3 minutes ago, rcfocus said:There must be other problem. Since your UM2+ is pretty new, how about sending it back to distributor for service?
I bought it used from one of the community members.
3 minutes ago, rcfocus said:There must be other problem. Since your UM2+ is pretty new, how about sending it back to distributor for service?
I bought it used from one of the community members.
Well...
You said "all belts are tight, the axes are aligned and bed is leveled".... please ensure the belts are not too tight. Please also ensure the short belts connected to X/Y motors are not too tight, not too feeble.
Lubricate all axes: the Z lead-screw, two Z rods, two X rods, two Y rods.
According to my experiences, you can also try to calibrate the X/Y pulleys. To move printhead in X direction, the two X belts are rotated at the same time by four pulleys. It is possible that the pulleys loosen over time. To ensure the two X belts are synced, we have to ensure the four pulleys are aligned (don't know how to express this!). For X axis, there are two rods and the two belts. You can loose the two screws on one X rod (rod#1). Please also ensure the the screws of pulleys on the other rod (rod#2) are well screwed. Then, tighten the two screws of the pulleys on rod#1. Now the four pulleys are aligned.
For Y axis, the structure is the same. So, do it again for Y axis.
Please also check the screws of X/Y motor pulleys. You have to ensure all pulleys are tightened.
Edited by rcfocus
Thank you - I tested all belts and they are not too tight nor loose. I had not checked the pulleys so did that too. When looking closely at the surface defects I can conclude that they do not seem like from X-Y motion stage. In other words, I am seeing 'fat' layer vs thin layers that extend in all directions. If it were a X-Y issue (i think) the defect would be shifted.
What you said "I am seeing 'fat' layer vs thin layers that extend in all directions." This could be caused by a dirty Z lead-screw. When a machine is used for a very long time. The Z lead-screw will be very dirty. Please fully clean the lead-screw and then lubricate it again.
But from the picture you posted, the lead-screw seems not dirty. Anyway, just try it and see if any difference.
Btw, a used hot-end will be very dirty too. You should either use a new nozzle or clean it up at first. I remember there is an article showing how to clean the nozzle (hot-pull & cold-pull).
Edited by rcfocus1 hour ago, rcfocus said:What you said "I am seeing 'fat' layer vs thin layers that extend in all directions." This could be caused by a dirty Z lead-screw. When a machine is used for a very long time. The Z lead-screw will be very dirty. Please fully clean the lead-screw and then lubricate it again.
I am going to clean it fully and re-lube with Magnalube and report back.
QuoteBtw, a used hot-end will be very dirty too.
I actually just installed the UM2+ Extrusion Upgrade Kit with the Olsson block on Friday 🙂
Out of curiosity I printed the mini speed test. See below. I notices that the inconsistent layer lines are a LOT LESS at higher speeds (like 70-80mm/sec). This gave me a new hypothesis: Maybe the axis need re-lubing and at high speeds the momentum overcomes certain resistance. In addition, based on @rfocosta advise, I thoroughly cleaned and re-lubed all axis and also ensure the axis are aligned and belts are appropriately tightened. This unfortunately, did not change anything. 😞
Second New Hypothesis: While re-lubing the axis I noticed that when I move the head by head in X-axis, instead of making a nice consistent zeeeeeeee sounds, it makes, zeee000zeeee0000zeeee000. Visual inspection shows that the back right X-pulley has a very slight runout. I guess this is because it is not centered on the 8mm rod?
What I fail to understand is that if this X-axis pulley was the source of inconsistency, why do I see it on all 4 sides of a cube and not just in the X-axis motion direction?
I think effects like this might also be caused by fluctuating temperature, because that changes viscosity and flow-rate (=less back-pressure when more liquid). And maybe also by a worn-out white teflon coupler?
But I don't know a good means to reliably measure the temperature from the outside? A thin and heat-resistant sensor that you could move into the nozzle from above might be best? (Of course after removing filament, removing bowden tube, and cleaning the nozzle.) And then let the printer run, "printing" this same object. Thus it goes through the same moves, same temperatures, same heated bed, but only without filament?
5 hours ago, geert_2 said:I think effects like this might also be caused by fluctuating temperature, because that changes viscosity and flow-rate (=less back-pressure when more liquid). And maybe also by a worn-out white teflon coupler?
But I don't know a good means to reliably measure the temperature from the outside? A thin and heat-resistant sensor that you could move into the nozzle from above might be best? (Of course after removing filament, removing bowden tube, and cleaning the nozzle.) And then let the printer run, "printing" this same object. Thus it goes through the same moves, same temperatures, same heated bed, but only without filament?
Yeah, the underlying issue does seem to be what I would generally call 'extrusion variability'. The reason I think it is not the 1) Extruder assembly 2) Nozzle 3) Heater Block or 4)Heater Catridge/Temperature sensor because before I had upgraded to the UM2+ kit, I had the same issue. The UM2+ kit basically is brand new extruder and the whole head assembly.
This leaves the mainboard (which could not be sensing the temperature accurately) or something about the motion system.
Edit: BTW, a good way to measure the nozzle without taking it apart would be with a thermal cam.
Edited by nighthowlersLong ago there has been a discussion if the heated bed could be the cause? If it would draw so much power that the ground-level (zero volt) would shift up, and if the temperature sensor would use that same ground wiring, this could cause errors in the temperature the main board "sees". And then it would adjust incorrectly.
But I don't remember exactly on what printer models this was (maybe UM2?), and what the conclusions or solutions were. Maybe you can find that again. It was several years ago.
If you could print the same object on a cold bed, using glue, maybe you could rule this out? Then at least any ground-level shifting can't come from the bed. It could still come from the nozzle-heater, but that draws less power, so should have less effect. If this would be the cause, of course. Could be something entirely different also.
18 minutes ago, geert_2 said:Long ago there has been a discussion if the heated bed could be the cause? If it would draw so much power that the ground-level (zero volt) would shift up, and if the temperature sensor would use that same ground wiring, this could cause errors in the temperature the main board "sees". And then it would adjust incorrectly.
Oh, that's a really interesting thought. Will test!
Hi nighthowlers,
You have some problems here.
First, forget about checking the heat bed common ground theory.
(Both, the heater and the PT100 sensor have separated ground, so a construction like that would give massive error
(not 50 deg. C). In such an error the printer would stop instantly (hi priority issue).)
This is an important finding:
Second New Hypothesis: While re-lubing the axis I noticed that when I move the head by head in X-axis, instead of making a nice consistent zeeeeeeee sounds, it makes, zeee000zeeee0000zeeee000. Visual inspection shows that the back right X-pulley has a very slight runout. I guess this is because it is not centered on the 8mm rod?
I've built my UM2 and have a plastic bag full of pulleys, that is not up to good standard, as they have some wobble. In your case, you may try to shim up the pulley using "shim" of brass.
Usually they are oval at one side, for testing you may use paper and then tighten the set screw.
Now a question; did you rice the nozzle temperature when you made the "circular tower" flow "speed" test?
IMO. I think your model look better at the lo part of the model, however this latter one look much better than the first model.
During normal printing, how much does the nozzle temperature fluctuate up and down? (should be ±1 deg. C)
Edit: Why is your first model white at the bottom or top?
What do you think?
Thanks
Torgeir
Edited by Torgeir
One thing that I did not mention, due to this wobbling the extruder will move a little back and forth in the X-axis.
Thanks
Torgeir.
1 hour ago, Torgeir said:Second New Hypothesis: While re-lubing the axis I noticed that when I move the head by head in X-axis, instead of making a nice consistent zeeeeeeee sounds, it makes, zeee000zeeee0000zeeee000. Visual inspection shows that the back right X-pulley has a very slight runout. I guess this is because it is not centered on the 8mm rod?
I've built my UM2 and have a plastic bag full of pulleys, that is not up to good standard, as they have some wobble. In your case, you may try to shim up the pulley using "shim" of brass.
Usually they are oval at one side, for testing you may use paper and then tighten the set screw.
I ordered two new pulleys from a Ultimaker supplier (fbrc8.com) and will change it. I think this could be playing a role.
QuoteNow a question; did you rice the nozzle temperature when you made the "circular tower" flow "speed" test?
IMO. I think your model look better at the lo part of the model, however this latter one look much better than the first model.
No, I used the GCode from a community member (from a file called Mini_Speed_Test in the Bondtech Extruder thread). I checked and the temperature is never changed in GCode. The lower layers may look better (and they are) but they suffer from the same "micro extrusion variability".
QuoteDuring normal printing, how much does the nozzle temperature fluctuate up and down? (should be ±1 deg. C)
Edit: Why is your first model white at the bottom or top?
The temps shown remain 209-210. So I would say yeah, between +/- 1 deg. C. Its not white, its just over exposed as I was using a macro lens and took the image in sunlight.
Okay friends, it looks like I am a LOT clsoer to the source of the problem. 😄
3 hours ago, geert_2 said:Long ago there has been a discussion if the heated bed could be the cause?
I think this is it!! Although I did not follow the floating ground aspect, my guess is that the PID needs better tuning. When the bed fluctuates power, it is causing micro-variations in the nozzle temperature. What bothers me is that such a minor temperature variation can cause this. I am in process of running an extensive log of temperatures and will report back. Thank you @geert_2 for bringing this on my radar!
2 hours ago, Torgeir said:First, forget about checking the heat bed common ground theory.
(Both, the heater and the PT100 sensor have separated ground, so a construction like that would give massive error
(not 50 deg. C). In such an error the printer would stop instantly (hi priority issue).)
@Torgeir, so yes, it is not related to ground, but looks like it is dependent on the heated bed being operational or not.
Note how much cleaner the edge is with bed heat OFF. Also note, the surface when reflecting light is a lot more consistent.
When looking in person (sorry hard to capture in photos), there is about 70-80% improvement in surface quality and the micro-extrusion variation.
This also explains why I saw slightly less of such variation on higher speeds (70-80 mm/sec), the reason being, the temperature variations are happening at a certain slow rate. When the print is sped up, the printer goes through many more layers before the temperature variation (or whatever the heated bed causes) has a change to impact it.
I am going to monitor and log detailed temperatures and try and tune the PID. If that doesn't help, I am gonna see what the underlying reason is and make a modified circuit possibly to separate any power impact of heated bed on rest of circuitry. I might run a separate power supply or add filtering circuitry.
Edited by nighthowlersHi nighthowlers,
Great finding.. 🙂
Also bear in mind that the power supply is an important one in here, this as you reduced the hard load of power supply might have an impact. But just go on, you're close I'm sure.
Way to go.
Torgeir.
Forgot this one, I'll think the later firmware use "bang-bang" mode not PID, cause PID created lots of harmonics radiation not allowed for devices like this. Also PID can make distortion to sensors in it's own system, right, -it disturb itself..
Just like an audio amplifier starting to squeak if the mic come to close to the speaker..
I.E. The high current pulses from the heat bed wires act as an antenna radiating high energy harmonics all around the printer.
Well, -can happen just like that, possible, -but just theory..
The electronic things can really make gray hair sometimes. 😞
Thanks
Torgeir
12 minutes ago, Torgeir said:Forgot this one, I'll think the later firmware use "bang-bang" mode not PID, cause PID created lots of harmonics radiation not allowed for devices like this. Also PID can make distortion to sensors in it's own system, right, -it disturb itself..
Just like an audio amplifier starting to squeak if the mic come to close to the speaker..
I.E. The high current pulses from the heat bed wires act as an antenna radiating high energy harmonics all around the
That's right, but I think the PID loop is pretty slow, of around 7 Hz, so EMI isn't an issue as much. If really, I can shield the cables. Fortunately, electronics is in my wheelhouse so I will figure this out and report back for posterity in case someone has these issues.
Okay, so I went with the hypothesis: that the bed takes up plenty of power and that causes temperature variations in the extruder. As a result we see micro extrusion variations.
Experiment
The idea was to run a print job with and without BED_HEAT and conclude if
I edited my Gcode for Cube test such that mid-way into the print it would turn the BED_HEATER to OFF. I I collected temperature data as frequently as Marlin would print to Serial. I wrote a quick analysis script to collect results from multiple runs and compute stats. See below.
Conclusion:
# HOTEND TEMP ANALYSYS Average difference from set point (all range): -0.045 degC [BED ON] Average difference from set point: -0.054 degC STD 0.757 VAR 0.572 [BED OFF] Average difference from set point: -0.036 degC STD 0.524 VAR 0.275
Dammit. 😞🙁
New Hypothesis: BED_HEAT is causing motion artifacts, position inaccuracy or extruder motor motion artifacts. It could also be an issue only on Z-axis? Cause that moves infrequently and when it does needs inrush current. There also seems to be a 'pattern' that likely mimics the ON/OFF of BED_HEAT.
Few ideas to test this:
Any other ideas why bed heater could be causing the lines in print?
Edited by nighthowlersNo idea. But as an engineer, I always check the power supply stability for boards (systems) with unknown issues. So, you should check the 2nd item, +24V, before other items.
The extruder temperature you measured is based on the serial output of Marlin. But the firmware doesn't know that the grounding level is dynamically shifted. So, the firmware (PID or other) will still try to make the temperature reading stable while the actual temperature is drifting.
I mean, you can NOT depend on the serial output. You have to use an external device to measure the nozzle temperature.
Hi there,
Sorry, but I do not think this is caused by temperature variation, the lines is to consistent so it has something with the X/Y axis -stepping error.
Actually, it might be the missed micro step problem, the stepper keep synchronized at every "full step", but can have micro step offset. I've never seen as much as this, but it is possible for sure.
This problem can be seen on some of the printed object we do, some more than others, depending on object slicing and printer settings. So many things we all know..
This may be caused by the stepper driver circuit and probable issues with this arrangement, -maybe this become an issue when the voltage drop to a critical point? Current will also drop..
However, I've never checked this as I'm using a power supply that deliver 12.3 Amp. 24 VDC.
Sure, electronics can surprise..
You can have a look at this tread:
Why does my print have tiny zebra stripes?
Joined this tread in page 5. In here you can see the kind of modification I’ve made on the main PCB in order to get read of most of the “sebra stripes” that was visible on some printed object.
I did this; "shortening of the ROSC (reference oscillator) adjusting resistor" modification, in order to avoid missing steps caused by lack of back EMF.
By pulling the ROSC pin to ground, mixed decay is set to be active 100% of the time, for both rising and falling currents, and prevents missed steps. (Allegro datasheet advice)
You can try this; just short the two ROSC resistors for X and Y by using a short link bridge across them.
(Other link to GND may cause positive feedback, -this is a warning from Allegro factory.)
Hmm. -Well, you got something to try..
Good Luck
Torgeir
9 hours ago, rcfocus said:The extruder temperature you measured is based on the serial output of Marlin. But the firmware doesn't know that the grounding level is dynamically shifted. So, the firmware (PID or other) will still try to make the temperature reading stable while the actual temperature is drifting.
I mean, you can NOT depend on the serial output. You have to use an external device to measure the nozzle temperature.
Hi rcfocus, that is a great point, and I will measure the V rails to make sure.
BTW, I am not sure what you mean by "ground level shifting". That is a term I've never come across and unfortunately unable to interpret. GND is simply a label reference for measurement purposes. In such a floating circuit what matters is the net potential difference, which should be 24V. Maybe what I interpret by ground level shifting is that when the power budget is exceeded, the power supply becomes current limited and is compensated by a reduction in voltage (i.e. 24V drooping).
Even if that is the case, I am not sure how it affects the PT100 measurement (it might but seems unlikely to my brain).
The PT100 temperature sensor is a differential sensor that is being amplified by the INA826, whose output is then sensed by the ATMEGA ADC. All of these are running off a +5V supply which is generated by the buck converter. Droop in +24V should not affect the 5V supply (unless supply falls below 18V -- A4403GEUTR has an input operating range of 9 to 46V).
So two things:
1. Check how much the 24V power supply, if any is changing
2. Measure EMI noise or such on 5V rails
4 hours ago, Torgeir said:Hi there,
Sorry, but I do not think this is caused by temperature variation, the lines is to consistent so it has something with the X/Y axis -stepping error.
Actually, it might be the missed micro step problem, the stepper keep synchronized at every "full step", but can have micro step offset. I've never seen as much as this, but it is possible for sure.
This problem can be seen on some of the printed object we do, some more than others, depending on object slicing and printer settings. So many things we all know..
I did this; "shortening of the ROSC (reference oscillator) adjusting resistor" modification, in order to avoid missing steps caused by lack of back EMF.
By pulling the ROSC pin to ground, mixed decay is set to be active 100% of the time, for both rising and falling currents, and prevents missed steps. (Allegro datasheet advice)
Yeah I think this is some sort of stepping issue as well that happens as a result of current deficit. Maybe. I will look into ROSC, thank you for all those pointers!
Edit: @Torgeir: I just read through that discussion about missed microstep. That is very interesting. I will try out the ROSC mod.
Recommended Posts
rcfocus 0
Z lead-screw wobble is fine. It won't affect X/Y precision. In fact, it is almost impossible to make a perfect lead-screw. So, additional axis rods must be used. The two big rods (at the left and right sides) parallel to the Z lead-screw is used for this purpose. They ensure the precision of X/Y direction.
There must be other problem. Since your UM2+ is pretty new, how about sending it back to distributor for service?
Link to post
Share on other sites