FBS Vibration Compensation

Kulitorum

See this article:

http://www.3ders.org/articles/20171020-university-of-michigan-professor-doubles-3d-printing-speeds-using-vibration-mitigating-algorithm.html

Here's the paper:

https://umich.app.box.com/s/n9cvs27ckehdr64gzv5igtmboykymgk6

Have some free time?

Kulitorum

dc42

Looks interesting! I'll get in touch with the authors.

robm

Yes please!
Haven't looked at the paper yet, but from the article it is interesting to note (1) it involves some printer specific vibration calibration, and (2) - perhaps due to (1) - the work is targeted at firmware rather than slicer software.

DocTrucker

Is the duet fast enough to sample position encoder signals to monitor the ringing? If so an auto calibration routine could be considered to study the ring and interpolate the printer dynamics.

If not traditional position encoders you maybe able to 'listen' to the vibrations and apply suitable correction. Essentially noise cancelling earphones for your printer.

brunofporto

@doctrucker and @dc42 You can measure your printer's natural frequencies using your smart phone accelerometer. If the Duet "talks" with this measurement and get the data from the the measurements it can automatically avoid the rotation speeds of the motors that can induce resonance on any printer.

Or you can implement the reading of a common i2c or SPI accelerometer at the head and close the control loop for good

So it could automatically set acceleration, speeds, jerk (the real one) and other parameters based on your particular printer.

The printer could make an automatic test g code making it "zig zag" from maximum travel at each axys and at different speeds until it have a good picture of the natural frequencies. Then the printer would set all the settings needed for this algorithm.

https://www.youtube.com/watch?v=Ws1JfHl3Y0o

DocTrucker

Some really interesting techy stuff there! I was thinking about a PhD project one of my peers was working on way back while doing research. His project had a large optical component but one of the ideas worked on was to allow the use of more flexible (i.e. lighter) robots and through the fast optical sensor detect oscilations and commanding an inverse oscillation to the position control to damp out the oscillation. Realistically for our machines I think we'd be talking about requiring FPGA sort of speeds along with the ultra fast accelerometers which probably would need most of us to up our game in terms of electrical noise control. The FPGA programming is a whole different mindset to micro controllers. Having said that with the parallel processing being demonstrated with the 2.0 release of firmware the micro controllers are proving themselves ever-increasingly more powerful.

Edit: Arduino have or are about to release an FPGA development kit. I may have a play to see if that is quick enough to get a good signal from an accelerometer strapped to the head or if noise is too bad!

Edit 2: My competence is more along the lines of a hobbyist hacker/tinkerer now rather than professional so don't be expecting my results too quick!