Acelerometers and input shaping

[peirof]

Hi,

I have seen in the forum a very interesting thing and it is the use of accelerometers to reduce the effects it can produce on the printed piece, when the hotend reduces (for example when reaching a corner) or accelerates, when it begins to print a straight line.

I have even seen in the forum some tremendously clarifying videos of what is sought with the adjustment of acceleration and jerk, I think.

But the truth is that it is not very clear to me.

Some questions arise:

• What accelerometer are compatible with Duet?
• Accelerometer conects to Duet or external pc?
• How to connect and configure an accelerometer?
• How are the results intérprete?
• Where could I see the accelerometer reading?

And more...

There is some guide/wiki/Web where I can find more info?

My ignorance is absolute, but the results obtained in the videos are impressive

I want to learn more about this...

Thx

[zapta]

@peirof , it's still in beta and the documentation is sparse and incomplete. Since you didn't ask about SBC, I presume that you run a duet in stand alone mode (so am I). The first step is to get, connect and configure an accelerometer as described here https://duet3d.dozuki.com/Wiki/Accelerometers

Next steps would be to enable the accelerometer and the input shaping plugins in DWC and make measurements. Ask for direction once you get to that area

BTW, after setting up a temporary accelerometer and enabling input shaping, I am not sure that using an accelerometer is required. An alternative approach is to take the input shaping test STL model and print it with layers of different input shaping frequencies, as we do with heat towers for temperature settings. I think that will also give a significant improvement and may be simpler since the accelerometer integration can be flakey due to long wires and interferances.

[oliof]

@zapta printing a reference model certainly can work, but takes more time. you can also just measure ripple distance on an existing print and multiply with print speed to get an approximation, which takes less time than fitting an accelerometer and doesn't need extra prints (the old DAA method as documented on the wiki, basically).

[zapta]

@oliof said in Acelerometers and input shaping:

@zapta printing a reference model certainly can work, but takes more time

If I count the time it took me to get the parts, understanding the connection diagram and configuration and debugging noise issues, it took me X20 more time than running the input sharping tower, and when done, I run a few input shaping towers anyway to confirm the results and the overall impact (on x, y, interaction with PA, and so on).

What we need is a reasonably priced and off the shelf small accelerometer board/module with a long thin and flexible cable that easily plug into duets and works reliably. Maybe a custom board with an MCU, accelerometer, and CAN connector (@T3P3Tony, this is your department ;-))

Edit: having to plug a single connector to the Mini5+ would also be a nice feature.

[PCR]

@zapta i have one with USB C ready

[zapta]

@pcr, that sounds intriguing

Please send us pictures when you can.

Edit: stumbled upon this one https://www.dailyduino.com/index.php/2020/06/01/stm32-can-bus/ . The STM32F103C8 are dirty cheap (maybe not at this time due to the global shortage), and well supported by platformio including with a hardware debugger. If they can do the job, it may be a good choice. What do you think?

[JoergS5]

@zapta said in Acelerometers and input shaping:

STM32F103C8

you need CAN-FD for Duet, F1 series only has CAN. STM32 with CAN-FD are the G0, G4, H7, L5 and MP1 series (maybe more to come).
There is a STM32 document about CAN-FD: https://www.st.com/resource/en/application_note/dm00625700-fdcan-peripheral-on-stm32-devices-stmicroelectronics.pdf

[zapta]

Thanks @joergs5, this is very useful. I have zero experience with CAN bus.