Makerfarm iv3 z-axis lift speed

[hernejj]

OK, after seeing Adamfilip's awesome HulkXY machine I am very jealous of the quick z-ais lift speed he has achieved. I was wondering if anyone knows why the iv3 has such a slow z-axis lift, and if it is possible/easy to change it?

I assume I can either change the speed/acceleration profiles within the Marlin control panel, or maybe even the code. Failing that, perhaps we can back off the microstepping on the z motors a bit to achieve more speed and torque?

I'm hoping that a faster z-axis lift will result in slightly quicker print times and less blobbing at the lift location.

[printbus]

Some of your questions are exactly what drove me to start Marlin Motion Related Configuration.h Settings for MakerFarm i3v. If nothing else, post Testing Z axis travel with gcode might at least be helpful - it includes some gcode snippets that you can use to test Z-axis settings with.

In a nutshell, I found the most significant improvement in z-axis movements were obtained by increasing the Z-axis acceleration from the MakerFarm default of 5 mm/sec per second to 500 mm/sec per second, and increasing the allowable jerk in Z axis to 10 mm/sec. I did find I could increase the z axis feedrate some, but not a lot. I found that the increases I could obtain by raising the z axis feedrate and reducing microstepping were somewhat marginal compared to the acceleration and jerk adjustments. Think about it - the default z-axis feed rate is 2mm/sec, but an acceleration term of 5 mm/sec per second says "take a long time to get to that 2 mm/sec feedrate". And this is only half of it! The acceleration rate is applied for deceleration as well. M5 rods rotate just a quarter turn to get you a 0.2mm layer shift. It'll spend nearly all, if not all, of that quarter turn accelerating and decelerating with the default settings. You've got to increase that acceleration setting to maximize the benefit from anything else you're trying.

The big z-axis difference between the OEM i3v and AdamFilip's printer is in the threaded rods. Makerfarm uses M5 threaded rods, with a pitch of 0.8mm per thread rotation. Adam is using ACME rods - probably 4-start ones with a lead of 8mm. This means that for a single rotation, he'll get 8mm of linear movement comparted to the 0.8mm in a standard i3v. Yes, in theory that means he'll be able to drive the z-axis pretty fast. ACME rods aren't a panacea though - they bring new issues like backlash and torque concerns that need to be considered.

I also found that my limitation in the i3v was in lifting the x-carriage. I could move the x-carriage down considerably faster than I could move it up, easily explained by the weight of the x-carriage and M5 thread resistance being tough for the NEMA17s to deal with. I was tempted to experiment with, but never did, a counterweight system with a simple wheel at the top of each side of the printer. The concept was that if I could take some of the x-carriage weight off the Z-axis rods, I could likely be able to drive the z-axis upwards 2 or 3 times faster that what I was achieving. EDIT: I guess I never got around to that because after seeing how much faster the z-axis layer shift moves were with just the setting adjustments, I saw little point to it.

[hernejj]

Printbus,
Wow! I'm amazed. I completely missed that thread. I'm very glad you pointed me to it. Thanks, once again, for this thread and for all you provide for us within the MakerFarm community.

I'll need to find time (hopefully tonight) to go over your thread in detail but just from skimming it I'm getting a good idea on where to go from here.

My iv3 currently has the stock 12BHH48 motors. I also happen to have some Nema 23 motors hanging around from a cnc router build that I have yet to even start on . I wonder if we could crank up the z axis feel rate with these motors? Even if so, I doubt the benefits would justify the cost of two Nema 23s.

[printbus]

That thread is brain-buster. I admit it. But after all, there IS a lot to consider as far as the internal workings of these machines. At least jump to the post on Printbus settings summary if you want to save a lot of time and just get to a set of settings to try. The rest of the thread is background research and testing that got me to those conclusions.

I'm still amazed that there's been little input from anyone else on how they were able to optimize the motion related settings on their printer.

Start with the acceleration and jerk settings on the Z-axis. Then I highly suggest setting the stepper motor drivers based on functional performance and not a voltage setting. You, as others have been, may be quite impressed with your *new* printer after that.

The fine pitch of the M5 rods is a driving factor in how fast Z can move. Stepper motors have a limit on how fast they can be rotated. Marlin has limits on how fast it can issue steps. The amount of microstepping plays a role. My point is there's a number of things to consider - more than just swapping the motors out. My main gripe remaining with Z was how long it would take to home after a previous tall print. In addition to contemplating a counterweight scheme for the x-carriage, I also thought about modifying Marlin so that I could take advantage of the weight and lower the x-carriage at a faster rate, leaving the upwards movement at the slow but improved performance.

[lakester]

Firstly, thank you!

It's funny..., when I was transcribing Makerfarm's stock config settings to Marlin 1.1.0, the Z settings really stood out. I'm not sure I correlated them with any particular artifacts I may be seeing..., they just seemed SO SLOW. Nevertheless, I didn't play with it.

One issue I've never really been able to overcome..., the one that it seemed no matter what slicer settings I used I'd get basically the same results, relates to retraction (or quasi lack of it).

This discussion I think has given me some new ideas, and the confidence to try some stuff out. So look out Z accel/jerk settings..., and I'm coming for you extruder as well.

Thx again!

[printbus]

One issue I've never really been able to overcome..., the one that it seemed no matter what slicer settings I used I'd get basically the same results, relates to retraction (or quasi lack of it).

I don't know if MakerFarm has changed newer versions, but the default config used to set the extruder term in DEFAULT_MAX_FEEDRATE to 22 mm/sec. Normal extrusion wouldn't be close to that, but the printer will try to hit that during retractions. That's a bit fast.

My first major settings improvement was when I followed Clough42's recommendation of 10 mm/sec. Experimenting with the setting, I eventually ended up at 15 mm/sec. DEFAULT_RETRACT_ACCELERATION is something else that could be played with, but the default acceleration value of 500 mm/sec per second seemed fine used with the lower feedrate.

These values apply to a Greg's Wade extruder. Extruders with other motor types or gearing would require experimenting. There are snippets of test gcode for the extruder motor in that brain-buster of a thread too.

EDIT: I'll add that learned a lot about my printer and retraction settings in printing the Fine Positive Features object in the Make:2015 shoot-out models. The Fine Positive Features model is the one with nine spires to it. I think it is a good model for retraction testing since it doesn't require a lot of filament, prints fairly quick, and involves a high ratio of retractions to normal extrusion.

[hernejj]

I flashed my firmware to the latest Marlin (1.1.0-RC5) last night and updated my motion parameters exactly as Printbus suggested in this post:
http://3dprintboard.com/showthread.p...ll=1#post41192

The results are quite pleasing. My z-axis is quite snappy now. All axes move and home noticeably faster. In a timed test of a single cable chain link the new settings are about 8% faster with no noticeable difference in print quality.

I'm a little concerned with the increase in jerk particularly. Before now, when ever I tried to print too fast (70-100 mm/s) The nozzle tended to "pull up" previously laid down filament strands while rounding corners. I assume the issue is the pressure of the filament being pulled sideways before it had adequate time to cool. With a higher jerk and faster acceleration I wonder if we will actually see a reduction in the maximum print speed.

- How fast can we print with default jerk/accel settings.
- For the exact same print, how fast can we print with the faster settings?
- For both experiments, what is the actual time required, and which is ultimately quicker?

We would want to choose a model with some steep corners to exacerbate any potential problem areas.
Should be a fun experiment to try out.

[printbus]

Glad I could kickstart your pursuit of improved settings!

On the jerk & acceleration, a counter-argument can be made that the slower the nozzle moves through a corner (or a z-increase like for a layer shift), the more risk of heat soaking from the nozzle to the previously printed layers, softening them up. Of course, adding some print cooling airflow for PLA can affect all this as well.

[lakester]

Have started the experiments. Using "fine-positive-features" as the test piece.

First test is with the super-slow profile I use for my pottery cookie-cutters. The next will be with that profile, sped up a bit..., past what I felt was the excessive caution of my standard set-up, but still pretty slow. I.e., first two prints are tweaking slicing profile only.

After that, will start working on the configuration settings.

The comments regarding cooling resonated with something I was thinking earlier today. In particular, it would be interesting to enable the fan as the upper portions of the "spires" are printed..., since I suspect a part of the problem is heat soaking of the ever smaller volume of material towards the top. I'll tweak the fan manually first during a print, and if that produces an useable results..., will see if a useful parametric setting can be found.

[lakester]

Reporting back, with a slightly off-topic follow-up. So...

I too have been running w/ Marlin 1.1.0RC5 with printbus' config settings. Two main observations:

1. The primary advantage to this has been the preservation of overall print quality at higher speeds. So on that basis alone: "win".

2. Slight, but repeatable improvement of issues I associate with retraction problems. I THINK this was the result of the improved z-lift performance (in addition to the more aggressive extruder jerk settings).

I did play around with cooling a bit. The first experiment was with "fine-positive-features". About the only thing cooling did was exacerbate some spider-webbing. Tried a couple of different approaches, but no real improvements.

Inspired by something in another thread, I decided to try printing "the cube" with 0% infill..., i.e., hollow.

FINALLY..., a scenario where cooling actually made a demonstrable, positive difference! (a first for me)

The layer times with a hollow cube were so low, that when the head advanced to and printed the next layer, the previous layer had not cooled sufficiently to support the new layer. The problems were most visible at the corners.

Increasing layer times by adding back infill OR turning on the cooling fan solved the problem.

The frustrating thing about the fan is that it is a rather blunt instrument. It is difficult to generalize and generate parameters that actually create intelligent, reliable use of the thing.

With "fine-positive-features", I assumed that dwell time on the tiny spires was a problem that the fan could help. I was imagining exploring the idea of driving the fan based on some kind of analysis of inter-retract times. Not sure I'll play with that or not, given the negative results I got with several different fan settings at a gross level.

On the other hand, the fan SEEMS to be very useful for thin-walled layers with rendering times under 10-15 sec or so. Not quite sure how I'll generalize this..., but it's something I'll keep in mind.