Resonance Compensation

dgrat

See: https://www.klipper3d.org/Resonance_Compensation.html
Looks like for Delta printers it would be even simpler, because all towers are equal.
Paper: https://www.researchgate.net/publication/224606974_Input_shaping_for_nonlinear_systems

sebkritikel

Discussed some here:
https://forum.duet3d.com/topic/18529/input-shapers-klipper/9?_=1602859383989

But I would LOVE to see this implemented in RRF. DAA is useful, but this appears to be much, much more powerful.

T3P3Tony

I looked into the klipper implementation of this earlier in the week (as it was mentioned at VERRF) and as far as I can see they are all a centred on single frequency and are just a band of frequencies of different ranges that are filtered. I presume the downside of this is that this start to limit the allowable accelerations, as the band increases.

@dc42 would need to confirm but AFAIK DAA is a single frequency, not a narrow band (its certainly not a wide band).

Overall though this is really useful research. It fits loosely to experimental work we are doing to improve printing so I will certainly delve further!

Robthide37

Klipper doesn't adjust accel and it has ability for 2 frequencies currently plus you can do the sam ething as daa and use a multiple if you have alot of frequencies. The best thing about input shaping is it actually alters toolpath so their is no loss of speed

arhi

@Robthide37 alters toolpath? in what way? I'm not sure I'd like toolpath to be changed?

Phaedrux

@arhi said in Resonance Compensation:

I'm not sure I'd like toolpath to be changed?

What if altering the tool path slightly ends up with a physical result closer to the actual desired tool path?

arhi

@Phaedrux dunno, have hard time believing that will be a universal solution

Robthide37

@arhi it is not universal you need to tune it or you can even use an accelerometer to provide the feedback you need to set the frequency and dampening level

GeneRisi

@T3P3Tony , I use klipper on my modified BigBox printer. Compensation can be done for both X and Y axis' . The result on my printer, with two lightweight carriages in an IDEX configuration, is pretty amazing (I can watch ringing ripples disappear when using their test pattern). Also, there are multiple algorithms that are available now for the resonance compensation, so I think Dmitri's work will be a great resource for development!