Stepper not running smoothly above 100mm/s

I have built a delta with Ramps1.4, 0.9degree steppers and DRV8825 drivers set at 16x microstepping.
The board now runs at 14v (without diode) and stepper drivers set at 950mv.

When doing long travels at speed over 100mm/sec and also homing or just z movement, the motors run ok the first 50% of the travel,
then it seems they get out of oomph, and starts to get jerky and stuttering, it sounds like they don't know which speed they want to move at.
When they start to do that (when printing) it jerks so bad my magnetic effector jumps off..
Shorter travels is ok like 50mm, but if it has to move 100mm or more it is begging for problems.

I have tried voltages on the stepperadjustments fron 0.5-1.1v without any difference. And running them at 32step is even worse..
As long as i am sure it won't reach speeds over 100mm/s it prints great.

I can set it to 130mm/s if i am sure the distances it have to travel is not too far.

Have tried lowering DELTA_SEGMENTS_PER_SECOND to 170 from 200 original in case it was ramps not keeping up.

I will upload a video showing the problem later.

What could be my problem?

Edited 2 time(s). Last edit at 04/16/2015 06:06PM by AndersE.

If your acceleration multiplied by steps per mm is bigger than 2,000,000 you may be suffering from a bug in marlin that I just squashed. [github.com]

---

[www.hydraraptor.blogspot.com]

Quote
nophead
If your acceleration multiplied by steps per mm is bigger than 2,000,000 you may be suffering from a bug in marlin that I just squashed. [github.com]

Well done catching this bug!

I was thinking along different lines, for example I have no idea why Anders is running his steppers at 1.9A, that would seem a little bit excessive to me, and perhaps he would be pushing the 5A poly fuse on the RAMPS a little bit too far?

Edited 1 time(s). Last edit at 04/16/2015 09:02PM by AndrewBCN.

I have set in marlin max acc to 1500 so i don't think it's that. Heck original setting is 9000mm/s

But i will fiddle around with the acc settings some..

And for the amperage of 1.9 is that my steppers are rated 2.4A, and i just remembered that they loose a lot of steps below 600mv.
I have lost steps at 800mv. But perhaps it has a connection to the above problem.

Actually i had acc settings to 4000 and printing moves to 2000, i tried lowering to 1000 for both, it made little improvment. And if got as low as 500mm/s it actually got much worse.

Here is a video of the problem. When moving like this it is no big deal, but when printing it is a real problem.

Today i also tried lowering delta segments to 150 and limiting max feedrate on all steppers to 200 .. but no luck.

stuttering steppers

Edited 2 time(s). Last edit at 04/17/2015 12:50PM by AndersE.

Are you sure you have the jumpers set to 1/16th? The bottom pin is the only one closed? I had some problems with skipping and prints stopping and starting with 1/32 stepping, probably because of the Ramps 16Mhz clock.

Did you try measuring the voltage at the 12V 5A connector on the RAMPS, and then right after the 5A polyfuse (right on its leg for example)?

Also, I must say that upon further testing and from what I am reading these days in various forum posts, I am beginning to think that the DRV8825 drivers do not behave very well when (mis)matched with certain steppers which otherwise have no problems being driven by A4988 drivers, so it could be that you are unlucky and have just such a mismatch.

Yes i am. I started with 1/32 and it was much worse than now.

Quote
nubby
Are you sure you have the jumpers set to 1/16th? The bottom pin is the only one closed? I had some problems with skipping and prints stopping and starting with 1/32 stepping, probably because of the Ramps 16Mhz clock.

Do you mean measure while the steppers are running? I can try that. Maybe i should connect my scope also and look for oddities. Heh..i'm only confident with the scope when repairing old arcade boards.
I agree, i thought they were good drivers.

Quote
AndrewBCN
Did you try measuring the voltage at the 12V 5A connector on the RAMPS, and then right after the 5A polyfuse (right on its leg for example)?

Also, I must say that upon further testing and from what I am reading these days in various forum posts, I am beginning to think that the DRV8825 drivers do not behave very well when (mis)matched with certain steppers which otherwise have no problems being driven by A4988 drivers, so it could be that you are unlucky and have just such a mismatch.

You need a DSO with a lot of storage to diagnose stepper pulse train issues. I used a 15 year old logic analyser with 64k sample memory to diagnose the extruder acceleration problem and I was only looking at a 1mm move.

Maybe the step rate simply causes the interrupt routine to take so much time the foreground can no longer do the delta transform fast enough, so you get pauses between each delta segment.

---

[www.hydraraptor.blogspot.com]

Two issues to look at:

1. Time to upgrade your electronics if you want faster speeds. Arduino Mega is barely fast enough to run a delta at slow speeds.

2. At high speeds and depending on what motors you use, you may need a supply voltage greater than 12V for the stepper motor drivers, in order to get the current in the windings to change fast enough.

My Kossel rebuild uses Duet electronics, and with a 12V supply I can achieve around 400mm/sec pure Z movement speed before the motors lose steps. For pure XY movement I can get 500mm/sec, I haven't tried higher.

Edited 2 time(s). Last edit at 04/18/2015 09:08AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Yes i already have a feeler out in the controller section.
I already run the steppers at 14v (not much above 12v i know). I'll settle for what i get for now. I am going to try 1/8th stepping also before i throw in the towel.

If it was lack of stepper voltage the motors would skip steps at high speed. From your description they are just jerky, in which case I am sure it is the firmware breaking the moves into segments. If they arrive to the planner fast enough they will get blended back into smooth motion but if the step interrupts use up the CPU time such that the delta calculations are late then the planner will do a full stop at the end of the segment and then accelerate again when the next one is calculated.

---

[www.hydraraptor.blogspot.com]

nophead:

Thats sounds like accurate description of my problem. When printing at high speeds it sounds like: zzzzzzzzzzz..daa..daa..daa..daa..daa And it's the daa.daa that gets violent.
But why does it do the same for a simple homing or just z moves?

How low can one go on the deltasegments in marlin, if it would help to lower it more?

I don't have a delta so I haven't studied the code in great detail, but my guess would be it doesn't do special cases for Z only moves, so it breaks all G1 moves into segments. Not sure about homing.

Another issue is three motors are always moving fast, whereas on Cartesian machines only two move fast, so the step interrupt will be a bit shorter.

DELTA_SEGMENTS_PER_SECOND will affect quality, so I presume you can go as low as you want until you find the quality not acceptable. Also as it is per second, quality will reduce as you go faster. Or put another way to print faster with the same quality you need to do more segments per second but you have less processing time available due to the step interrupts.

A way to reduce the interrupt load would be to lower the thresholds where it switches to dual and quad stepping here: [github.com]. That might be a better option than using 8x microstepping, or reducing segments. Using 0.9 degree motors is doing you no favours here.

---

[www.hydraraptor.blogspot.com]