Core One: Belt tuning temperature dependency
After keeping belts in tune at room temperature (as most people will probably do) I repeated belt tuning immediately after a 12h print at 55°C chamber temperature where the printer was thoroughly "heat soaked".
To my surprise, I had to loosen screws 3/8 turns, which is a significant change. So it seems the metal is expanding more with temperature than the belts.
My own conclusion (printing mostly ASA at 55 deg) is to aim for the low end of the pitch range at room temperature in the future.
RE: Core One: Belt tuning temperature dependency
Kevlar, which I understand to be the reinforcing material in the belts, actually has a negative coefficient of thermal expansion. Hence the belts will contract (slightly) while the metal frame expands.
RE: Core One: Belt tuning temperature dependency
I noticed that aswell, tried for fun at 7-8deg which is the workshop idle temp during winter, got very high values but didnt change anything based on that, will only do it at room temp.
RE:
Also noticed that, even at just 30°C chamber temperature. At 20°C the belts where at ~93Hz. At 30°C they were at 98Hz.
I wonder if it would make sense for things like Input Shaping and Phase Stepping to have multiple presets for different temperature ranges.
RE: Core One: Belt tuning temperature dependency
...
I wonder if it would make sense for things like Input Shaping and Phase Stepping to have multiple presets for different temperature ranges.
This starts to look bad at the 55°C chamber temperature ceiling - which, arguably, is even lower than many people would wish for.
Looking at my own observation, I suspect that the calibration I had at the time of my first post was done on a "warmed-up" printer: Input shaper parameters were unusually high, probably tuned after warming up the printer. And looking at some calibration cubes I printed with the original configuration, print performance clearly degraded: The front right corner was inconsistent even at 0.2 mm structural speeds in generic ASA.
After relaxing and retuning belts then calibrating input shaper, print quality improved visibly.
So yes, it does matter.
A possible conclusion is to heat soak at 55 °C, then belt tuning, then input shaper cal. This might work with manual controls and an external heater to save time, but the nozzle needs to stay cold or it'll melt the accelerometer clip.
RE: Core One: Belt tuning temperature dependency
I discovered the same thing. I'm printing a lot of PC in jet black which is showing ghosting and ringing. I have a blanket over the chamber so I'm hitting 58-60°
I'm going to do this later, preheat with a hairdryer, crank the bed to 120°. Let it soak until the chamber temp is 55+ then rerun input shaper cal.
RE: Core One: Belt tuning temperature dependency
I am also currently testing the belt tension at different temperatures in the chamber. While testing, it occurred to me that we could continue where Prusa left off.Are you familiar with the picture of the Core One from the Prusa blog, where two stepper motors with belt drives were mounted on the front adjustment screws?
Something like that, but with worm gears moved to the inside. For testing, I would use an Arduino and write a code that readjusts a number of steps depending on the temperature and also resets them as soon as the temperature drops again.
I can't say anything yet about how high the belt tension differences are at maximum chamber temperature, but I think this is a very exciting topic.
RE: Core One: Belt tuning temperature dependency
I am also currently testing the belt tension at different temperatures in the chamber. While testing, it occurred to me that we could continue where Prusa left off.Are you familiar with the picture of the Core One from the Prusa blog, where two stepper motors with belt drives were mounted on the front adjustment screws?
Something like that, but with worm gears moved to the inside. For testing, I would use an Arduino and write a code that readjusts a number of steps depending on the temperature and also resets them as soon as the temperature drops again.
I can't say anything yet about how high the belt tension differences are at maximum chamber temperature, but I think this is a very exciting topic.
How about adding a load cell and do away with belt tuning altogether? If you have the possibility to adjust on the go and measure on the go this should be a simple regulation circuit.
/Anders
RE: Core One: Belt tuning temperature dependency
How about adding a load cell and do away with belt tuning altogether? If you have the possibility to adjust on the go and measure on the go this should be a simple regulation circuit.
That would indeed be a good idea. Automatic adjustment of the optimum belt tension during operation.For the first step, I will first test how regularly the belt tension changes with increasing temperature. Here, you could use the load cell to register the changes.
RE: Core One: Belt tuning temperature dependency
How about adding a load cell and do away with belt tuning altogether? If you have the possibility to adjust on the go and measure on the go this should be a simple regulation circuit.
That would indeed be a good idea. Automatic adjustment of the optimum belt tension during operation.For the first step, I will first test how regularly the belt tension changes with increasing temperature. Here, you could use the load cell to register the changes.
Yes, gather measurements first, then decide if regulation on the fly is worth the trouble. I was also thinking to do away with the manual maintenance step but that is a slightly different use case coupling belt tensioning and homing with the complexity of square gantry.
/Anders
RE: Core One: Belt tuning temperature dependency
Yes, gather measurements first, then decide if regulation on the fly is worth the trouble
Normally, tinkering with such an expensive printer is a waste of time and money.
But I knew that beforehand and deliberately looked for improvements that we could perhaps “explore” and fix together. It's simply in my job description —I can't stop myself.
Who wants a printer that works perfectly and has no faults? 🤣 🤣 🤣
RE: Core One: Belt tuning temperature dependency
Yes, gather measurements first, then decide if regulation on the fly is worth the trouble
Normally, tinkering with such an expensive printer is a waste of time and money.
But I knew that beforehand and deliberately looked for improvements that we could perhaps “explore” and fix together. It's simply in my job description —I can't stop myself.
Who wants a printer that works perfectly and has no faults? 🤣 🤣 🤣
I understand perfectly. There is a reason I live in a house I designed and built for myself by myself. I am trying really hard to use the printer as a tool for other hobbies but these small projects keeps creeping up on me. For belt tensioning I am running a long time test, I adjusted them when swapping out tension pulleys for ones with heat inserts. I checked a week after the installation and belt tension where still same. I have left them untouched since and the printer just works. I am not that tough on print quality though since I print most functional parts for myself.
/Anders
RE:
Good idea with the automatic belt tensioner, but I'm not sure if there is risk to make matters worse if adjusted during the print.
When the belt contracts at temperature, it's likely its own "spring constant" absorbs the change - it gets tighter but actual length remains the same.
Now if I move a pulley I'm changing the geometry, and that might translate to printhead position.
I haven't really thought it through - if the pulley were halfway between stepper and print head along the belt distance, there would be no change. With the print head at the stepper end of the belt (as a thought experiment) the pulley movement would translate 1:1 to head movement. In reality I suspect we're closer to the latter than the former.
RE: Core One: Belt tuning temperature dependency
>> if the pulley were halfway between stepper and print head along the belt distance, there would be no change
edit window expired... actually it would also move in this case, so it's even worse. Question is whether the required offset would matter but I think it does.
The pulley sits on a M4 screw, correct? (or was it M5?) M4 has a thread pitch of 0.7 mm, let's say we compensate 2/8 of a turn over the print, that's 175 µm of which maybe 100 µm show at the head. This may be acceptable if the error it fixes is worse than that. For VFAs alone, I doubt it.
RE: Core One: Belt tuning temperature dependency
I understand perfectly. There is a reason I live in a house I designed and built for myself by myself. I am trying really hard to use the printer as a tool for other hobbies but these small projects keeps creeping up on me. For belt tensioning I am running a long time test, I adjusted them when swapping out tension pulleys for ones with heat inserts. I checked a week after the installation and belt tension where still same. I have left them untouched since and the printer just works. I am not that tough on print quality though since I print most functional parts for myself.
I will do as you described. When I implement a change, I will test it for a few days first to take various boundary conditions into account.
Good idea with the automatic belt tensioner, but I'm not sure if there is risk to make matters worse if adjusted during the print.
As I said, it's all just theory so far. I'm just excited by the idea and the process of putting it into practice.The first priority is to build a kind of measuring system to collect some data on how a cold and warm belt behaves.
I would have to check the data sheet for the Powergrip Gates belt to see what the length tolerances are at certain temperatures. Or whether there are any at all. I mean, these are belts that have to implement controls and precise movements.
RE: Core One: Belt tuning temperature dependency
>> if the pulley were halfway between stepper and print head along the belt distance, there would be no change
edit window expired... actually it would also move in this case, so it's even worse. Question is whether the required offset would matter but I think it does.
The pulley sits on a M4 screw, correct? (or was it M5?) M4 has a thread pitch of 0.7 mm, let's say we compensate 2/8 of a turn over the print, that's 175 µm of which maybe 100 µm show at the head. This may be acceptable if the error it fixes is worse than that. For VFAs alone, I doubt it.
I agree this may open a new can of worms and may not be worth it in the end, but if one does not try one will not learn, curiosity is a curse sometimes.
the pulley is on an M3 30mm screw, tiny, and forces are not straight but offset about 45 degrees with a support on one side to hold in position.
/Anders
RE: Core One: Belt tuning temperature dependency
I agree this may open a new can of worms and may not be worth it in the end, but if one does not try one will not learn, curiosity is a curse sometimes.
That's how it is, and there's not much I can do about it. Curiosity is there and wants answers, and the learning curve wants to evaluate..... 🙄 🤣
ley is on an M3 30mm screw, tiny, and forces are not straight but offset about 45 degrees with a support on one side to hold in position.
That's how it is. There are a few places that could be sources of error.
I'm curious to see where my findings will lead.
RE: Core One: Belt tuning temperature dependency
>> Curiosity is there and wants answers
Absolutely, and I think you're on to something here.
Question is whether it would be preferable to compensate in software e.g. switching between two input shaper configurations based on temperature. But the little voice in my head (the one with a strong allergy to "canned worms") - immediately asks, wouldn't a discrete change show in the print? So we need interpolation, and a simple idea suddenly gets fairly complex. Still, I'm sure it could be done e.g. do the math before the print (e.g. agree on an input shaper topology that works well enough for all temps) then look up parameters from precalculated tables at runtime. Oh well, just thinking aloud 🙂
RE: Core One: Belt tuning temperature dependency
Yes, it's the engineer's mind and it can't be shut off. I'm more of an amateur tinkerer and only half my ideas pan out. Listen to the nay-sayers, then forge ahead! I wonder if a heavy bi-metal strip could be incorporated somewhere to compensate. We know the CTE of the steel frame. If the belt could be measured that would be a start. Maybe the belt dominates everything.