From the original Synthetos forum (2011–2025)
I'm using tinyG to print ceramic paste. I've recently begun running prints that run for as long as 24 hours. There seems to be some scaling shift in the extrusion rate after about 18 hours. The trouble seems to start at about the same time (but not exactly - so I don't think it is the gcode).
Here is a link a timelapse video of the issue - notice when the thickness shifts dramatically at about 2/3 of the way through.
https://drive.google.com/open?id=1f2XsBDeJBZB_1GIsos6kXTSWd8oGsHjb
This might be helpful. Here is a link to my current settings as a .txt file
https://drive.google.com/open?id=1A-CAQ8cWxuO5nKYzZCyJUVJiYLaFLFno
That's a pretty cool project... would be nice to see part of it in normal speed. Curious how often you have to replenish your ceramic paste supply... could it be as simple as a difference in the mix.
Thanks for the reply yawstick. I'll try to get a real time video but it will be a couple of days.
The tank is big enough for an entire print - I had the same thought about the viscosity changing from settling so I recently built an agitator for the tank. But the trouble remains.
The pump is iterative and squeezes out droplets - the droplets per linear distance nearly double at the point of 'trouble'. So the pump appears to be cycling (nearly) twice as much as I'm asking it to.
Is the pump driven by a stepper motor?
Yes - the pump is driven by a nema 17 stepper (mapped as A, m4)
Here is a link to the Gcode in question - one thought is that this is the longest build that I have run and the extruder values are quite high... Is there a limit to how large the values can be for tinyG to interpet?
https://drive.google.com/open?id=1ErkHDe21Id3d8xbWbCtr_cfjwjFO_ZIO
Interesting Project.
I see nothing obvious in your settings, but must admit we see very little rotational axis examples.
What do you use as a Gcode sender?
Do you see anything in the status reports indicating an increase in the rotational velocity as the material delivery increases?
Like the others, I have to wonder about the viscosity of your material changing over a long period of time, perhaps due to the shaking?maybe the slurry separates over time, with the heavier material depositing slower, then a more liquid flow occurs during the later stages?
Perhaps try running and manually keeping the reservoir at approximately the same level of fill?
The trouble seems to happen shortly after the extruder value exceded 25k...
Thanks cmcgrath5035.
I've ruled out a viscosity shift as the source of the problem because in my process, the 'droplets' are created by one pump cycle, which is one complete revolution of the stepper controlling the pump. When the trouble arises, the amount of droplets per linear distance approximately doubles. Meaning that it is not simply an increase in flow but rather an increase in pump cycles.
I'm sending the code with CoolTerm - and recently began sending the code from a Raspberry Pi 3 with CoolTerm Linux (which I prefer because it frees up my laptop). There are no signs of trouble, but I have to admit that I rarely understand what is being spit out in the CoolTerm status scroll. I suppose it would be interesting to try sending it from my windows PC.
Sure sounds odd.
Are you sure it's double the material, rather than half the X,Y velocity?
Is it pure linear Gcode, or have you crafted in loops or "macros"?
CoolTerm on RasPi should be ideal for this.
Have you tried making just one item to see if issues is Z related rather than duration of job?
It seems like you may be on to something with the a axis getting too big... unlike the other axis that go back to home or at least near to it... sounds like your a axis just keeps getting bigger and bigger. I'm not a power user or anything but would it be possible to reset the a axis after each stroke so your number doesn't get so large.
Long-winded. Apologies, please bear with me -
I'm becoming certain that the issue is either my tinyG settings or something that is in the firmware... I haven't seen any fully operational 3d printers utilizing the tinyG, only short clips of initial tests. if anyone is doing it I would love to get a glimpse of their axis settings.
Thanks cmcgrath5035 - the x,y velocity is constant (plus the pump is stepper controlled), and there are no loops or macros in the code.
Thanks yawstick - I rebuilt my grasshopper definition yesterday so that the extruder value resets to 0 after every curve and the extrusion issue has manifest itself again in an identical pattern.
It looks like my squirrely extrusion rate is exactly repeatable. I have essentially verified that there is nothing in the gcode that should cause this. (A friend ran a simulation with the code that would have shown a change in extrusion rate).
I'm not able to verify that it is not an issue with Coolterm sending - I've added line numbering to my gcode this morning and should be able to run code tomorrow that will allow me to trace the placement errors, if there are any...
I fired 2 vases from separate runs of the same code (links to pictures bellow). I counted the layers to confirm that the rate shifts at the same place - there are actually 2 rate shifts: one at the 24th layer and one at the 56th layer. 1 droplet = 1 cycle of the pump. You'll see in the images that the amount of droplets per linear inch varies at these points:
starting (bottom): 3 cycles per inch
middle: 2 cycles per inch
top: 4 cycles per inch
https://drive.google.com/open?id=1XMMTOHm81IH_v9_QFHlidlF0K5G8V38h
Interesting... just looking at the pics it appears to be related to the size of the vase... meaning as it gets smaller in the middle section... it appears the ratios change between the axes. There are lot of unknowns here... it appears that each stroke of the pump is small at the beginning and end... and fat in the middle of the stroke. Is that by design or the nature of the pump. Without seeing it in real time... kind of shooting in the dark. Probably shooting in the dark regardless ;)
Thanks for staying interested yawstick.
Here's a link to a video of the whole setup.
https://drive.google.com/open?id=1pqqo1DqOCpieuAcskBBuQdGygUudnrld
You'll see that these are all parallel walled forms so the linear distance of each layer is identical - this is going to slightly confuse the subjects, as the eratic patterning of the form is created by dropping the pump stepper voltage with the onboard pot until enough steps are skipped to disrupt the the regular pattern (otherwise the layers would build like a brick wall and simply be less interesting).
I don't think this is current related because it wouldn't be repeatable. On my current tests the pot is turned up enough that there don't appear to be missed steps.(regardless, missed steps would not manifest identically in two separate runs of the code).
I'm printing vases that increase in diameter at the same time as those which reduce in diameter - so the linear distance of each layer is 'relatively' constant.
To generate the extrusion code with rhino/grasshopper - I'm turning curves into a series of points which are spaced 5mm apart all along the curve. With my current flow rate, the pump should complete 0.625 cyles per 5mm, and this stands up if you review the gcode. Doesn't stand up in the print.
It is very interesting what you are doing.
With regards to yawsticks observation that pump motion seems to be slow-fast-slow, that is likely due to jerk - the concept of jerk is to manage momentum as smoothly as possible; a move form stop accelerates slowly at first, then acceleration increases to achieve the set max velocity, then the motion decelerates on a similar curve. You could try to modify $ajm to see how that affects results.
Some comments relative to your manipulation of stepper A drive current to induce waves. The on board pots adjust the amount of power delivered to the stepper, not the voltage. The stepper driver monitors the current as it ramps up in the motor windings and cut off pulses of full voltage when the set current (by the pot) is achieved. The current thru an inductor (the winding) changes based on the drive impedance of the stepper driver and the inductance of the winding. I would expect operating, as you are, right on the threshold of missing steps, would produce rather random results and be subject to thermal issues, since the stepper will tend to miss more steps when driven by marginal currents at high temperatures. That does not explain the repeatability of your erratic results.
This topic is from the original forum (April 2018). Your reply starts a new discussion linked to it.
Back to TinyG Support or latest activity.