VC4 * 500 Gantry warping issues as printer heats
Hi, My VC4 * 500 build is pretty much completed and I have been trying to run through an exhaustive set of testing and calibrating the printer. I am coming over from a VC3.0 500 that I built in 2021. One of the big issues with the VC3.0 was the bed mesh never being stable on that printer until I heat soaked it for 30-45 minutes before I used it. I tried a number of things to get around that but the only thing that ever worked was just letting it warm up and then it went on its marry way.
With the VC4 I was hoping to get away from warming up the printer but it seems to have similar issues as the VC3 though Beacon help a ton vs the original bltouch and then superpinda on my VC3. Last night as I was doing some testing trying to fine tune my first layer on the VC4 I had it dialed down to almost perfect. I took a break for about 30 minutes and came back and planned to just try to find tune that last 0.005 in baby steps. When I tried picking up from where I left the printer it was like I forgot to save my z offset. It was completely off and I had to back off where I left off by 0.05 to get it back to clean again. The only change between those tests was I left the bed on and the front door of the enclosure. I did a few more tests and it seemed to be staying right where I left it so I called it a night.
Today I figured I would document z-tilt changes and heightmap changes as I warmed the printer bed up from ambient temperature, where it sat overnight, up to a bed temperature of 70C and then eventually a bed temperature of 100C. The plan was to start with 70C and keep measuring the z-tilt changes and heightmap changes every 15 minutes until those stabilized and then move the bed temperature up to 100C and repeat the same process.
91 Replies
For some background the printer is a VC4 500 that is fully stock besides using an Octopus Pro with a 48v power supply for the CoreXY and Hybrid Steppers. It is fully enclosed with a 3mm ACM back & sides and a 6mm polycarbonate top and front.
TL;DR at 70C it took nearly 75 minutes for the printer to fully stabilize, with most of the changes happening in the first 30 minutes. It then took another 60 minutes for the printer to fully stabilize at 100C when I moved up to the higher temperature, again as with the 70C test most of the changes happened in the first 30 minutes. If I started a print at the wrong time during that heating cycle my first layer would have been off. It may have been to much squish or not enough which could have contributed to a failed print later in the cycle.
First up with just the starting point from the printer being off overnight. I have a slight bump on the left side of my x rail/gantry that I cannot get rid of. I have removed the rail from the X at this point and cleaned the rail and the extrusion and it will not go away. I have used a high quality certified straight edge and the bed is flat and I cannot see light under it. Otherwise its not a bad bed mesh of only 0.233mm on a 500mm printer. Way better than my VC3 500 so I was happy.
The challenge is as the printer starts to warm up to a 70C bed temp. You can see the z-tilt measurements change even though the printer was never stopped. On top of that you can see the height map change. I left the head parked at 20, 427.5, 2.113 where it defaulted do after a z-tilt. This first picture is the z-tilt and height map about 5 minutes after starting to heat up the bed. It had just reached 70C when I took it. I then captured 5 more tests with the bed left at 70C and the enclosure unopened. Before each height map I z-tilted the printer and captured the changes from the previous test after the printer had leveled the bed.
Here are the 5 following at 70C
It seemed to have stabilized at 75 minutes after it reached 70C bed temp, or 80 minutes from ambient temperature. At this point I moved the bed up to 100C and continued the same procedure, z-tilt and then heightmap and documenting the changes every 15 minutes.
At this point things seemed to have stabilized. While this was going, I was researching here on Discord and found an older troubleshooting thread where Mikkel asked someone to loosen the X gantry rail while the printer is hot and then retighten it from the middle out. So I decided I may as well try this and as I was loosening the rail I felt the extrusion snap back into place. I tightened it up from middle outwards and then immediately reran the test. As you can see it was a major change and my height map went back almost to what it started with. I was happy with that and figured problem solved.
Real life, aka my job that pays for my addictions, aka hobbies, called and I got pulled away from the printer for about 5 hours. I let the printer cool down during this time and when I redid a height map it was completely bowed the other direction this time. I suspect if I went back through the previous testing, it would get to a good place once the entire printer had heated up but I would have sub par first layer prints until it stabilized again.
Finally I left the printer at ambient temp and loosen the x gantry rail again and tightened it back from the middle outwards and immediately did another z-tilt measurement and height map and the printer is back to pretty much where I started.
I know there has been "discussions" over the years on various RatRig forums like here and Facebook and similar type discussions happening about Voron as well having some sort of bimetal effect between the aluminum extrusion and the steel linear rail. I don't know what to say besides showing this information I captured that there is obviously something like that happening between the X extrusion and the X rail. VC4 is such a massive upgrade to VC3 and with out quick and detailed Beacon scans the bed, I suspect most people are not going to notice this but its happening. Its causing sub par prints either the first lay or first layers where the squish is going to be changing for the first 30-40 minutes of the printer warming up. I am wondering what others are seeing and is the printer how it is "good enough" or is there anything mechanically I am missing that is exacerbating my problems. Thanks!
Great job documenting this. It has me concerned; I hope someone from RR jumps in here. I don't look forward to long heat soaks before I start printing.
One solution would be titanium, but that's expensive and has issues of its own. I wonder if there's a way to fix the rail firmly in the center and allow slippage toward each end of the rail?
Given that the expansion rates can be predicted, perhaps RR could compensate for this in code?
possibly but the printer does not necessarily know how long its been heating up, etc. I would love it RR could come up with some way of better mounting the rail to allow for expansion but I am not sure if there is a way without inducing noise that would reflect in the quality of the prints
If they knew the temperature of the extrusion, they wouldn't need to know how long it had been heating. Does the toolboard have a temp sensor they could read?
Honestly I am not sure this is such a big issue. I really wonder how much things change by the time a full first layer gets laid down. I stressed about this a lot with the 3.1 500 and went with a carbon beam. Honestly I didn’t see it help at all. The beacon and a good z-offset solved my issues.
And yes I did heat soak for 20min.
my issue is either crapy first layer or some of the first say 10 or so layers look like z banding issues but its just the gantry settling in. Its not like the VC3.0 when I first started as it took for every to mesh the bed with a BLtouch and by the time it was done it was a mess. Between how fast beacon is and the high mesh density its I think good enough. My 2nd prints of the day are perfect.
I agree the Beacon is the game changer. I could never get a good first layer before no matter how long I heat soaked.
I could get "decent" first layer on the VC3.0 if I heat soaked and printed small parts and used Helge's PAM setup
my goal is to print a perfect 500x500 single layer print on the VC4 🙂
I just wish I could find a 500mm smooth pei and use glue. Or even some of these crazy other beds I use on my Bambu. I don’t use the textured bed at all on it
I have a smooth one I bought for my VC3. Never used it. Its just stuck to one of my cabinets in the garage
I love Prusa's satin sheets. Oseq are supposed to be very similar or the same. I asked them about 500 size and they said custom sizes were coming in the future.
I'm still concerned with the length of time needed to heat soak before I can expect a good print.
I'll never be in a 'production' environment where I'm printing continuously. My situation, once the system is tuned, will be to print a part, then shut down until I'm ready to print the next part. Waiting an hour or more before I can print is sub-optimal at best.
I hope RR looks into a software solution based on gantry temperature. All it should take is a cheap sensor and SMOP (small matter of programming).🙂
Best way is to submit a pull request
Not necessarily the best way for me!
This seems to be a common problem exacerbated by the size of the printer. It would be nice if RR found time to code a solution. If not I can always get one of those titanium replacements for the aluminum extrusion, but as I've already spent quite a bit more than I had initially anticipated I'd like to avoid the extra expense.
Who do you mean when you say "it would be nice if RR found time to code a solution"?
Rat-OS is open source software created and maintained by Mikel not RR. Further, Rat-OS is klipper, so you are really asking Mikel and the klipper developers to do something for free for you. Thus it is better to work with them and submit a pull request. I think you will find it isn't as easy as you are making it out to be
The Toro ti tube is the one customisation I'm doing as I build, precisely to reduce or eliminate the need to lengthy heat soaking. I think software compensation would be nice, but not trivial. Rather than trying to estimate actual current bowing related to some temp sensor (where? How is it wired? How reliably do readings map to actual bow?), maybe compensation could be based on live scanning with the beacon sensor.
I tried printing a 350x350 part last night and it was a nightmare. Major microstepping needed to get the print to not be way to close. As soon as Tori has more tubes in stock I am placing my order. This printer is turning out to be worse than my VC3.0 from a gantry movement point of view
Following this for a bit now, and what bothers me the most is the groove/ridge on the left side.
Something is up with that. Since you straight edged the bed and can rule out defects there, the only thing that would show a pattern the length of the y is something on the x.
The sensor is getting closer to the bed around X100, Y0 to X100, Y500.
So if the bed shows no defects and the gantry doesn't have a dip in that spot, could it be something with the magnet or the pei sheet?
I'd start with taking a mic to the pei, and see if there's a ridge around that spot, or rotating the plate 90° and rescanning. If the ridge moves, bingo.
I'm not sure how to best check the magnet, maybe the straight edge again?
I saw the ridge before I even put the pei and magnet on the bed. Its either the X extrusion or the X rail causing it. I have cleaned under it 3 times but have not taken it off as I didnt realize there was a problem until the toolhead was fully assembled. Its a 0.1mm bump and while I plan on fixing it somehow or someway, I dont see how that is causing these problems. I am saying movement of 0.6mm depended on how hot the printer is. Even at 0.8mm I suspect klipper + beacon would deal with it fine but the problem is the height map changes as I am printing so at the beginning its a certain height map and when it finishes a 350x350mm first layer its changed so some of the print is not close enough to the bed while other areas its way way to close. Nothing klipper+beacon can do once its done the heightmap and started the print. It assume everything is going to stay as it was when it captured the heightmap
Interesting. If you have a dial indicator and a magnetic arm, you could tram the bed using the second x carriage. Maybe it'll show something, and it's at least non destructive.
I do have one. I am on the 4th test print back to back this morning and its working even better. I am going to see how it finishes the first layer and if it does I am going to let the printer cool down for a bit and then warm it up to 105-110C from 90 where I have been at to see if maybe the warmer bed temp will stabilize the entire gantry, etc.
Waiting for the ChickenBrand gantry stabilization heater kit bom & github
I just want to figure out a procedure that works to get this thing to print big stuff reliably. I ordered a torque screw driver to see if I have the x rail to tight. I did it not that tight but looking at the numbers @tg73 posted I suspect mine are high. We shall see. I may try to fully take the rail off this afternoon as well and see if I can figure out if its the rail or extrusion and get rid of the bump. Spent to much time fiddling with this. Got other things to do and things to print 🙂
Tram the bed from the second carriage before touching anything. You need a baseline for this. Otherwise you'll blow double the diagnostic time. I'm not flat rate anymore so proper diagnostics isn't a luxury now. 🤣
There is the bed temp at 110c for 30 minutes after being at 90C for 2 hours. Thats over 0.8mm from ambient to hot.
It's expanding and bowing. Are the corners binding and not allowing it to expand? Maybe the sliders for the bed balls in the support arms are not angled correctly?
With the front bias of the warp, that's where I'd look
It’s the linear rail to extrusion difference in expansion. If I loosen the rail up I literately feel it snap back to flat. The problem is if I tighten it back down when it’s cold, it’s bowed the other direction.
Damn, ok.
https://discord.com/channels/582187371529764864/1256094330167230535/1256097811343085658 that was it after it high tightened the rail and let the printer cool down
The challenge is it keeps changing. If the change is slow enough it will get absorbed in the print but it seems like it’s changing fast enough that it’s causing problems. Doing some testing at 110c right now to see if hotter might be more stable
I wonder if a torque pattern would help. Like center outwards?
Tried that as well. When I built I started from one end to the other. When I did the hot tightening I did it the center out and still had the problems. I then tired center in from cold and same thing. That’s what it’s current set at.
With the torque screwdriver going from tight to looser from center out might be worth a try.
It's going to need some slip, apparently
It sure seems like it. Hopefully the torque screwdriver will be here this evening and I’ll do some further testing after this 110c testing
One last thing, if the ends of the extrusion are binding against the side it might bow as the 2020 is the weak point. But snapping back when loosening the rail would maybe rule that out.
I have not loosened the 3 bolts on each side that attach the extrusion to jointer plate. I have only loosened the screws in the rail. I can give that a try in a few. Easy test to see if it makes a difference.
Might want to take a look at https://www.printables.com/model/230716-ratrig-v-core-3-extrusionless-x-gantry-mod
Dubstep discovered this exact same issue years ago
Printables.com
RatRig v-core 3 extrusionless x-gantry mod by elcojacobs | Download...
V-core 3 mod to remove the aluminium extrusion from the x-gantry and run on just the rail or a top and bottom rail | Download free 3D printable STL models
I have all of the parts for that on my VC3. I planned to upgrade it to that mod
You probably could ask elco to modify it for vcore4
I think going to the titanium tub would be a better option on vc4. With most of the movement parts being metal I would not want to introduce plastic being a core part holding the printer together. Also the stock rails on vc4 probably not long enough for a mod like this without a long printed part holding it. So that means two new rails and at that point the titanium tube is cheaper for a 500mm machine
I have had the printer up to 110c bed for over two hours at this point. The printer had been enabled the entire time so the z-tilt should not have changed but over the last 30 minutes it changed another 0.06mm but the bed range improved from 1.2mm to 1.1mm (ambient is 0.233). Not sure if that means it’s stabilized or not. Going to let sit for another 30 mins.
I'm sorry if I mislead you with my SMOP comment. It was an inside joke when I was a developer back in the 80's and 90's. I didn't intent to imply that it was in any way easy. I also didn't understand that Mikel was doing this for free; I assumed he was a RR employee.
Rather than trying to estimate actual current bowing related to some temp sensor (where? How is it wired? How reliably do readings map to actual bow?), maybe compensation could be based on live scanning with the beacon sensor.It seems to me the temperature of the extrusion, measured in the center, would be enough to calculate the bowing based on the expansion factor of steel vs aluminum. If the Beacon is able to map while a print is in progress that would be a better way to approach the problem as it would work with standard hardware. I didn't expect RR to ship a product that is basically unusable in its stock configuration (which if @chicken's experience is typical, it is). Is torque specified in the build docs?
no. I can tell you what torques I've been using so far (I've just done the y axis rails).
I can tell you what torques I've been using so far (I've just done the y axis rails).Yes, please and thank you!
This really isn't adding much to the actual troubleshooting this thread is for.
I don’t think RR had torque recommendations. I don’t have a torque screwdriver but one should be here in a few hours.
This chart was just posted in #v-core-4
From hiwin
MGN12 on aluminium 0.98nM
HG15 on aluminium 2.06nM
(hiwin spec)
Other torques based on typical specs for the bolt size and my own feel range from 0.5nM (leadscrew nut attach to coupler), 1.1nM (pom inserts, z arms to carriages, limit asemblies, leadscrew constraint brackets), 1.5nM (leadscrew to stepper couplers). Everything except rails, frame quick connects and the base plate has loctite 222.
I think it's very common to overtorque the rails, which stops them from expanding and contributes to bimetal effect (so says mikel I think).
Purple? Not Blue?
M6 and below, 222 is recommended AFAIK. Notably with smaller threads. Let me see if I can find a ref.
How to choose the right LOCTITE® threadlocker
With so many options, sometimes it can be hard to choose which threadlocker is right for you. Luckily here at LOCTITE®, we have provided you with some answers to help you choose what threadlocker is the best fit for you.
I found it on the back of the package LOL
2nd Amazon order today forgot the thread locker ordered 222 and 242
what's 242 and what will you use it for?
Blue medium strength M6-M20
A lot of people assume he is. But he isn't, and he and others that have worked on Rat-OS did it for free
I built my vc4 z axis and y rails today. I had wrongly torqued to the middle column of the hiwin specs, but now rectified. I have a digital torque driver which is great for this scale of torquing. I'd love a certified calibrated one, but can't justify the cost. But for sure better than the bare-handed gorilla.
I just tried to sponsor him on GitHub, but they no longer take PayPal for sponsorships. I've emailed them to find out how I can pay for the sponsorship. Given what he has created for us, IMO we should all be chipping in.
Ok I let the printer heat soak for 6 hours and was still measuring changes on z-tilt each time. About 0.04 to 0.06 each time. The bed mesh stabilized at 1mm in change from ambient to heated. I found my dial indicator and set it up under the middle of the X gantry and reset it to 0mm. The middle of the gantry dropped 0.79mm from just loosening the x rail screws. The steppers are all still enabled so it was not the bed moving. This test was done after about 1 minute of the front door being off. I heard the gantry move when I was getting near the last of the screws and saw the dial indicator moving in the negative range, it moved most of this difference when I loosened the last screw on the left side (right screw was still tight). To me this shows the bimetal effect. I am just trying to figure out why my machine is nearly unusable for any large prints and others seem to be printing ok.
Pictures were rotated so you could read the indicator
After 40 minutes of cooling the gantry has settled down another 0.18mm. The rail is still loose so this is just the gantry itself of the bed as the dial indicator is sitting on it. I may try retesting this again to see if loosening the x extrusion bolts to the y mounts is binding it up at all as well.
Do you have a tool for accurate torquing of the rail screws? The correct torque is really quite low https://discord.com/channels/582187371529764864/1256094330167230535/1257103727660105819, and I think most people would naturally end up with 1.5-2.5nM without a torque tool. You need correct torque on all rails.
Discord
Discord - Group Chat That’s All Fun & Games
Discord is great for playing games and chilling with friends, or even building a worldwide community. Customize your own space to talk, play, and hang out.
And it's vital to tighten the screws in order from one end to the other, or outwards from the centre.
Must be nice. Civilization, I mean. Next day is best I can get and only if it's in the Vegas warehouse
I've been using my beam torque wrench and some printed clickers that spec out a ~1.5 and ~1 Nm.
Yah San Francisco Bay Area most things are overnight with more and more same day or before 7am delivery .
So nearly 2 hours of cooling the gantry settled down 1.04mm. 0.79mm was the rail holding it up and 0.25 was the extrusion itself cooling down. Most of that was in the first 30-40 minutes. That 1mm is pretty much the exact changes that beacon picked up in the height map as the printer heated.
2020 might be a bit skinny for a 500
Taking the entire x rail and extrusion out to inspect it. See if I can find that bump.
Will that fit to tighten the Y rails? I can't tell the size from the pic
Sorta 🤷♂️
Clearance is clearance
the rail is wavy and the x extrusion is bowed
thinking of rotating the extrusion 90 degrees and seeing how things tighten up
That might help. My extrusion passed the straight edge test. Didn't check the rail.
0.98 n m is tighter than I thought
I have one coming on Tuesday, not that exact one, it's supposed to have better screen. We'll be set to assemble anything after this.
Makes me wish I still had access to a large surface plate table. Wondering if you had a warped rail to start with . A bent extrusion would have shown up during installation.
I'm still using VC3.1 400 with beacon but i used to have similar issue until i made some changes which greatly improved my first layer. I will put them here if you would like to try them out. All of the changes are made in the software printer.cfg not to the printer
on my EV3, i noticed that the tilt probing is not centered on the rear so i made some changes to that too
I increased the number of probing point to 21x21 and reduce the compensation on the mesh
I also reduced the tolerance of the Z adjustment to 0.005. It takes the printer 4-5 tries to get it but eventually it does better than that
[bed_mesh] horizontal_move_z: 5 mesh_min: 20,20 mesh_max:390,390 probe_count: 21,21 fade_start: 1.0 fade_end: 10.0 mesh_pps: 2,2 algorithm: bicubic bicubic_tension: 0.1 [beacon] serial: /dev/beacon x_offset: 0 y_offset: 18.2 mesh_main_direction: x mesh_runs: 1 speed: 15. lift_speed: 80. contact_max_hotend_temperature: 275 [z_tilt] z_positions: 0,0 200,400 400,0 points: 60,60 200,370 360,60 retry_tolerance: 0.005
Thanks for the info. If you are already using beacon you should take a look at what the RatOS team has been doing for gantry twist compensation in RatOS 2.1. Its mostly being talked about in the VC4 channel but I suspect the beacon channel as well. I went back and tore apart my gantry to deal with that bump and then ran into so other issues with the CoreXY system. I should be back at the point of playing with this later today
Thanks .. I will look into the gantry twist
GitHub
RatOS/site/docs/configuration/beacon_contact.md at documentation_v2...
The preconfigured Raspberry Pi image that makes it easy to run Klipper + Moonraker + Mainsail on your printer. - HelgeKeck/RatOS
bed
?
Hey @chicken, would you mind if I pm'd you about this issue? Looking into v4, but having seen such bowing with v3 I'm wary and trying to see how it is for people affected
Yah go for it. Stepping out to go help fix something at my parents house but I should be back in 90 mins or so