Actual speed versus planned speed

Zesty_Lykle

Hi All,

We were asked to do a few videos of high speed printing, just to show that the Nimble could handle it.
We set a high speed in the slicer and then tweaked the % slider in the web interface to up the speed.
We were supposed to go at a speed of about 300 mm/sec.
But that didn't look right and we were called out on it.

So I did a fresh test. All the % settings in S3D are at 100% and original slicer speed was set to 200 mm/sec.
The g-code at the start confirms this: G1 X24.034 Y-73.971 E0.0062 F12000

Loaded it up and started printing. Increasing the speed from 35% to 100%
This is the result at the 100% mark: https://youtu.be/IcLxbGk9-iA

Now compare that to this clip: https://www.youtube.com/watch?v=V1r2HtWPGw8

Ours does not look as fast and watching it in real life it doesn't look as fast.

2 questions:
1 Any idea what is going on?
2 Do you know of a reliable way to measure actual nozzle speed?

Thanks

W3DRK

It sounds to me like your maximum speed is being limited by a low acceleration setting. In other-words, the path isn't long enough for it to reach maximum speed before it needs to slow down and change direction. What are your current acceleration settings? It may also be limited by whatever you have the maximum axis speeds set to as well.

Zesty_Lykle

Hi
The path is more than long enough, approx 65 mm.
My accel setting is: ah, oh, hmm. 800
Still that should be enough shouldn't it?

Lykle

W3DRK

Wow that's super low. To put that into perspective I safely use an accel of 3000 on my Kossel.

Zesty_Lykle

No idea why I set it so low. Set it at 1500 now and it has increased the perceived speed. Still not sure if the end speed is near the planned speed.

Still trying to figure out a way to measure actual speed.

Thanks for your help. It helped a lot.

T3P3Tony

Lykle - also check you maximum speed for the axis and for the extruder that are configured in firmware. Movement speed will be limited so it does not violate maximum speeds for all the axis involved.

Dougal1957

Lykle My default speed for a delta is 18000 yours at 12000 is limiting you to a max of 200 mm/s 18000 takes you to 300 if you want to say try at 400 you need to set 24000 as for Accels the default is 1000 for a mini Kossel but if your printer is built nice and strong I would try upping it to 2000 and maybe behond also you may need to adjust the Max Speed Changes as well the default is 1200 (I believe that this equates to the Jerk setting?)

to make sure you are getting the max out of it I think you need to print a rectangular box where the max length is as long as you can Print this may allow you to get upto the speeds that you want?

Doug

W3DRK

It will move (within reason) at whatever speeds you command assuming it's not being limited by:

X,Y,Z motor maximum speed - These settings obviously need to be high enough to keep up with the commanded feed rates. This is easy with a Cartesian printer but trickier with a delta since motor mm/sec doesn't directly correlate to effector speed.

Extruder motor maximum speed - This must be high enough to push filament fast enough to keep up with a high speed print. If the extruder maximum mm/sec isn't high enough (unlikely) the firmware will "govern" the other movements to compensate, effectively limiting your printing speed.

Acceleration - If your acceleration is too low, unless the path is very long, the motors may not have enough time to ramp up to full speed before they need to ramp back down to change direction.

Processor speed - However unlikely it is possible to exceed the limits of the Duet/Duet WiFi MCU, especially when using 0.9 degree/step motors. I use 0.9 deg steppers with x16 micro-stepping on my Kossel 250 and have no issues moving edge to edge at 300mm/sec.

EDIT: I forgot MCU speed

Dougal1957

@W3RDK that is a very good explanation thanks for that??

Zesty_Lykle

OK guys, great. That will keep me busy for tomorrow.
Why did I start this project again?

W3DRK

For what it's worth I'm planning to pledge for one of your extruders. I am curious though, and I'm sure this isn't the place to ask…but how do you deal with any torsional spring effect of the cable between the motor and drive gear? Surely there must be some energy being stored in the twisting of the drive cable. Doesn't this effectively behave similarly to the elastic properties of filament in a bowden tube?

Hugues

You can also mount a volcano and 1.2 mm nozzle on your extruder.
If you extruder can push filament through it at the high standard print speed ( like 120mm/s on 0.1mm height) you should convince lot of pro user… ; -)

I also notice a big difference when extruder motor are setup to low or standard accel/jerk and that's worst with geared extruder. Mine are higher than movement.

elmoret

Prusa has a great acceleration calculator here:

http://prusaprinters.org/calculator/

Zesty_Lykle

Hi all,
Where do I find the time?!?!
I have run a Volcano with a 1.2 nozzle, and that one works. I think I went up to 80 mm/sec just to see it print, but have not done any real performance testing on speed with the Volcano.
Speed seems to be tricky to measure anyway.

I had upped the max speed, so that was not the factor. It simply was the acceleration. I am now printing visibly faster. Also, the printer is starting to shake a little.

Still have not figured out how to measure accurate nozzle speed. Maybe a ruler behind it, and shoot a video then calculate the speed by examining the frames. Phew, no time.

@W3DRK, no problem. There is a thread here somewhere, but I can answer here.
Yes there is flex in the drive cable, it is minimal as we stay well within the torque limits. But the flex goes into the gear system and is reduced 30:1. So it is reduced a great deal, so much so that it is not a factor. I use 1.2 mm retraction on my Delta and Brian uses 0.9 on his. So you can see it is minimal.

deckingman

@Zesty_Lykle:

Hi all,

Still have not figured out how to measure accurate nozzle speed. Maybe a ruler behind it, and shoot a video then calculate the speed by examining the frames. Phew, no time.

It's difficult to do. On a simple left to right move, starting from rest, there is a very quick "jerk", then it accelerates up the speed, maintains that for a while (assuming that the move was long enough for it to have reached maximum speed), then it decelerates down to a stop, changes direction and starts over. Do you take the bit in the middle as being he nozzle speed, or do you take an average of the entire move? Which figure is more meaningful?

IMO, it's all a bit academic as most machines are limited by how quickly the filament can be melted. I went to great pains in selecting motors that would give me the highest torque and acceleration but in practice, I can't anywhere close to using those numbers. Even non-print moves are set well below what is theoretically possible for the motors taking in to account the weight of the axes because it's just too scary. According to the calcs, I should be able to accelerate the X axis at a tad over 3,000 mm/sec/sec up to around 780mm/sec. In practice, I use about a quarter of the acceleration and half the maximum speed - it's far less scary.

Just my twopence worth.

T3P3Tony

Lykle a more realistic real world test could be printing something and seeing how long it takes?

bot

Just feed filament as fast as the motor can push it, at temp… you can calculate the volumetric rate of extrusion, and therefore the maximum possible feedrate at a given layer height/extrusion width.

You can also calculate how fast the nozzle will be moving at any point in time using calculus (I believe -- I'm no mathematician) and plugging in the known values from firmware and the gcode file.

Adamfilip

I havent found that its possible to actually print anything near those speeds..

are people printing good quality at 300mm/sec??

deckingman

@Adamfilip:

I havent found that its possible to actually print anything near those speeds..

are people printing good quality at 300mm/sec??

I doubt it. Most machines are limited by how fast the hot end can melt the filament and that is largely due to the size of the melt chamber and how long the filament stays in it, and also to some extent how hot you run it at. Some people push the temperature up in order to achieve higher speeds but that can lead to all sorts of other problems. Even if the hot end is capable of melting filament that fast, 300 mm/sec may be possible for longish moves but imagine infilling something that is only 10mm wide. For one thing, you have to accelerate, move, decelerate stop and change direction and repeat so you'd be unlikely to ever get up to 300mm/sec on such short moves. Even if you had instantaneous speed changes doing that 10mm zigzag will mean moving the the hot end back and forth 30 times a second. Even if that was possible without the machine shaking itself to bits, and even if the extruder was capable of such fast moves, the filament won't be - it's too "spongy" to be stopped and started that quickly.

peridot

Your best bet for hitting printing speed limits is a largish smooth object printed in spiral vase mode - there are no corners or stops anywhere, so as long as the radius of curvature isn't too tight the printer can get up to its maximum speed and stay there.

For my printer I get fairly reasonable non-printing motion up to 300 mm/s, but print moves are limited by how fast the extruder can extrude. Currently my extruder is tuned to quite conservative settings (in terms of maximum speed, acceleration, and jerk) because I had some trouble. Maximum speed you can tune by simply extruding into air and seeing when you start skipping steps; of course this will depend on filament type and temperature. Acceleration and jerk settings for the extruder I find a bit more mysterious, but they're probably more relevant to retraction than to printing. Since it's your extruder you're testing, this makes sense; if you've got a good mechanical system, your print speed is limited by how fast your extruder can push your filament through your hot end.