Trying to increasing print surface quality (E3D on Flashforge?)

[jimc]

you guys have already mentioned it and are all over the cause of it. its not your steppers. it all this boils down to resonance from sudden direction changes. almost every machine does it to some degree. some are better than other but this is what core xy movement gets rid of or atleast limits. unfortunately printers with core xy are extremely limited but you will be seeing it more and more on newer printers being released. this can also come right down to the material your printer is made of, how rigid the construction of the frame is and if the bed is cantilevered like on a mb/ff and tons of others and also the weight of the moving print head. you can lower the acceleration which can help a ton but this can also just be inherent in your printer's design and you will never get rid of it 100%.

[Dargonfly]

Wow, that is one hell of a difference.

Are you changing the acceleration in replicatorG > machine > onboard preferences?

Care to post your numbers please.

Is this the same problem?
http://3dprintboard.com/showthread.p...rring-features

Yep, same problem!
I'll post some numbers that I've used below;

Awesome stuff!! Glad someone is working on it, that damn shadowing has always been an issue for me too - pretty much on any sort of hole that's printed or detail on the vertical. I would love to know what you did to get that final print, my FF is to the point now where it's done over 2000 hours, and although it's printed some awesome stuff I have never really gotten down to the nitty gritty.

I thought originally like Curious, that you were probably thinking these machines were capable of alot more than they really were, but as you have proven that's not the case and if you spend the time and effort you can get them printing just as good as a 3k ultimaker.

Just a 3k Ultimaker? We'll get it to print much better than that!

you guys have already mentioned it and are all over the cause of it. its not your steppers. it all this boils down to resonance from sudden direction changes. almost every machine does it to some degree. some are better than other but this is what core xy movement gets rid of or atleast limits. unfortunately printers with core xy are extremely limited but you will be seeing it more and more on newer printers being released. this can also come right down to the material your printer is made of, how rigid the construction of the frame is and if the bed is cantilevered like on a mb/ff and tons of others and also the weight of the moving print head. you can lower the acceleration which can help a ton but this can also just be inherent in your printer's design and you will never get rid of it 100%.

Yes, I understand. Perhaps I'll make some parts out of aluminum as soon as I've got a CNC router to improve it a tiny bit more.
And even though my photo doesn't show it, in the last picture there is also some of this shadowing going on, but it's almost not visible (and you can't feel it with your nail) - so with some finish (paint) it'll look very clean.

Ok. Some numbers I tested yesterday - as a disclaimer; I just lowered it by a large amount and didn't test anything time-effeciency-wise yet. NOR have I tested any preferred settings. This will all be done over the following weeks.

Not at home, so have not got exact screenshot, but the acceleration settings can be changed by going to: RepG->Machine->Onboard Preferences->Acceleration

There you will see:

I am using Sailfish 7.7 and I pressed 'Reset motherboard to factory settings' to get standard values.
Then I took the 'X acceleration rate (mm/s²)' and 'Y acceleration rate (mm/s²)' and set them to 100 (because standard is "1000" and not "1" ... man I would've figured this out sooner if us Dutch people would use international standard regarding commas and dots...)
Also changed the 'X/Y max junction jerk (mm/s)' to 8
Press 'Commit Changes'.
..
and that's it that's were the fine-tuning will start. (because this specific setting made it so that prints take about twice as long! - but you'll really see and feel what the machine is doing, which is very interesting and useful! I think that with a lowered acceleration setting you can easily up the maximum speed.)
If you want to test stuff, good luck and please post pictures and/or settings!

[Mjolinor]

I don't think you can increase the max speed, the acceleration is there in order to allow faster overall movements.

If you turn acceleration off completely your machine will start jumping at every change of direction.

By reducing the acceleration you are not changing the behaviour at the corner but are just slowing the head between corners. If you increase speed you will start getting an overshoot at the corner.

[Dargonfly]

Maybe I didn't explain it right:
With the test print (20x20x20mm cube) and the slow acceleration settings it was very clear what the machine did:
It came to a stop at every corner and started accelerating to max speed (took about 5mm (0-30mm/s)) and then it started deaccelerating about 5mm before the next corner.
That means that for 10mm in the 'middle' it was running at a max speed of 30mm/s. I think it can easily aim for a max speed (software-wise) of 60mm/s. This would probably mean that is would accelerate for the first 10mm, and then deaccelerate the last 10mm. (in this particular example.)

Because I feel now that the max. speed isn't hurting the machine, but the large acceleration is. But to get my prints to finish before 2017, I'd like it to print as fast as possible while remaining a good quality. I believe the formula for this is low acceleration and 'decent' max speed. Like a truck! Because currently I was running it like a scooter (high acceleration, low max speed) - but that doesn't work out because the mass of the extruderassembly is much fatter.. (like a truck.)

Also, stuff like overshoot can be adjusted with the JFK? JAK? JAJK? values on the onboard preferences. I adjusted these perfectly at my previous settings, but I'll need to set them again as soon as I have my speed-quality set up perfectly. Stay tuned!
(link: http://www.makerbot.com/sailfish/tuning/ with some explanations of above mentioned values and calibration methods)

[curious aardvark]

I honestly can't see any difference.

But when you've got things sorted out - I'll probably copy your settings :-)

[Mjolinor]

But when you've got things sorted out - I'll probably copy your settings :-)

Chinese ancestors?

[curious aardvark]

Chinese ancestors?

Just intelligent ones :-)
Using what works is just common sense and if someone has more time to work that out than I do - only an idiot wouldn't use that knowlege.

And now I look, I can see the waves, and they are only at one end of the print. So the acceleration thing makes a lot of sense. Once the head gets up to speed the waves stop.

I don't envisage it ever being an issue visual wise, with what i'm liable to make - but anything that increases precision of prints without massive effort on my part, has to be a good thing :-)

[Mjolinor]

I don't think it is related to the head coming up to speed at all. The waves will still be there but reduced as the overshoot is reduced because it reaches the corner more slowly. If it is doing an X move then the wave is created by the previous Y move. It is a property of everything in the universe be it central heating controllers or speedometer needles on cars, they all suffer from it and it cannot be cured, only minimised.

The proper way to minimise it would be with PID controllers on servo drives that have feedback. That would be one expensive printer.

[Dargonfly]

I'm trying to find some good settings (will update later this weekend) but I was wondering if I could calculate 'perfect' settings based on the hardware.
So perhaps Mjolinor knows this one ('cause you seem to know a lot of this technical mumbo-jumbo)
How often per second does the motherboard / controller talk to the stepper motors?
What's the smallest step the steppers can make the extruder move in X Y directions?
What's the smallest length of material that can be extruded in one step?

I'm thinking I could use that to calculate which values allow for no floating point values going to the stepper which perhaps allows some small improvements?

[Mjolinor]

What it is possible to calculate has already been calculated.

The oscillation is partly mechanical, it is stretch in the belts and movement in the end pieces caused by the impossibility of going from moving to stopped in zero time. Some of it is electrical but for exactly the same reasons, it is not possible to go from current flowing to no current flowing in zero time.

If you google "Ringing" there are lots of explanations of what it is and what causes it electrically but you will also see that it is applicable to everything including any mechanical, hydraulic or pneumatic system.