Notifications
Clear all

Accelerometer?  

Page 2 / 3
  RSS
Netpackrat
(@netpackrat)
Reputable Member
RE: Accelerometer?

As I understand it, it's a one time calibration that you run.  You don't need to do it again unless you make physical changes to the printer.  I recently (as in a couple of weeks ago) finished the build of a Voron 2.4 using a kit from LDO Motors.  The kit included the "input shaping tool kit" consisting mainly of the accelerometer and a connection PCB that you install on the Raspberry Pi controller board, a cable and some attachment hardware.  You temporarily install the accelerometer to the print head and run its cable to the connector board on the Pi.  After the input shaping calibration has been run, you remove the accelerometer and its cable.  I have not done that yet...  I am going to print with it for a while to let it run in, then readjust the belt tension and run the input shaping calibration.

 

Posted : 03/04/2023 10:16 pm
(((GRIFFCOMM
(@griffcomm)
Estimable Member
Topic starter answered:
RE:

It wouldnt be "active", ergo once stored its fine, but then what happens as the belts start to wear or you move to a new desk, you need to re-calibrate.  So how often does the XL check? is it live?

I really hope this can be turned off, my MK3s works fine, i dont want a feature on when it appears to be half baked (which is very surprising given Prusa past history).

This post was modified 12 months ago by (((GRIFFCOMM

Many Thanks

Posted : 03/04/2023 10:34 pm
addohm
(@addohm)
Estimable Member
RE: Accelerometer?

Fine, until something changes.  Being able to selectively calibrate, or actively calibrate is sort of necessary.  I certainly don't want my printer to be down for weeks or months just because of a mishap or a mod unknowingly throwing it out.

Posted by: @griffcomm

It wouldnt be "active", ergo once stored its fine, but then what happens as the belts start to wear or you move to a new desk, you need to re-calibrate.  So how often does the XL check? is it live?

I really hope this can be turned off, my MK3s works fine, i dont want a feature on when it appears to be half baked (which is very surprising given Prusa past history).

 

Posted : 04/04/2023 1:57 am
Martin_au
(@martin_au)
Reputable Member
RE: Accelerometer?

I haven't dug into the maths much (yet), but input shaping to account for different surfaces that the printer is mounted on sounds very close to the concept of people levelling their bed with a level bubble. There should be a couple of inherent frequencies in the Prusa Mk4, which are likely to be very consistent across all the Mk4s (and likely the Mk3 as well). Putting a Mk4 on a wobbly surface will expose another set of wavelengths to the accelerometer as the printer wobbles around. Correcting for those wavelengths is not correct or useful. The only issue I can see with calibrating on a hard surface, then moving to a wobbly surface is that allowing the printer to move may may provide some minor amount of damping change under some circumstances, but that would vary depending on speed and frequency of printing and wouldn't be something correctable by input shaping anyway (also, likely to be too minor an effect to worry about). 

Posted : 04/04/2023 2:51 am
(((GRIFFCOMM
(@griffcomm)
Estimable Member
Topic starter answered:
RE:

Actually, this looks more like it compensates for whats on the BED, not just the printer, once your print is above 3 inches tall, resonance is a HUGE issue, we have had to re-design modals as the whole thing would would resonate the print and it was unable to print anymore (the modal was shaking due to whats being printed)... oce video i saw was how the moving bed could move to NOT resonate the print on it, so it clearly ISNT a static measurement as it will change for every single print on the bed, this has to be a live reading that changes every 1mm of plastic is placed on the bed, as that will change the resonance felt through the printer and culd reuslt in the print shacking but the bed with the print and the head remaining in line (which is what you want with this).

 

This is very different to the XL for which the bed doesnt move forward and back, so i wonder if this wasnt fully thought through when saying the Mk4 will have this feature, the whole print moves on the Mk4, it doesnt on the XL, so if anything the Mk4 would benifit from this way more than the XL would (or the same due to belt lengths).

This post was modified 12 months ago by (((GRIFFCOMM

Many Thanks

Posted : 04/04/2023 3:17 am
Martin_au
(@martin_au)
Reputable Member
RE: Accelerometer?

 

Posted by: @griffcomm

Actually, this looks more like it compensates for whats on the BED, not just the printer, once your print is above 3 inches tall, resonance is a HUGE issue, we have had to re-design modals as the whole thing would would resonate the print and it was unable to print anymore (the modal was shaking due to whats being printed)... oce video i saw was how the moving bed could move to NOT resonate the print on it, so it clearly ISNT a static measurement as it will change for every single print on the bed, this has to be a live reading that changes every 1mm of plastic is placed on the bed, as that will change the resonance felt through the printer and culd reuslt in the print shacking but the bed with the print and the head remaining in line (which is what you want with this).

 

This is very different to the XL for which the bed doesnt move forward and back, so i wonder if this wasnt fully thought through when saying the Mk4 will have this feature, the whole print moves on the Mk4, it doesnt on the XL, so if anything the Mk4 would benifit from this way more than the XL would (or the same due to belt lengths).

Pretty easy to have an static value that's updated during the slicing process. That's the solution I'd expect. I'm pretty sure they don't have an accelerometer on the bed.

Posted : 04/04/2023 5:02 am
jseyfert3 liked
(((GRIFFCOMM
(@griffcomm)
Estimable Member
Topic starter answered:
RE: Accelerometer?

I meant on the Mk4, wouldnt need it for the bed on the XL as the bed only drops, the bed on the Mk4 slides back and forth AND vibrates to hell when doing infill, we had major issues with some designs, had to re-design them to stop this vibration.  

Many Thanks

Posted : 04/04/2023 9:14 am
Thejiral
(@thejiral)
Noble Member
RE: Accelerometer?

 

Posted by: @martin_au

I haven't dug into the maths much (yet), but input shaping to account for different surfaces that the printer is mounted on sounds very close to the concept of people levelling their bed with a level bubble. There should be a couple of inherent frequencies in the Prusa Mk4, which are likely to be very consistent across all the Mk4s (and likely the Mk3 as well). Putting a Mk4 on a wobbly surface will expose another set of wavelengths to the accelerometer as the printer wobbles around. Correcting for those wavelengths is not correct or useful. The only issue I can see with calibrating on a hard surface, then moving to a wobbly surface is that allowing the printer to move may may provide some minor amount of damping change under some circumstances, but that would vary depending on speed and frequency of printing and wouldn't be something correctable by input shaping anyway (also, likely to be too minor an effect to worry about). 

I am not an expert but I have done input shaper calibration on a Voron 0.1 myself. Correct me if I get things wrong but I don't think resonance works like you describe it here. Where the wobble is coming from is secondary. What matters is at which frequency of your back and forth movement of the specific axis leads to the strongest accelerations on the Extruder (for a core xy, for the bed slinger also on the bed), that's where your resonance frequency is. What you are measuring with the accelerometer is exactly what matters as long as you screwed the accelerometer well in place without adding much wobble additionally which is deviating from the actual movement of the extruder. 

Resonance frequencies are a delicate thing and can change substantially with only minor changes in the setup. What the printer is standing on has a big influence, I mean it is often even easy to hear the difference. My Mk3s was quite noisy from vibrations before I put a stone slab below it. Afterwards much of that vibration noise was gone. That is resonance in action. Obviously it had changed. If you ignore that, you can just as well print without input shaper altogether. 

You don't need an accelerometer to calibrate input shaper of course, you can also print calibration prints. Apparently one can get decent calibration results that way too. It is a lot more bothersome than some automated calibration via accelerometers however. 

Mk3s MMU2s, Voron 0.1, Voron 2.4

Posted : 04/04/2023 2:13 pm
Goyetus
(@goyetus)
Eminent Member
RE: Accelerometer?

In marlin 2.1.x  the only way to set input shaping  is with "print test files".   Accelerometers are not supported.
Since mk4  is  in Marlin  and going to Klipper,   meanwhile they are in Marlin,   It´s nosense to give acelerometers to the user,   but in my opinion,   1$ only .... they should have integrated in the canbus and in the bed.  

 

 

Posted by: @thejiral

 

Posted by: @martin_au

I haven't dug into the maths much (yet), but input shaping to account for different surfaces that the printer is mounted on sounds very close to the concept of people levelling their bed with a level bubble. There should be a couple of inherent frequencies in the Prusa Mk4, which are likely to be very consistent across all the Mk4s (and likely the Mk3 as well). Putting a Mk4 on a wobbly surface will expose another set of wavelengths to the accelerometer as the printer wobbles around. Correcting for those wavelengths is not correct or useful. The only issue I can see with calibrating on a hard surface, then moving to a wobbly surface is that allowing the printer to move may may provide some minor amount of damping change under some circumstances, but that would vary depending on speed and frequency of printing and wouldn't be something correctable by input shaping anyway (also, likely to be too minor an effect to worry about). 

I am not an expert but I have done input shaper calibration on a Voron 0.1 myself. Correct me if I get things wrong but I don't think resonance works like you describe it here. Where the wobble is coming from is secondary. What matters is at which frequency of your back and forth movement of the specific axis leads to the strongest accelerations on the Extruder (for a core xy, for the bed slinger also on the bed), that's where your resonance frequency is. What you are measuring with the accelerometer is exactly what matters as long as you screwed the accelerometer well in place without adding much wobble additionally which is deviating from the actual movement of the extruder. 

Resonance frequencies are a delicate thing and can change substantially with only minor changes in the setup. What the printer is standing on has a big influence, I mean it is often even easy to hear the difference. My Mk3s was quite noisy from vibrations before I put a stone slab below it. Afterwards much of that vibration noise was gone. That is resonance in action. Obviously it had changed. If you ignore that, you can just as well print without input shaper altogether. 

You don't need an accelerometer to calibrate input shaper of course, you can also print calibration prints. Apparently one can get decent calibration results that way too. It is a lot more bothersome than some automated calibration via accelerometers however. 

 

 

Posted : 04/04/2023 2:26 pm
Thejiral liked
addohm
(@addohm)
Estimable Member
RE: Accelerometer?

Completely agree.

Posted by: @goyetus

In marlin 2.1.x  the only way to set input shaping  is with "print test files".   Accelerometers are not supported.
Since mk4  is  in Marlin  and going to Klipper,   meanwhile they are in Marlin,   It´s nosense to give acelerometers to the user,   but in my opinion,   1$ only .... they should have integrated in the canbus and in the bed.  

 

 

Posted by: @thejiral

 

Posted by: @martin_au

I haven't dug into the maths much (yet), but input shaping to account for different surfaces that the printer is mounted on sounds very close to the concept of people levelling their bed with a level bubble. There should be a couple of inherent frequencies in the Prusa Mk4, which are likely to be very consistent across all the Mk4s (and likely the Mk3 as well). Putting a Mk4 on a wobbly surface will expose another set of wavelengths to the accelerometer as the printer wobbles around. Correcting for those wavelengths is not correct or useful. The only issue I can see with calibrating on a hard surface, then moving to a wobbly surface is that allowing the printer to move may may provide some minor amount of damping change under some circumstances, but that would vary depending on speed and frequency of printing and wouldn't be something correctable by input shaping anyway (also, likely to be too minor an effect to worry about). 

I am not an expert but I have done input shaper calibration on a Voron 0.1 myself. Correct me if I get things wrong but I don't think resonance works like you describe it here. Where the wobble is coming from is secondary. What matters is at which frequency of your back and forth movement of the specific axis leads to the strongest accelerations on the Extruder (for a core xy, for the bed slinger also on the bed), that's where your resonance frequency is. What you are measuring with the accelerometer is exactly what matters as long as you screwed the accelerometer well in place without adding much wobble additionally which is deviating from the actual movement of the extruder. 

Resonance frequencies are a delicate thing and can change substantially with only minor changes in the setup. What the printer is standing on has a big influence, I mean it is often even easy to hear the difference. My Mk3s was quite noisy from vibrations before I put a stone slab below it. Afterwards much of that vibration noise was gone. That is resonance in action. Obviously it had changed. If you ignore that, you can just as well print without input shaper altogether. 

You don't need an accelerometer to calibrate input shaper of course, you can also print calibration prints. Apparently one can get decent calibration results that way too. It is a lot more bothersome than some automated calibration via accelerometers however. 

 

 

 

Posted : 04/04/2023 3:46 pm
jseyfert3
(@jseyfert3)
Reputable Member
RE: Accelerometer?
Posted by: @griffcomm

Actually, this looks more like it compensates for whats on the BED, not just the printer, once your print is above 3 inches tall, resonance is a HUGE issue, we have had to re-design modals as the whole thing would would resonate the print and it was unable to print anymore (the modal was shaking due to whats being printed)... oce video i saw was how the moving bed could move to NOT resonate the print on it, so it clearly ISNT a static measurement as it will change for every single print on the bed, this has to be a live reading that changes every 1mm of plastic is placed on the bed, as that will change the resonance felt through the printer and culd reuslt in the print shacking but the bed with the print and the head remaining in line (which is what you want with this).

 

This is very different to the XL for which the bed doesnt move forward and back, so i wonder if this wasnt fully thought through when saying the Mk4 will have this feature, the whole print moves on the Mk4, it doesnt on the XL, so if anything the Mk4 would benifit from this way more than the XL would (or the same due to belt lengths).

That could be calculated though. The software already knows how much mass is added during the print and where. So, at least in theory, you can compensate for this extra mass without needing live feedback.

Posted : 05/04/2023 12:02 am
(((GRIFFCOMM
(@griffcomm)
Estimable Member
Topic starter answered:
RE: Accelerometer?

Maybe possible, but as this whole thread is all about the accelerometer, its way easier to have that installed then the system can read that and compensate for everything that might be thrown at it in real life with out any virtual math hokum going on.

Many Thanks

Posted : 05/04/2023 12:37 am
addohm liked
jseyfert3
(@jseyfert3)
Reputable Member
RE: Accelerometer?
Posted by: @griffcomm

Maybe possible, but as this whole thread is all about the accelerometer, its way easier to have that installed then the system can read that and compensate for everything that might be thrown at it in real life with out any virtual math hokum going on.

The people that invented inertial navigation for submarines just cried...but the submarines are navigating just fine on that "virtual math hokum".

Posted : 05/04/2023 1:15 am
Netpackrat
(@netpackrat)
Reputable Member
RE: Accelerometer?

 

Posted by: @goyetus

In marlin 2.1.x  the only way to set input shaping  is with "print test files".   Accelerometers are not supported.
Since mk4  is  in Marlin  and going to Klipper,   meanwhile they are in Marlin,   It´s nosense to give acelerometers to the user,   but in my opinion,   1$ only .... they should have integrated in the canbus and in the bed.  

I thought they said their implementations of input shaping and pressure advance would be Klipper inspired, not that they were going to Klipper.

Posted : 05/04/2023 1:28 am
Shushuda liked
TT
 TT
(@tt-3)
Active Member
RE: Accelerometer?

Well, that is not completely accurate.

"Inertial navigation is a self-contained navigation technique in which measurements provided by accelerometers and gyroscopes are used to track the position and orientation of an object relative to a known starting point, orientation and velocity"

Posted : 05/04/2023 1:32 am
jseyfert3
(@jseyfert3)
Reputable Member
RE: Accelerometer?
Posted by: @jseyfert3
Posted by: @griffcomm

Maybe possible, but as this whole thread is all about the accelerometer, its way easier to have that installed then the system can read that and compensate for everything that might be thrown at it in real life with out any virtual math hokum going on.

The people that invented inertial navigation for submarines just cried...but the submarines are navigating just fine on that "virtual math hokum".

Which is maybe not the most apt analogy, because inertial navigation relies on accelerometers.

But, they use a known initial position, with accelerometers then measuring accelerations when the position is used, in order to track position over time. And by track, I mean calculate, because position is the 2nd integral of acceleration. So it's 100% reliant on that "virtual math hokum" to know where it is.

Posted : 05/04/2023 1:33 am
jseyfert3
(@jseyfert3)
Reputable Member
RE: Accelerometer?
Posted by: @tt-3

Well, that is not completely accurate.

"Inertial navigation is a self-contained navigation technique in which measurements provided by accelerometers and gyroscopes are used to track the position and orientation of an object relative to a known starting point, orientation and velocity"

I was writing an addendum (see above) when you posted this. Yes, inertial navigation requires accelerometers. But also requires math, calculus specifically, in order to calculate the position of the object. You can't do it without continuous double integration of the accelerometer data.

Posted : 05/04/2023 1:36 am
Virgile eligriV
(@virgile-eligriv)
New Member
RE: Accelerometer?

I'm wondering how this will work for the kit version, wait and see...

Posted : 05/04/2023 11:26 am
morgun_ua
(@morgun_ua)
Member
RE: Accelerometer?

While the weight of the bed is a static a easily pre-calculatable value, you cannot say the same for mass. And the latter is what matters the most, while your G-CODE can state that x will move 420mm with 10000acceleration it doesn't guarantee that printer won't override those parameters.  That could work in theory if we'd something like an Apple ecosystem, where we'd had tight software/hardware integration and all parts calibrated and paired for one exact printer with all the downsides. Once any of those will became variable, process of precalculating it in software becomes practically impossible, so we need a hardware accelerometer to compensate for that.

If you think it's still anyhow possible, consider the table your placing the printer will have it's own weight and its resonance frequency which will change depending on the mass and what on it. 1Kg of liquid will behave differently that a 1Gg of filament lying on the table next to the printer. Even stepper motors will very their speed and acceleration depending on their temperature.

Posted : 29/08/2023 8:48 am
Rainer
(@rainer-2)
Reputable Member
RE: Accelerometer?

Maybe I'm misunderstanding it because of the language, but I have the impression that it is assumed that the accellrometer is used during printing.

However, it is only used to determine the resonance frequency and recommend values for damping and acceleration.

Therefore it does not matter whether this is determined with a test print or accellrometer. In print, it doesn't matter at all.

Unlike the Corexy, the bedslinger's mass increases in Y, prusa compensates for this by changing frequency and acceleration with increasing mass.

Posted : 29/08/2023 9:46 am
Page 2 / 3
Share: