Belt tension adjustment
Input shaping is obviously a key ingredient to making the Core One fast. The resonance frequencies which the input shaper aims to compensate will depend critically on the belt tension (among other factors). With the long belts in a Core XY setup, which are each guided via several idlers, I would assume that belts will be more prone to losing tension over time than in the classical bed-slinger setups.
Prusa has decided against installing an acceleration sensor (which would allow automated measurement of the resonance frequencies) as a standard feature. Input shaping relies on factory presets, and hence on constant belt tension throughout the operating life of the printer. I assume this means one should check the belt tension regularly, and adjust it when needed.
In the 1.0 version of the Core One's user manual, all I can find is the suggestion to print a circular shape to check for loose belts -- that seems like a very coarse way to check belt tension, which only detects major slack in the belt. What is the recommended procedure for checking belt tension? How often should it be done?
Also, how can the tension be adjusted? The manual just refers to the online help on that topic, which however does not seem to exist yet. Many thanks for your insights!
RE: Belt tension adjustment
And so it begins…
RE: Belt tension adjustment
Well, I guess we will see proper instructions and target values for the tension when the assembly instructions for the kit are published.
RE: Belt tension adjustment
The belts are tensioned from the front left and right with an inbus.
I think that here, as with all Prusa, the belt tuner is used.
RE: Belt tension adjustment
The belts are tensioned from the front left and right with an inbus.
I think that here, as with all Prusa, the belt tuner is used.
That makes sense, thank you. It's strange that the handbook is not more specific here (and even misleading, suggesting to "print something round" to check belt tension). It seems like a slightly rushed conversion from the MK4 series handbook -- I also came across the old MK4 build volume somewhere.
RE: Belt tension adjustment
I just wanted to add that the accelerometer and calibration procedure that works with the MK4 also works with the CORE One. I wish it were included in the toolhead but I guess that would've required a new Love Board at the very least. Then again, the CORE could've gotten the new board, and upgrades could've used the MK4 one with the option for a paid upgrade to one with the integrated accelerometer....
Input shaping is obviously a key ingredient to making the Core One fast. The resonance frequencies which the input shaper aims to compensate will depend critically on the belt tension (among other factors). With the long belts in a Core XY setup, which are each guided via several idlers, I would assume that belts will be more prone to losing tension over time than in the classical bed-slinger setups.
Prusa has decided against installing an acceleration sensor (which would allow automated measurement of the resonance frequencies) as a standard feature. Input shaping relies on factory presets, and hence on constant belt tension throughout the operating life of the printer. I assume this means one should check the belt tension regularly, and adjust it when needed.
In the 1.0 version of the Core One's user manual, all I can find is the suggestion to print a circular shape to check for loose belts -- that seems like a very coarse way to check belt tension, which only detects major slack in the belt. What is the recommended procedure for checking belt tension? How often should it be done?
Also, how can the tension be adjusted? The manual just refers to the online help on that topic, which however does not seem to exist yet. Many thanks for your insights!
RE: Belt tension adjustment
I think that Prusa is of the opinion that it is not absolutely necessary to use the acceleration sensor
Have you tested it on the MK4S?
It comes out with approximately the same values as Prusa uses.
The difference is probably less than the measuring tolerance of the sensor.
RE:
To me it seems like a brave decision on Prusa's part to omit the accelerometer and rely on long-term stability and unit-to-unit reproducibility of the belt tension. It would be a pity if that ends up giving the Core One a mixed reputation regarding print quality. They obviously have gained experience with the MK4S, but that machine has a very different drive geometry -- will the much longer belts in the Core One be as well-behaved?
But maybe I'm overthinking this. The Input Shaper might still limit the acceleration to values which keep the system a fair margin away from its mechanical resonances. In which case, let's say, a 10% shift of the resonance frequency might still be tolerable?
RE:
I noticed the same thing on the MK4S. However on the CORE One I got different values and Input Shaping methods than the stock values. Of course it's possible this feature needs more testing and firmware tweaking, as there is currently zero updated information about it yet on their site. Or that I missed some important required step.
I think that Prusa is of the opinion that it is not absolutely necessary to use the acceleration sensor
Have you tested it on the MK4S?
It comes out with approximately the same values as Prusa uses.
The difference is probably less than the measuring tolerance of the sensor.
RE: Belt tension adjustment
The difference is probably less than the measuring tolerance of the sensor.
The measuring tolerance should be very small. One scans the frequency and measures where the vibration amplitude becomes largest. (Best determined from a fit to all data, after completion of the scan.) The sensor's amplitude measurement is a non-critical relative measure only, and the drive frequencies during the scan are known very accurately since they are derived from the xBuddy board's quartz timebase.
RE: Belt tension adjustment
Are you sure that all sensors measure the same thing? They are relatively cheap parts and not professional equipment.
But it is also not the case that if you are a little off Input Shaper no longer works. You may then only have 92% but still much better than without.
RE:
Are you sure that all sensors measure the same thing? They are relatively cheap parts and not professional equipment.
No, I did not claim that all sensors measure the same, quantitatively. What I tried to explain was that they don't have to: They measure acceleration, and all they need to report is "I see more acceleration (i.e. a higher vibration amplitude) now than I did a second ago" or "I see less now than a second ago". The critical metric is the excitation frequency at which this occurs, and that is known very precisely since it is driven by the stepper motors, based on a quartz timebase.
Knowing the amplitude of the vibration (determined by the Q factor of the resonance) is also important for optimal input shaping, and measuring that would indeed require a quantitative accelerometer. But I don't see how the Q factor would vary much over time or between printers -- it is determined by the moving mass and the dampening forces. So using default values for that should work well. In contrast, the frequency depends on the belt tension and is expected to vary over time and temperature -- hence should benefit from calibration.
But as I mentioned, I can imagine that knowing the exact resonance frequency is not that critical. That will depend on how aggressively you want to drive the printer, i.e. how close to its resonance frequencies you want to operate it. I assume Prusa has decided to apply a margin there, sacrificing some speed in favor of decent operation even with sub-optimal input shaping parameters.
RE: Belt tension adjustment
All theory, I see it practically.
If the print is good without the sensor and has no artifacts, it won't be any better, or only marginally better, with it. So why install one?
If you want to know exactly, get one, if you don't, save the money and buy a roll of filament.
I admit that I am also the one who bought the sensor.
But many people won't care as long as the prints look good.
Prusa is not known for pushing the limits in terms of speed.
Then things might look a little different.
The right frequency is probably more critical with a Corexy.
Let's just wait and see.
RE:
All theory, I see it practically.
If the print is good without the sensor and has no artifacts, it won't be any better, or only marginally better, with it. So why install one?[...]
Let's just wait and see.
My question did start out as a rather practical one: When I build the Core One, how do I set the "right" belt tension -- i.e. the one which matches what Prusa set when they determined the stadard parameters for the input shaper? And when I use the Core One, how often should I expect to check and adjust the tension, and how easy is that?
I agree on the "wait and see" bit. Once Prusa published the assembly instructions for the kit, they should provide clarity on this. I'm just too keen and curious to simply sit back and wait for my new toy. 😊
RE: Belt tension adjustment
I can understand that, I feel the same way.
Prusa will only publish the instructions when the first kits are delivered.
Nero 3D has tightened the belts in its Core One Live Stream
Just by feel, without measuring.
RE:
I did the IS calibration on my Core One yesterday. The values have changed by 5Hz for me. But what was much more important was that the filter method changed from EI to ZVD. Visually, however, I couldn't see any difference in the prints (before vs. after).
I think that Prusa is of the opinion that it is not absolutely necessary to use the acceleration sensor
Have you tested it on the MK4S?
It comes out with approximately the same values as Prusa uses.
The difference is probably less than the measuring tolerance of the sensor.
RE:
I would like to know how to adjust the belts as well, I got my core one yesterday and every time I turn it on and try to start a print it wants to recalibrate home and just sits in the front right corner wacking the sides over and over for 5 to 10 minutes with repeated errors about homeing and calibration and print head collisions but doesn't ask me to do anything. It eventually seems to give up trying to calibrate home and just goes ahead and starts the print.
I read that could be due to loose belts? I'm coming from a mk3s and a mk4 so this is my first core xy machine.
RE: Belt tension adjustment
Ok, There are two holes in the front with adjustment screws for the belt tension, I tightened each one up by about 1 and a half turns, ran a homing calibration which it did quickly this time and started a print with no issue so hopefully that fixed this problem.
RE: Belt tension adjustment
I think these two shapers are quite similar. If I remember correctly, they also switched between them in the MK4 at the beginning.
Have you ever done the Klipper test print? You can see the differences better there.
I did the IS calibration on my Core One yesterday. The values have changed by 5Hz for me. But what was much more important was that the filter method changed from EI to ZVD. Visually, however, I couldn't see any difference in the prints (before vs. after).
I think that Prusa is of the opinion that it is not absolutely necessary to use the acceleration sensor
Have you tested it on the MK4S?
It comes out with approximately the same values as Prusa uses.
The difference is probably less than the measuring tolerance of the sensor.
RE: Belt tension adjustment
Ok, There are two holes in the front with adjustment screws for the belt tension, I tightened each one up by about 1 and a half turns, ran a homing calibration which it did quickly this time and started a print with no issue so hopefully that fixed this problem.
Combined with the "no firmware, no test print" observation, this really starts to sound like you received a unit which was not quite ready for shipment? Maybe you should submit a summary of your findings to Prusa's support team -- to put a stake in the ground in case you run into more issues and this turns into a warranty discussion, and to make them aware that there may be a gap in the quality control process.
Also, Prusa should really publish a proper procedure for checking and adjusting the belt tension sooner rather than later. Not every user will have the experience to apply the right (?) number of turns by seat-of-pants.