with resonance compensation! :D
But now I know I need to go back and re-tune pressure advance :)
Thanks and don't stop with these great, short Klipper tips & tricks!!!
what setting is minimal layer speed in prusa slicer, i think you glossed over that one?
thank you for putting this recap video.
Could you print a mount for the accelerometer, leave it on the bed, mount the accelerometer into it and then run the test? In my mind that would give you the most accurate results as it would be specifically recording the forces the print was subject to whilst being printed.
Yup. And thats been tried and known to work
this is a very useful and nice video! specially the tape test. it's also short which rocks! Thanks!
What Would I do without your videos :)
What if the gap never opens? Will it be safe to run the printer at 7000 accel?
Yes
Nice Guide
This video explained so much. Thank you.
I should have watched this as I was clueless what I'm going to do after finding out resonances from input shaper measurement
Very informative video, I have learn a lot from your video. Input shaping is a great technique to improve print quality without sacrificing time. I also have developed my own version of input shaper from scratch.
Nice video I like the content! Keep it coming! Greeting from Belgium
I love this channel❤️
Do you control acceleration using your slicer? I thought the docs recommended disabling all acceleration control...
It needs to be disabled for the calibration print, but afterwards you can adjust it.
@@CanuckCreator Do you run jerk settings too?
Great video as always 👍🙂
You have made my way to Klipper!
Great video, cant wait to play with frequency domain in fdm. best mount for accelerometers, is often strong magnets. Often used on industrial rotating machinery. Simple, rigid and doesn't unscrew.
A magnet is definetly not the best way to mount an accelerometer!!. This is only done for quick measurements to get an indication. For proper measuring it's rigidly mounted to a machined surface or glued with a special ceramic glue.
With a magnet you filter out your high frequencies.
Thanks again, this videos are helpful.
Great video thanks for the help. My only critique is that at 0:14 you ended your sentence with a preposition. “…you should be running your printer at.”
Next time try, “…at which you should be running your 3D printer.”
On my printer the gap didn't widen at all.
There was some other "slight" problem more visible. I got really bad layer shift on Y axis. The bed is too heavy for high acceleration.
Have the same doesn't widen also
Good way to attach accelerometer to the bed is to just print the mount and not remove it from it. You can then just screw the adxl to the printed mount.
Wouldn't you expect to see different resonance data when you mount the accelerometer at different locations on the bed? For the hard mounting, you mounted it at the corner while for the tape, you picked the front/center. Also, the way you had the cables running put a load on the bed. Wouldn't you want to be careful to arrange them to minimize load in the direction of movement? Finally, doesn't the mass of the print and height above the bed strongly influence the vibration of the bed/part? Seems that calibrating with an empty bed won't give good results when you have a few hundred grams of plastic on there. Looks like you really need to look at the movement of the other axies when looking at bed vibration at least for a bed flinger.
Excellent tutorial. Only error is the starting height being 1250. It’s actually 1500. Common error because the formula isn’t commonly shared.
Here’s the formula though:
value = start + factor * ((floor(z_height / band) + .5) * band)
lol, literally today he changed the formulas on the resonance compensation page to be something clearer for people to use:
TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=ACCEL START=1500 STEP_DELTA=500 STEP_HEIGHT=5
Great vid. Looking to take my accel from a glacial sub-600 to 3000+! 🚀🚀🚀🐇🐇🐇
What layer height should I use if using .6 nozzle should it be .3 when I printed the test the gap was never closed or barely changed thru whole print even at 7500 accel
Level of Information condensation is superb. Liked. Subscribed.
Regarding resonances: do you see substantial differences with feet/base setups? E.g one shall calibrate in final printer Position?
Thanks! 👍
Can you use magnets to attach the accelerometer? Or would they get thrown off?
Great practical explanation, but those Cubes show a LOT of ringing, can't it be helped a bit more?
i have a question. so i built myself a Voron 2.4 and it runs beautifully. my only problem is that no matter what speed i specify in my slicing software, when i preview. the speed is always brought back to 35mms. what am i missing?
What about if it recommends ZV? Should you still go MZV?
External stepper motor driver TMC2225 with Input Shaping
GH Smart Shaper
Due to the shortage of electronic components (Raspberry boards are practically unavailable) and to the required skills, Klipper is not yet a solution within everyone’s reach.
GH Enterprise has therefore found a solution that allows you to add Input Shaping to 3D printers in a simple and immediate way, avoiding the use of external host cards such as RPI4.
Just replace the X and Y axis drivers with the GH Smart Shaper drivers, add some configuration commands to your g-code and you are ready to print with input shaping.
This sounds like an add?
Also that method has downsides like it only works on machines with independent xy motors. So not on deltas or corexy machines and also it uses the print and measure method which is not as accurate as the accelerometer method
@@CanuckCreator In a couple of days it will also work on Corexy, we are completing the documentation, on Delta it will not be possible
the smart shaper now also works with corexy printers
Which value is right on the Klipper Resonance Compensation page, the START=1250 in the command or the 1500 at the description?
If the starting acceleration is 1250, and each band is printed at +500, than the last one is only 6750, right?
The acceleration isn't changed mid bad, only at the start of a band, or am I reading it wrong?
Its easiest and simplest to go 500 per full band and use the value for the last "fully complete" band before the gap starts to open.
Thanks, now at least that part is clear. It's just the Klipper RC description that is feels controversial. (1250 != 1500)
Oh, now that I didn't expect. Thank you very much!
The Klipper behavior with TUNING_TOWER when BAND parameter is active is confusing, unfortunately. START says 1250, but the actual start is 1500, as the docs correctly suggest. You can also see it in the console as the print progresses. So the correct accelerations in that tower are 1500, 2000, 2500, and so forth.
Thanks, will look at the console when I try it out.
Kudos for the overview. I wonder if you can speak to the use of using smoothing limits on the more aggressive input shaping algorithms? Also , you said use the highest peak with mzv? Or was it the recommended value but with mzv? Thanks?
It really depends on your use case and printer, id rather run tad lower accels and MZV and get crisp corners vs more aggressive input shapers at higher accels
As for MZV i use peak
@@CanuckCreator Can you say more about the value of limiting smoothing with more aggressive input shaping algorithms?
Hi Nero! I wanna ask the afterburner mod, what you have on your switchwire. Can you share the stl files? I doesn't fint dual bowden afterburner on thingiverse. Thanks Man 🍻🤘😁🤘
input shaper is possible with an a4988 driver?
I have no ideas why but my ringing tower doesn’t have the void that we are supposed to measure at. I’ve done it twice, all of the Cura settings have been set and verified multiple times. And the command prompts have been entered in order and where copied directly from the linked website. Does anyone know why that may be?
Same here. Mine is the same from bottom to top.
I liked the smash button just like you said!! (9:20)
I would really like a bit more detailed explanation on choosing the right value for input shaper.
I suppose "it depends". On your printer, on the models that you print, on how much hurry you have.
gonna try hot glue for mounting to the bed.
Do you still recommend MZV over the recommended EI shaper on dewlta printers?
I printed the ringing tower twice with the acceleration tuning. But I am not getting any ringing and alle these half circles are closed… any ideas? Tried discord but the don’t seem to know
whant to build a voron buut im couple months in 3d printing. should i start whit a SW or all in a Voron 2? have no problem to take some time to tinker here and there
What extruder do you use on this printer?
Could you provide a link to the test cube you use in many of your videos? I have looked at many of them and can''t find where to get it. Thanks
Its on the voron github under test prints in the v2 and othrr printer repos
@@CanuckCreator Thanks for the information. I finally found the link github.com/VoronDesign/Voron-2/blob/Voron2.4/STLs/TEST_PRINTS/Voron_Design_Cube_v7.stl
EVERY FUKKKIN TIME IT GETS ME. holy shit.
"like that smash button", godamnit you caught me off guard. now I have to press like.
I've got a question regarding the fact that we have to retune linear/pressure advance everytime we change input shaping settings. I've also read that after changing acceleration values the pressure advance calibration also has to be redone. Is it true? If so, how does that work if we vary the acceleration for inner/outer perimeter in one print, doesn't this mess things up?
Correct, IS will affect PA
Usually you just have to rerun PA after doing a new IS tune and thats it
9:20 indeed I will like that smash button
When i print at speed of 100 print start worping from corner and go really bad. When i print at speed of 60 part didn't worp. Should i add powerfull pot colling or anything else. Please suggest
Thanks for this explanation.
On my side, the Tower printing is not a success.
If I have understood, the speed should increase regularly due to TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=ACCEL START=1500 STEP_DELTA=500 STEP_HEIGHT=5.
My speed is 45mm/s and is not changing... :|
An idea?
Note: I'm using a Creality Ender 3 S1 Pro + Creality Sonic Pad
Why are the Axis not in line with the printer Axis
Here's a question. I have tuned my switchwire esteps flow input shaper and tuned acceleration. I have one problem? When printing the first layer, the perimeters and brim come out perfect. Then when it comes to the solid infill it looks good and smooth but it looks like it's starting/stopping too soon and late. I haven't had a chance to less with it lately, but do you think this is related to pressure advance? I just printed a hollow single wall test cube and there walls are within .01 so I doubt my issue is mechanical related. Thanks Nero!
May be a slicer setting at play such as coasting
When I test this the band is never closed 🤷🏼♂️ even when starting not at 1250 but on 250
But have no idea what I do wrong
same mine started from bottom almost closed and stayed the same most of print
You HAVE to mount the accelerometer with tha Z axis pointing UP
Please take a look at the datasheet before mounting the Accelerometer.
For best result it has to be oriented correctly.
XY accelerometer plane has to be parallel to XY printer plane.
Figure 51. Noise vs. Output Data Rate PAGE 33
Now, how to find the speed limit for printing? There is any method with klipper?
That will depend alot on the flow rate of your hotend, the plastic being used and desired finish quality. Theres no set test for determining it.
Good job and thank you! Lol 9:21
does anyone know the name or have a link for that XYZ cube? I see it in a lot of videos but no one links or shows any reference to it.
Thanks
bueler? bueler?
Thanks bud!
Thanks! :)
im a bit confused on settings name in prusa vs. Cura, can anyone help a poor soul like me what settings I should set to what in cura?
where can i get the voron calibration cube stl file?
its on the voron github, check under the v2 folder and test prints
@@CanuckCreator thankyou , i will check it
Where do I go to change these values?
square_corner_velocity
ISET_VELOCITY_LIMIT ACCEL_TO_DECEL=7000
SET_PRESSURE_ADVANCE ADVANCE=0
gotta go fast!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Like some other people here mine doesn't widen even at 7k accel. idk what im doing something wrong
im using prusa slicer and i cant for the life of me get this print to work and have it increce speed
does not start with 1250, starts with 1500!
I liked the smash button :)
1:23 800mm/s IS fast! 😂
Acceleration is measured in mm per second per second NOT mm per second which is speed.
So I did this, and my curved gaps are all full. I dont see any big gaps like in your prints.
Make sure you like that smash button
cant get the thing to stay stuck to the bed for the life of me after the first layer the speed picks up and rips the thing right off
wow idk how but somehow my cura was trying to slice the model with a 1.0 mm nozzle changed it back to .4 and its magically printing perfect again
Which slicer do you use in order to control the acceleration for outer perimeters separately?
Cura and prusaslicer / superslicer have this function
Like that smash button
When doing professional vibrations tests I have seen them using bees wax to fix the sensors. Its like a piece of gum.
Do you really have to waste that much filament?
I suffer from premature articulation so I almost made an inappropriate joke there..
Only Benchies
#klipper #calibration #voron
Detecting ringing in the mechanism with a sensitive instrument is fine, but until you have worked out the relationship between detected ringing and visible ringing in the prints, the data is next to useless!
And once you have that relationship worked out, you would have determined your 'best acceleration'!
Seriously, is maximum acceleration so variable that it needs real-time control?
"Printing PLA today...need to bump the acceleration on the X axis +0.03%"...
If my printer needed to have its acceleration adjusted 'on the fly', I'd be looking for a new printer.
I am following the video to the T but my print doesn't make a gap at any point in the vertical grove that he shows to measure with calipers. Does anyone have an idea what I am doing wrong?