hi all,
since i didn’t find anything about this in the klipper documentation: what is the reasonable value range for square corner velocity?
that the default value is 5 is clear. in my case the printer is an ender-3 pro. but also in general and for other printers i would be interested in what is considered reasonable.
cheers
tom
Generally it is not recommended to change this value. It makes a lot more sense to “play” with your maximum acceleration and see what you printer can take.
But the headroom of a stock Ender 3 likely is quite limited.
and why does this not make sense under klipper - is no recommended range communicated - and is the value adjustable then at all? in marlin a change of the jerk value makes clearly visible sense.
It’s your printer, do what you want with it. The default value of 5 generally works well on every printer, but you can set it to whatever you want.
what a pointless answer.
You seem to think we owe you something more than a recommendation to use the default.
you owe me zero and now don’t bother me with your psycho problems anymore. period.
I suggest you keep your tone civil. This is neither endorsed nor accepted here.
SCV is a pretty complex subject and highly depending on your hardware. A general statement cannot be given except that your best bet is to stick with the default or experiment and find out what works for you.
Typically it does not yield any significant value in terms of print time and potentially just makes your quality worse.
i am simply done with such forum participants who have no clue themselves but then unasked dumb down others with blah like ‘It’s your printer do what you want with it’ or 'you seem to think we owe you something '. talk like this doesn’t help anyone and is not supposed to.
ok. got it. if this setting is so fragile that actually only one value is valid for all printers and you can’t control the behavior in the corners properly with it the bottom line is that either the algorithm for cornering is not very good or its integration into the overall calculation is not very well done. the odd thing about it is that pressure advance doesn’t work very well with the bowden extruders either.
p.s. however, my main question would have been to you - because i could already see here that you have a plan of klipper - which value range you would consider reasonable to try out with the ender-3 pro. and it’s not about speed but about getting the cleanest corners possible. and there would have been helpful something like: try for the ender-3 pro with bowden something between 4.1 and 5.9 or 1.0 and 10.0 or 1.0 and 100.0. just a normal answer like that. but on the one hand to say ‘scv is a pretty complex subject and highly depending on your hardware’ - but on the other hand the value should always be 5 makes rather little sense for me.
There is probably no generic answer.
My personal feeling would be to carefully calibrate Pressure Advance and leave SCV at its default value.
Actually from what I’ve seen in my testing and profiles of people tuning their printers for high speeds and quality, SCV differs a lot. Also my way of dealing with it is very different from the recommendations in the docs, but your results may vary.
I’m not an expert, but my way to calibrate SCV was:
for some reason, the parts sliced with orcaslicer 2.1.1 is playing with this value midprint. in my chanse it jumps from 8 to 13 every now and then(voron 2.4r2). these seem to be the jerk values/
My couple of cents:
square_corner_velocity model is not really particularly physical (it calculates and uses the cornering radius that’s just a model and does not exist in practice). Two reasons it is used in Klipper is because we (a) need a model to calculate motion for segmented smooth surfaces, where toolhead changes the direction at the corner only very slightly (though Klipper also calculates centripetal acceleration for such cases), and (b) because linear pressure advance model does not work very well for low velocities, so decelerating too much at sharper corners will lead to poor quality of such corners because oftentimes linear pressure advance cannot stop oozing from the nozzle at low speeds.
So, in short, if you reduce square_corner_velocity, you get more physically sound kinematics calculations at sharp corners, but you’ll get worse results on corners due to pressure advance. If you increase square_corner_velocity, pressure advance works better at sharp corners, but the kinematics of the toolhead at corners becomes “borked” (basically, the steppers cannot sharply increase or decrease their velocity, the same goes for the toolhead, and this means they’ll do accelerations/decelerations themselves due to some flexibility of a system (belts, frame, etc.), but this will not be accounted for by Klipper, and may result in greater ringing and other defects at corners). Also if you use input shaping, increasing square_corner_velocity rapidly increases smoothing from input shaping.
Therefore, there is not ‘one size fits all’ answer. Klipper uses an scv==5 as a good enough middle ground. Depending on what you are trying to achieve, you may want to experiment adjusting it based on the points I mentioned above. E.g. if one wants to print a benchy really fast, they probably don’t care about the quality of it too much, so they may raise scv basically up to the point that printer still does not skip steps. Otherwise, you may be able to raise it, but only a little bit. For input shaping smoothing, the typical range of ‘still OK’ scv values is around 5-10 mm/sec, and you can pass --scv=... parameter to the shaper tuning script to see the effect of different scv values on shaper smoothing (at some point you’ll see that the recommended accelerations will start to decrease in order to maintain the reasonable smoothing as you increase scv).
As for bowdens and pressure advance, you are right that PA does not work very well with bowdens. I think that’s pretty much true for all firmware out there, not just Klipper. So that’s the big downside of bowdens, in my opinion, and that’s why I think they have less and less value and practical applications in the modern world now that there are very lightweight direct extruders widely available.
Well, i wasnt looking for good corners whenever i did my tuning on my ender 3 pro and got a sub 10 min benchy (9:54) on stock motors and belts.
What i did was… i used orca slicer and right clicked spawned a cylinder that was about 20mm and another 40mm.
I noticed if i had high scv it wouldn’t lay down the filament well and would fail around smaller corners like the smaller cylinder, but would be fine on the bigger so i just tuned it till it did the smaller cylinder and that was my max. That was like 15 scv i cant remember, 30 was too much tho for printing round stuff but i did try 40 on a benchy and it did print. I ended up on 15 for speed and when i redid a calibration for making the 3d benchy letters on back of boat to look better… after i had fun with all the speed i have a value of 3.33 scv yeah 3.5 probably would work but i like angel numbers and they have worked for me in overclocking my computers throughout the years.
I may seem crazy. Well I am crazy…
Do not do this to your printer if you want long term use out of it.
I am testing how long this will last at 1.2 amps on both x and y.
Motors are louder and wayyy hotter. I zip tied a heat sink on x the other day snd i have had the y axis amps high even when i was on marlin and since marlin i have had a broken bambu hot end heat sink, i used high temp gasket maker to bond it to the y axis motor. Lol yeah im jank and i work with what i got.
God bless and have fun printing!!
(Explaining mods)
(Raw uncut 9min benchy)
Hello @CodeMasterCody3D !
Thank you for your input.