On Friday a french replicape user and I had a chat. Turns out that for Core XY and H-bot printers he and members of the fablab nearby found a significant speed difference between the Replicape and other boards of a similar level (Smoothie, Duet). The same speed difference did not exist for the fablab’s delta printers.
This is truly unfortunate and points to potential implementation problems of the path planner for those motion kinematics.
I’ve created a “beta-test” image of Kamikaze 2.1.0 with the development branch of redeem installed instead of the stable version. In addition there is the CPU frequency fixed to maximum speed, something that can cause issues for kinematics with lots of calculations for stepper motion calculations (especially on a Delta printer).
I am posting the link to this image here, so that anyone who thinks their printer is too slow can give it a try and let us know if this fixes the speed.
What I would ask is that you give feedback on whether or not this helps your printer! Either post here, comment below or join us on Slack and say something in the general or support channels - this is critical information for us to be able to fix the issue once and for all.
How far is it off? A multiple of 70.7% (sqrt(2)/2) or 50% is pretty easy to inadvertently get with CoreXY/HBot kinematics because the rotation transform is also a compound pulley mechanism. Sailfish and Repetier both had issues with that in the past.
Hello Ryan, I wrote a post to explain the problem but I think it need to ne moderated. It’s about 40 to 100% slower than estimation, and real print time on same printer it’s another firmware ils really 30 to 100% slower ( may be a little more on complexe peint)
@Tan_Tuan_TRAN can you share a picture or link of the part? One of the things we suspect is a slowdown due to very complex geometries. A simple square part would be slow due to the default jerk/acceleration values
i have left all by default, but today, i have try to tweak a little x-y acceleration an djerk, but, i didnt remember the time i have done before tweaking them. I’ll publish the time it take with the new image and set this acceleration to default
The layer shift sounds like skipped steps, and it can be difficult to find out the exact amount to go in current… I’ll get Wiley to look at your reports here, since he’s writing the path planner
@Jon_Charnas i have tune up the current, but same. but i seen to be more than a layer shift. with old version, i have acc at 2 and djerk at 0.02, i dont notice this issue.
