A few people asked where to buy a Z1 preloaded Fysetc MGN7H rail for Voron 0.2. I told Fysetc and they added it to their Aliexpress store - s.click.aliexpress.com/e/_DFnWBDL The link is an affiliate. If you buy please let me know in the comments if the rail is as good as mine was.
I never did a measurement about it, but I had sensed this problem intuitively and added a counterweight to the loom on my End5+ machine. Your video has inspired me to characterize the difference it makes.
Damn that's dissapointing that the Fysetc kit is plagued with so many problems. Seems like you're better off with another kit maker for the Voron V0.2.
Yeah, it is a small 3D printer and their decision to include such a thick wire loom with loose rails is so bad. They should have rather included the CNC printhead bracket with regular extrusion and the CAN BUS.
@@PrintingPerspective sadly pretty much all kits with a picobilical board have these thick looms. You have 14 (?) wires and the not the soft one with the silicone insulation, personally im not surprised. Formbot was so grateful to have all rails as z0 light preload and a x rail in a separate package with z1 high preload. My rails were flawless, but i switched to a mgn9 for x as i retrofitted a pandoras box gantry for out of build space travel A bit off topic, but voron should make the top hats of their printers bigger, doesn't matter if that ruins backwards compatibility with older frames or not. If you want gentle bends on a v0 bowden tube, it will most likely rub. On a trident 300 amd 350, it will also rub at some point, maybe not on the 250. Ratrig for example has differently tall top hats depending on the printer size on the v core enclosure, thats how it show be done
I wouldn't want the vorons enclosures downgraded to the level of the vcore's though. Super complex stl, very expensive to have made, absolutely not hermetically sealed... and clearly the printer was not designed to be enclosed like the V2. I don't currently have a cable-chain-less voron so I don't know avout that, but I would have assumed that if you went without cable chains you would go CANbus and it would be fine.
@@ludovicurbain4048 i would just like that the tophat also scales with size. Its not only the umbilical wiring if you have one, but its mostly the ptfe tube, especially if you want to print carbon fiber filaments. The rubbing on the top panel scratches it over time, which looks quite ugly and i personally had some homing failures due to the ptfe tube getting between endstop and x axis during y homing. Could have been prevented with more head room. You would have the same issues with a v2 at higher z heights, but you could simply limit the max z height for that, although on a 250mm one thats not that cool as it can only print around 200mm tall before smashing the ptfe tube into the top panel, which is also am after thought im my opinion, the v2 frame only allows for maximum advertised z height without a direct drive
@@kilianlindlbauer8277 indeed it makes sense. Vorons have always been designed with printer density in mind. Maybe other v0 users have figured out a way around that issue.
I built a V2 120x120 because of things like this, I could see the head move on those 7mm rails, the 12mm of the V2 are perfect. Even though the bet is small, having QGL for perfect flatness and TAP for auto z offset is a dream
thanks for this. I bought a super cheap V0.1 kit Fysetc but immediately ditched the supply control board for a BTT M5P. I have also noticed that the linear rails are sloppy/loose - I will be replacing them with some better quality ones. I will probably source the X carriage from pcbway, so once again thank you for posting the link. I am looking forward to your future videos.
I made mine from compression molded chopped strand carbon fiber. You can 3D print the molds and it is cheaper if you’re on a budget. It is messy though, not ideal if you don’t have a garage or something.
I've needed to replace the umbilical cable because of loose connections. I made my own with more flexible silicone cable and it has been working decently since. I didn't go as in depth as you did but my kit FYSETC kit definitely shares the issues (it's an older one where the top hat was even shorter, 70mm...)
One thing i know is that 3d printer never using silicon cable. But all of drone racing configuration using silicon cable on all wiring. Maybe there are custom cable that using silicon cable
I just bought a Formbot V0.2 ket and they included a Z1 preload rail for the X axis. I also went with Formbot, because they included a better/simpler controller board based on Raspberry "Pico" (aks rp2040) Then I replaced the Kigigami bed mount with a Fysetc CNC version. THAT is really worth it. You can bend a Kirigami with you fingers. It is not accurate and the Z motion was not smooth. Now with CNC frame Z motion is perfect.
Funny this popped up when I'm chasing down a .1mm over 150mm inaccuracy on my delta printer. Just rebuilt it with some home made mag rods, so they aren't quite perfect as a result I'm getting so a minor discrepancy between my x z and y axis measurements (3 axis split apart by 120 degrees, so you need to calibrate in 3 axis instead of just 2 on cartesian). While it can be a bit more of a pain to tune a delta, you can actually get them to print even more accurately than most cartesian printers you take the time.
Great video. I appreciate your methodical approach to this problem. I noticed some play in the x linear rail on mine too but didn't realise the affect would be do much! At 7:01, are you sure that the shorter bowden tube length didn't have any effect on your y accuracy? It looks like it will apply some bending torque to the print head at higher y values due to the loop being a little too short.The shot of the machine fitted with the CAN toolboard has a much longer loop length.
Thanks, I doubt it had any significant effect. I was using a PTFE tube with 2.5mm inner diameter and it is very flexible compared to those stiff wires. I wish it were possible to transfer the feel through a video because it is so stupid how stiff that wire loom really was. :)
I want to thank you ffor your insight on this little printer, it's sad because they opened a local warehouse on my country so I can get it for 300 bucks with some coupons and cc promotions. Since you have contact with fysetc, do you know if this kit is revised/improved? Otherwise it's not worth it over a formbot kit I think
Those PTFE wires were very stiff, and there's not that much room for a longer cable when enclosed. More flexible cables probably would be a better solution.
I asked "KGT Factory Store" on AliExpress if I can get Z1 preload on the rail. They said order the regular rail and make sure to inform them what you need.
@@PrintingPerspective ok thanks! I have some kgt rails and I like them but didn’t knew I could have asked for different preloads. Z1 preload would be a medium preload rail right?
Here are the preload levels in general - www.automation-overstock.com/hiwintech/preloads.gif I like their MGN12 rails, but the MGN7 with Z1 preload seemed too much for the Voron 0.2 or it was just a bad quality one, who knows.
can you please make a tutorial ablut that canbusfix/CM68 debian 11 update thing? fysetc's doc doesn't work or I am just way to stupid to do it myself :((
Bro... that board is such a headache that it would be so much easier just to replace it. There are so many things missing in those image files... If you are stuck not knowing how to update the image, there will be so many more difficult things to configure... And most important they don't know how to fix the overloading MCU issue with slow CAN BUS rates...
@@PrintingPerspective well that's massive headache.. I was trying to get crowsnest on the same kit but it doesn't support debian 10 buster so I tried updating it but got stuck at rkdevtools
I know it would be a ton of work, but I wish there were numbers for each mod from the start. Just in case one mod accomplished almost all of the improvements.
Well, that's the honest truth. Calling the kit "Pro" and it having such bad printing accuracy is just unacceptable. They could have easily included the right upgrade parts and it would have been a killer deal.
The cable should have ground, power, and a couple of wires for serial data. CANBUS is one option. The cable should be rated for continuous flex, as used for robots or cable chains. The extruder board encodes and decodes the data, drives the stepper motor, manages all of the sensors, etc. An optical filament runout sensor should be on top of the extruder. In a perfect world, the hardware and data format would be an industry standard, making it much easier to develop third party extruder assemblies that could be optimized for flexible material, carbon fiber filament, etc.
How are you able to achieve such high accuracy on your bambu lab X1C printer? I purchased 5 of them for my farm and when I tested all of them, they were no where close to your measurements, and ranged somewhere in the 0.3-0.4 accuracy range. I contacted them about fixes but they said it’s within tolerance ranges and I have nothing to worry about.
Not sure if you are using the adxl from the ebb board, but it's bad position, plus the flimsy mount, will always gives you strange shaper results. Also, Klipper docs states that 250k baud rate for a can interface is not enough for shaper, even 500k can saturate the whole thing and mess with the measurements.
@@PrintingPerspective Reducing the speed at which the entire input shaping accelerometer test runs can help get at least accurate (if not still flawed) results.
when you re-ran input shaper, did you use the adxl on the ebb36 board? I have found them to be very inaccurate because they just aren't mounted securely enough to the toolhead.
@@PrintingPerspective I had the same thought about the ebb board. I saw someone testing CAN board vs hotend shroud vs nozzle mount, and CAN performed much worse compared to nozzle & shroud mounting (both nozzle and shroud gave pretty similar readings if I remember correctly). I wish I could tell you where I've heard this, but I can't remember.
Man i love what you do, all your videos are both super easy to follow, yet you give the best description of anything 3d print I've seen of the platform. I just subscribed to you on Patreon and are excited to read all you post present past and future. Keep up the good work
Have to agree, the cable is too stiff and the rail has visible play on my printer as well. Right now I work an a V2.4 before I spend time on those issues. I also had stability issues with the display and already swapped for an USB one.
I am measuring the length of the prints printed on the X vs Y axis with a jig for consistent readings. I linked the video description to the slightly updated test on the Printables.
What do you think about using one multichannal cable? I've been thinking about it for a while now but haven't tried it yet. Some VGA-type cables have enough channels, but I'm not sure if they can handle the power for the hotend.
For a CAN BUS? If so I just used twisted and shielded cable for the data + 2 wires for the power, and then both sleeved in. But VGA cables I think are way too thin, they are data cables so were never designed to carry power.
@@e.f.k.3547I wonder how well a usb c cable would do, with newer phones that have charging capability of 120w wonder if they would work here. Although that's at 20v and not 12v so amps would go up. I've seen people use mini dvi cable or vga can't quite remember and that apparently worked well.
I had similar issues with play in the x carriage of my ldo v0.2 (still better than the fysetc but not perfect). Ended up going with an mgn9 honeybadger rail, honeybadger alu x carriage, and my custom microsaul toolhead. No play whatsoever, and my is results were insane
You touched on all important aspects but you missed why the light weight beam is worse despite lighter weight. It's weak in torsion. You need a CF tube that is drilled on both sides so that you can screw the rail all the way through to the other side of the beam with long screws and have a printed insert inside that the screw goes through. That way you will have less weight AND more stiffness than the stock aluminium part. It doesn't matter how stiff the beam is if the rail it self can wobble on it because of a weak connection to it. The aluminium one is just too thin so won't be any good. If you combine that with a rail with the highest preload (z2) then that will be great. I recommend the hollow mgn9 from fysetc.
Why don't understand so many people the basic rule of market? Be cheaper as your opponent. That's the reason why Creality sells so well and why there is no pre-loaded rail.
Wires come in many different variations past solid and stranded. I see it in the automotive world all the time. Some wires are far more flexible than others.
I wouldn't worry about that on a bed slinger, unless it is ridiculously thick. Plus you can adjust steps per mm as both axis have independent motors unlike on corexy.
How did you find squaring the frame? I've had mine for about 2 years now and soon going to do a lot of maintenance work on it and i need to improve the frame squareness. Any tips? My machines the 400, thanks. And yeah an amazing printer!!!
Smooth idler pulleys are the Debble!, found moving to toothed idler pulleys a long time ago before resonance compensation was a thing helped a lot. Effectively the teeth act as springs on smooth pulleys About bearings, linier rod bearings anyway. I have a Sermoon D1, all the bearings are clocked at 12. So all motion forces act on 3 and 9 while 12 and 6 hold the carriage in Z direction. Doing that removed all the slop in my printer. And of course the machine is squared and the bed is trammed, having a quality square really helps. The frame doesn't have to be perfect as long as the rails are able to be, wasted a lot of time on that. Note, if you do get it perfectly square you WILL find out if there's slop/tolerance gaps in your machine. Found this out on my old wooden Replicator, you could hear the carriage and bearings rattle after, even was visible on the prints. So, you might want some binding to press the bearings against something.
You may want to rerun your IS results while adjusting your accel_per_hz. The messy x rail results have a super low power being a scale of 0-1.75 1e3. You are just measuring noise.
@@benclimo461check the klipper documentation. You’ll use the accel_per_hz setting under input shaper. Make sure it’s with a nozzle ADXL or something. CAN board ADXLs aren’t very good for actual results
Yeah with DIY printers you really want to have a good kit already, because rebuilding the printer later with replacement parts and figuring out issues is not that fun.
I have fysetc v0.1 kit and I'm happy over 2 years. I had to change only X axis rail because of play. Otherwise all parts were good. I run CANbus ebb36 on my another printer vzbot235 where I use 500k baud rate with SKR Pico because with 250k ADXL results are not correct. So far no problem with overloading MCU. Just yesterday I updated klipper and had to update also MCU and EBB and it went smooth.
This isn’t an uncommon issue in precision automation. First thing I would want the cable connected at a place as far from a point of leverage as possible. So loop it down and connect it parallel with that linear support. Now it’s incapable of rocking the tool head no matter how much tension it comes under. I’ve never been in a position to redesign the cable however I can add length to it so it never goes into tension. Building a structure over the machine (or using the ceiling) to support excess cable. It basically “dangles”. We’re hillbillies but get the job done. 😂
4:08 ---i bet you didnt clean the rail , bearing block and lubricate it with proper silicone lubricant for this type of application (yes you need to do this with brand new rail ...purely because it isnt premium quality rail already lubed with 500$ per kg high speed machining grease , most of the time they need cleaning so clean it😅 Edit--- This is also created in manufacturing when plastic parts on the end of bearing block arent aligned well , balls inside get stuck on the edge of plastic when ball comes out of a steel part (block) . Simple solution to it is to just tap the plastic up/down until it move smoothly / loose screws a little , align and it should work perfec If not...you can unscrew plastic parts and check if the edge of plastic isn't already bend inside etc ...i use something sharp to cut edge the plastic hole at an angle so the balls can move smoothly in and out....works every time PS--- idea for next episode how heavy can the X axis be to still go 200mms with at least 5k acceleration i bet you cant beat my old behemoth from like 9-10y ago .... around 4 kg X axis 😆😆 (yes , it was big...and heavy😆)
Nowadays you don't really need to do a lot of cleaning with rails from my experience. I just clean the existing oil from the ball bearings with IPA and use Mobill EP2 grease. Works perfectly on Fysetc rail, and all other rails.
@@PrintingPerspective AH i forgot This is also created in manufacturing when plastic parts on the end of bearing block arent aligned well , balls inside get stuck on the edge of plastic when ball comes out of a steel part (block) . Simple solution to it is to just tap the plastic up/down until it move smoothly / loose screws a little , align and it should work perfect If not...you can unscrew plastic parts and check if the edge of plastic isn't already bend inside etc ...i use something sharp to cut edge the plastic hole at an angle so the balls can move smoothly in and out....works every time
On the rail I have one suggestion. I had a lose fitting rail and tried to squeeze it with a bench vise. Worked perfectly after. Not saying that's the way to go about. But there is nothing to lose anyway.
I wonder if this is why my fysetc 0.1 has always been useless over Y100. I have never been able to get it to stick properly to the back of the bed and the extreme position flexing the head would explain it.
yep - having issues with my Fysetc 0.2 Pro... ran well at first (30 hours) but now having shutdown after 20 min with undervoltage errors. Now support is telling me to re-flash the firmware...
A Flashforge adventurer 5m Pro test would be very interesting as it shined in all the tests I saw regarding print quality. But as you mentioned in your videos it would be nice to know the real performance and not the performance on prints where everything looks good.
My 300mm V2.4 is the most precise printer that I have today. But it has drag chains with multiple PTFE wires (not CAN BUS). The bigger the printer the less it should affect in theory because all the parts will be bigger and more rigid which minimizes the negative effect of the thick wire loom.
Yeah, they are lacking but my god, compared to Fysetc docs they seem good lol. I have this in the config for the Manta E3EZ and BTT ADXL345 V1.0: [adxl345] cs_pin: PC15 spi_software_miso_pin: PC11 spi_software_mosi_pin: PC12 spi_software_sclk_pin: PC10 [resonance_tester] accel_chip: adxl345 probe_points: 150,150, 25 # middle of bed as an example
Well without having seen his model, it's a easy effect to make in a test print. You want peak acceleration, so a sharp turning point. A 4 leg star would be great. Print in vase mode or single wall, custom speed settings (could increase for every 20-50 layer). Please note that this will not be accurate (true measure), so you're only looking for the difference between X and Y
Thanks. I made a jig to measure length consistently with digidal calipers. There's a link in the video description to the updated version of tests, which includes that jig.
Replacing the stiff PTFE / Plastic Cable Loom in a Sleeving with one made of flexible Silicone Cables without Sleeving would probably have fixed the issue.
I am sure that would help, but considering how much rewiring you would need to do, CAN BUS would be so much easier if you don't run the Catalyst board.
Silicone isn't too great in places where you need to have some wire on wire movement. proper PTFE insulation on wires and slightly loose wire sleeve ( even PET one ) is great. On negative side wires with proper PTFE insulation are not cheap , on plus side they are kinda "soft". But i don't like it either and i will move it to CAN with proper cable for that , which is really nice and again - expensive.
A few people asked where to buy a Z1 preloaded Fysetc MGN7H rail for Voron 0.2. I told Fysetc and they added it to their Aliexpress store - s.click.aliexpress.com/e/_DFnWBDL
The link is an affiliate. If you buy please let me know in the comments if the rail is as good as mine was.
I never did a measurement about it, but I had sensed this problem intuitively and added a counterweight to the loom on my End5+ machine. Your video has inspired me to characterize the difference it makes.
Damn that's dissapointing that the Fysetc kit is plagued with so many problems. Seems like you're better off with another kit maker for the Voron V0.2.
Yeah, it is a small 3D printer and their decision to include such a thick wire loom with loose rails is so bad. They should have rather included the CNC printhead bracket with regular extrusion and the CAN BUS.
@@PrintingPerspective sadly pretty much all kits with a picobilical board have these thick looms. You have 14 (?) wires and the not the soft one with the silicone insulation, personally im not surprised. Formbot was so grateful to have all rails as z0 light preload and a x rail in a separate package with z1 high preload. My rails were flawless, but i switched to a mgn9 for x as i retrofitted a pandoras box gantry for out of build space travel
A bit off topic, but voron should make the top hats of their printers bigger, doesn't matter if that ruins backwards compatibility with older frames or not. If you want gentle bends on a v0 bowden tube, it will most likely rub. On a trident 300 amd 350, it will also rub at some point, maybe not on the 250. Ratrig for example has differently tall top hats depending on the printer size on the v core enclosure, thats how it show be done
I wouldn't want the vorons enclosures downgraded to the level of the vcore's though.
Super complex stl, very expensive to have made, absolutely not hermetically sealed... and clearly the printer was not designed to be enclosed like the V2.
I don't currently have a cable-chain-less voron so I don't know avout that, but I would have assumed that if you went without cable chains you would go CANbus and it would be fine.
@@ludovicurbain4048 i would just like that the tophat also scales with size. Its not only the umbilical wiring if you have one, but its mostly the ptfe tube, especially if you want to print carbon fiber filaments. The rubbing on the top panel scratches it over time, which looks quite ugly and i personally had some homing failures due to the ptfe tube getting between endstop and x axis during y homing. Could have been prevented with more head room. You would have the same issues with a v2 at higher z heights, but you could simply limit the max z height for that, although on a 250mm one thats not that cool as it can only print around 200mm tall before smashing the ptfe tube into the top panel, which is also am after thought im my opinion, the v2 frame only allows for maximum advertised z height without a direct drive
@@kilianlindlbauer8277 indeed it makes sense. Vorons have always been designed with printer density in mind. Maybe other v0 users have figured out a way around that issue.
I built a V2 120x120 because of things like this, I could see the head move on those 7mm rails, the 12mm of the V2 are perfect. Even though the bet is small, having QGL for perfect flatness and TAP for auto z offset is a dream
thanks for this. I bought a super cheap V0.1 kit Fysetc but immediately ditched the supply control board for a BTT M5P.
I have also noticed that the linear rails are sloppy/loose - I will be replacing them with some better quality ones.
I will probably source the X carriage from pcbway, so once again thank you for posting the link.
I am looking forward to your future videos.
I made mine from compression molded chopped strand carbon fiber. You can 3D print the molds and it is cheaper if you’re on a budget. It is messy though, not ideal if you don’t have a garage or something.
I've needed to replace the umbilical cable because of loose connections. I made my own with more flexible silicone cable and it has been working decently since. I didn't go as in depth as you did but my kit FYSETC kit definitely shares the issues (it's an older one where the top hat was even shorter, 70mm...)
One thing i know is that 3d printer never using silicon cable. But all of drone racing configuration using silicon cable on all wiring. Maybe there are custom cable that using silicon cable
I just bought a Formbot V0.2 ket and they included a Z1 preload rail for the X axis. I also went with Formbot, because they included a better/simpler controller board based on Raspberry "Pico" (aks rp2040)
Then I replaced the Kigigami bed mount with a Fysetc CNC version. THAT is really worth it. You can bend a Kirigami with you fingers. It is not accurate and the Z motion was not smooth. Now with CNC frame Z motion is perfect.
Funny this popped up when I'm chasing down a .1mm over 150mm inaccuracy on my delta printer. Just rebuilt it with some home made mag rods, so they aren't quite perfect as a result I'm getting so a minor discrepancy between my x z and y axis measurements (3 axis split apart by 120 degrees, so you need to calibrate in 3 axis instead of just 2 on cartesian). While it can be a bit more of a pain to tune a delta, you can actually get them to print even more accurately than most cartesian printers you take the time.
I think using a flexible pcb or flat power ribbon cable would be the way to go.
Great video. I appreciate your methodical approach to this problem. I noticed some play in the x linear rail on mine too but didn't realise the affect would be do much! At 7:01, are you sure that the shorter bowden tube length didn't have any effect on your y accuracy? It looks like it will apply some bending torque to the print head at higher y values due to the loop being a little too short.The shot of the machine fitted with the CAN toolboard has a much longer loop length.
Thanks, I doubt it had any significant effect. I was using a PTFE tube with 2.5mm inner diameter and it is very flexible compared to those stiff wires. I wish it were possible to transfer the feel through a video because it is so stupid how stiff that wire loom really was. :)
I want to thank you ffor your insight on this little printer, it's sad because they opened a local warehouse on my country so I can get it for 300 bucks with some coupons and cc promotions. Since you have contact with fysetc, do you know if this kit is revised/improved? Otherwise it's not worth it over a formbot kit I think
Dont need to go canbus, just need propper cables, with the propper length, that original cable seems a bit short.
Looks like more troubles not because wire harness thickness, but because it a too short and pulls printhead
Those PTFE wires were very stiff, and there's not that much room for a longer cable when enclosed. More flexible cables probably would be a better solution.
Why not dwsign it to have the loom come out at a 90?
Ugh i have this kit in the box sititng for two months waiting to be built now 😢
Very informative video!
One question btw, where did you bought the KGT linear rail with Z1 preload?
I asked "KGT Factory Store" on AliExpress if I can get Z1 preload on the rail. They said order the regular rail and make sure to inform them what you need.
@@PrintingPerspective ok thanks! I have some kgt rails and I like them but didn’t knew I could have asked for different preloads. Z1 preload would be a medium preload rail right?
Here are the preload levels in general - www.automation-overstock.com/hiwintech/preloads.gif
I like their MGN12 rails, but the MGN7 with Z1 preload seemed too much for the Voron 0.2 or it was just a bad quality one, who knows.
very nice video, thank you!
can you please make a tutorial ablut that canbusfix/CM68 debian 11 update thing? fysetc's doc doesn't work or I am just way to stupid to do it myself :((
Bro... that board is such a headache that it would be so much easier just to replace it. There are so many things missing in those image files... If you are stuck not knowing how to update the image, there will be so many more difficult things to configure... And most important they don't know how to fix the overloading MCU issue with slow CAN BUS rates...
@@PrintingPerspective well that's massive headache.. I was trying to get crowsnest on the same kit but it doesn't support debian 10 buster so I tried updating it but got stuck at rkdevtools
I order my first kit voron , the v0.2 r1 from formbot. I hope that it's a better kit than this :)
the kit is great! well except for the thick cable...mine died after 200h of printing
I am afraid this is Formbot kit. Or Fysetc. They use this setup.
Make your own loom.
I know it would be a ton of work, but I wish there were numbers for each mod from the start. Just in case one mod accomplished almost all of the improvements.
man, you tore Fysetc a new one at the end....
Well, that's the honest truth. Calling the kit "Pro" and it having such bad printing accuracy is just unacceptable. They could have easily included the right upgrade parts and it would have been a killer deal.
Well they just did that on the rev V1.1 check the new BOM of course I am in the same Boat I bought this in January
notif gang?
Pretty wild that compute, clocks, and datarates are still something of a limitation in 2024. I mean, BTT sells board with 550Mhz main MCU's ...
The cable should have ground, power, and a couple of wires for serial data. CANBUS is one option. The cable should be rated for continuous flex, as used for robots or cable chains. The extruder board encodes and decodes the data, drives the stepper motor, manages all of the sensors, etc. An optical filament runout sensor should be on top of the extruder. In a perfect world, the hardware and data format would be an industry standard, making it much easier to develop third party extruder assemblies that could be optimized for flexible material, carbon fiber filament, etc.
How are you able to achieve such high accuracy on your bambu lab X1C printer? I purchased 5 of them for my farm and when I tested all of them, they were no where close to your measurements, and ranged somewhere in the 0.3-0.4 accuracy range.
I contacted them about fixes but they said it’s within tolerance ranges and I have nothing to worry about.
Not sure if you are using the adxl from the ebb board, but it's bad position, plus the flimsy mount, will always gives you strange shaper results. Also, Klipper docs states that 250k baud rate for a can interface is not enough for shaper, even 500k can saturate the whole thing and mess with the measurements.
Yeah, most likely.
@@PrintingPerspective Reducing the speed at which the entire input shaping accelerometer test runs can help get at least accurate (if not still flawed) results.
when you re-ran input shaper, did you use the adxl on the ebb36 board? I have found them to be very inaccurate because they just aren't mounted securely enough to the toolhead.
Yes, but I made a custom bracket that holds the PCB in place. It felt pretty secure but I guess I will never know without more testing.
@@PrintingPerspective I had the same thought about the ebb board. I saw someone testing CAN board vs hotend shroud vs nozzle mount, and CAN performed much worse compared to nozzle & shroud mounting (both nozzle and shroud gave pretty similar readings if I remember correctly). I wish I could tell you where I've heard this, but I can't remember.
@@habag1112 I believe it was ModBot. th-cam.com/video/ufmUmeDnr0Y/w-d-xo.html
Man i love what you do, all your videos are both super easy to follow, yet you give the best description of anything 3d print I've seen of the platform.
I just subscribed to you on Patreon and are excited to read all you post present past and future.
Keep up the good work
nice video!!!!
Thanks for this video. My V0 is in serious need of a re-build/upgrade so I'll look in incorporate some of your experience.
Have to agree, the cable is too stiff and the rail has visible play on my printer as well.
Right now I work an a V2.4 before I spend time on those issues.
I also had stability issues with the display and already swapped for an USB one.
Good to know than the onboard CAN won’t perform.
I would have tried to use that - guess I’ll use an USB CAN from the get go.
I don’t understand the test methodology of how your measuring that the machine is becoming more accurate
I am measuring the length of the prints printed on the X vs Y axis with a jig for consistent readings. I linked the video description to the slightly updated test on the Printables.
What do you think about using one multichannal cable? I've been thinking about it for a while now but haven't tried it yet. Some VGA-type cables have enough channels, but I'm not sure if they can handle the power for the hotend.
For a CAN BUS? If so I just used twisted and shielded cable for the data + 2 wires for the power, and then both sleeved in. But VGA cables I think are way too thin, they are data cables so were never designed to carry power.
No, in this case I meant without can-bus. I also know that a VGA cable wouldn't work but maybe there is another type of cable.
@@e.f.k.3547I wonder how well a usb c cable would do, with newer phones that have charging capability of 120w wonder if they would work here. Although that's at 20v and not 12v so amps would go up. I've seen people use mini dvi cable or vga can't quite remember and that apparently worked well.
Formbot kit have x-axis preloaded on default, it works fine.
I had similar issues with play in the x carriage of my ldo v0.2 (still better than the fysetc but not perfect). Ended up going with an mgn9 honeybadger rail, honeybadger alu x carriage, and my custom microsaul toolhead. No play whatsoever, and my is results were insane
Nice, they should be insane, those are some nice upgrades.
You touched on all important aspects but you missed why the light weight beam is worse despite lighter weight. It's weak in torsion. You need a CF tube that is drilled on both sides so that you can screw the rail all the way through to the other side of the beam with long screws and have a printed insert inside that the screw goes through. That way you will have less weight AND more stiffness than the stock aluminium part. It doesn't matter how stiff the beam is if the rail it self can wobble on it because of a weak connection to it. The aluminium one is just too thin so won't be any good. If you combine that with a rail with the highest preload (z2) then that will be great. I recommend the hollow mgn9 from fysetc.
lol the wobbly printer
Why don't understand so many people the basic rule of market? Be cheaper as your opponent. That's the reason why Creality sells so well and why there is no pre-loaded rail.
Wires come in many different variations past solid and stranded. I see it in the automotive world all the time. Some wires are far more flexible than others.
if i ever get metal parts for a voron its gonna be from pcbways metal 3d printing service cuz it has to be 3d printed
Love It - fast and full of pin point conclusions!
Makes me wonder whether some of the fancy cable chains for the ender 3 cause similar issues? Gonna be a shame since I love how they look.
I wouldn't worry about that on a bed slinger, unless it is ridiculously thick. Plus you can adjust steps per mm as both axis have independent motors unlike on corexy.
always use class 6 cables and if you can afford use silicone
I cannot recommend a vcore-3 ratrig enough I am in love with it and after the initial setup have been extremely happy with it!
Yeah, I need to get my hands on it.
How did you find squaring the frame? I've had mine for about 2 years now and soon going to do a lot of maintenance work on it and i need to improve the frame squareness. Any tips? My machines the 400, thanks. And yeah an amazing printer!!!
Smooth idler pulleys are the Debble!, found moving to toothed idler pulleys a long time ago before resonance compensation was a thing helped a lot. Effectively the teeth act as springs on smooth pulleys
About bearings, linier rod bearings anyway. I have a Sermoon D1, all the bearings are clocked at 12. So all motion forces act on 3 and 9 while 12 and 6 hold the carriage in Z direction. Doing that removed all the slop in my printer. And of course the machine is squared and the bed is trammed, having a quality square really helps. The frame doesn't have to be perfect as long as the rails are able to be, wasted a lot of time on that.
Note, if you do get it perfectly square you WILL find out if there's slop/tolerance gaps in your machine. Found this out on my old wooden Replicator, you could hear the carriage and bearings rattle after, even was visible on the prints. So, you might want some binding to press the bearings against something.
I just built my fysetc kit last week. I did not have the uneven foot issue you had at all. That is a crazy amount of bounce
Because it was a wrong part in the kit, I just reprinted and everything is fine. Plus it might depend on the date of the kit and the region.
You may want to rerun your IS results while adjusting your accel_per_hz. The messy x rail results have a super low power being a scale of 0-1.75 1e3. You are just measuring noise.
Thanks, will try to play around with that.
How do you do that? I haven't heard that before
@@benclimo461check the klipper documentation. You’ll use the accel_per_hz setting under input shaper. Make sure it’s with a nozzle ADXL or something. CAN board ADXLs aren’t very good for actual results
"why is the ldo kit so much?" well I thought I could do it cheaper and Im at the same price in the end
Yeah with DIY printers you really want to have a good kit already, because rebuilding the printer later with replacement parts and figuring out issues is not that fun.
I have fysetc v0.1 kit and I'm happy over 2 years. I had to change only X axis rail because of play. Otherwise all parts were good. I run CANbus ebb36 on my another printer vzbot235 where I use 500k baud rate with SKR Pico because with 250k ADXL results are not correct. So far no problem with overloading MCU. Just yesterday I updated klipper and had to update also MCU and EBB and it went smooth.
note: the print parts infill density
other note : why did you not try dowden condiguration
This isn’t an uncommon issue in precision automation. First thing I would want the cable connected at a place as far from a point of leverage as possible. So loop it down and connect it parallel with that linear support. Now it’s incapable of rocking the tool head no matter how much tension it comes under. I’ve never been in a position to redesign the cable however I can add length to it so it never goes into tension. Building a structure over the machine (or using the ceiling) to support excess cable. It basically “dangles”. We’re hillbillies but get the job done. 😂
4:08 ---i bet you didnt clean the rail , bearing block and lubricate it with proper silicone lubricant for this type of application (yes you need to do this with brand new rail ...purely because it isnt premium quality rail already lubed with 500$ per kg high speed machining grease , most of the time they need cleaning so clean it😅
Edit---
This is also created in manufacturing when plastic parts on the end of bearing block arent aligned well , balls inside get stuck on the edge of plastic when ball comes out of a steel part (block) .
Simple solution to it is to just tap the plastic up/down until it move smoothly / loose screws a little , align and it should work perfec
If not...you can unscrew plastic parts and check if the edge of plastic isn't already bend inside etc ...i use something sharp to cut edge the plastic hole at an angle so the balls can move smoothly in and out....works every time
PS--- idea for next episode
how heavy can the X axis be to still go 200mms with at least 5k acceleration
i bet you cant beat my old behemoth from like 9-10y ago .... around 4 kg X axis 😆😆 (yes , it was big...and heavy😆)
Nowadays you don't really need to do a lot of cleaning with rails from my experience. I just clean the existing oil from the ball bearings with IPA and use Mobill EP2 grease. Works perfectly on Fysetc rail, and all other rails.
@@PrintingPerspective AH i forgot
This is also created in manufacturing when plastic parts on the end of bearing block arent aligned well , balls inside get stuck on the edge of plastic when ball comes out of a steel part (block) .
Simple solution to it is to just tap the plastic up/down until it move smoothly / loose screws a little , align and it should work perfect
If not...you can unscrew plastic parts and check if the edge of plastic isn't already bend inside etc ...i use something sharp to cut edge the plastic hole at an angle so the balls can move smoothly in and out....works every time
On the rail I have one suggestion. I had a lose fitting rail and tried to squeeze it with a bench vise. Worked perfectly after. Not saying that's the way to go about. But there is nothing to lose anyway.
I wonder if this is why my fysetc 0.1 has always been useless over Y100. I have never been able to get it to stick properly to the back of the bed and the extreme position flexing the head would explain it.
Might be the case.
My kit from FormBot came with a preloaded X-axis rail.
Can you link the rails?
Unfortunately I can't because I don't think they sell those. I will suggest Fysetc to make it available for purchase.
Fysetc added the rail to buy to their Aliexpress store, please check my top pinned comment.
yep - having issues with my Fysetc 0.2 Pro... ran well at first (30 hours) but now having shutdown after 20 min with undervoltage errors. Now support is telling me to re-flash the firmware...
That kit such a headache...
thats a shame about the fsytec kit, I thought they would be better
True, I don't get how the quality of their kits can differ so much.
A Flashforge adventurer 5m Pro test would be very interesting as it shined in all the tests I saw regarding print quality. But as you mentioned in your videos it would be nice to know the real performance and not the performance on prints where everything looks good.
I will include it in one of the future print quality review videos.
I’m new to Klipper, how do you get to the screen that shows the input shaper graph?
Just go to the official Klipper documentation page and check measuring resonance tab. They're all the commands listed how to do it. :)
How does wire loom affect a larger voron like 300 mm?
My 300mm V2.4 is the most precise printer that I have today. But it has drag chains with multiple PTFE wires (not CAN BUS). The bigger the printer the less it should affect in theory because all the parts will be bigger and more rigid which minimizes the negative effect of the thick wire loom.
How do you wire the motors xy because the came with real short cables for the e3ez?
Extension cables or cutting the connector and resoldering the wires.
@@PrintingPerspective can you share your printer.cfg I am doing almost the same change on my fysetc
Im interested in getting canbus on the manta e3ez. Phaetus rapido/orbiter v2 running with apogee toolhead. Is this possible?
It has a CAN port, but I didn't have time to test how well it works.
@@PrintingPerspective no worries. I gave up on putting a BTT adxl on my printer. It shows up but the documentation is severely lacking
Yeah, they are lacking but my god, compared to Fysetc docs they seem good lol.
I have this in the config for the Manta E3EZ and BTT ADXL345 V1.0:
[adxl345]
cs_pin: PC15
spi_software_miso_pin: PC11
spi_software_mosi_pin: PC12
spi_software_sclk_pin: PC10
[resonance_tester]
accel_chip: adxl345
probe_points:
150,150, 25 # middle of bed as an example
You can tighten the carriages on the rails with the screws on the side im pretty sure
Those are just for disassembling the carriage, at least on MGN rails.
Awesome video. May I ask how you measured the Print Length Differences? :)
Well without having seen his model, it's a easy effect to make in a test print. You want peak acceleration, so a sharp turning point. A 4 leg star would be great. Print in vase mode or single wall, custom speed settings (could increase for every 20-50 layer). Please note that this will not be accurate (true measure), so you're only looking for the difference between X and Y
Thanks. I made a jig to measure length consistently with digidal calipers. There's a link in the video description to the updated version of tests, which includes that jig.
@@PrintingPerspective perfect thank you very much!
Replacing the stiff PTFE / Plastic Cable Loom in a Sleeving with one made of flexible Silicone Cables without Sleeving would probably have fixed the issue.
I am sure that would help, but considering how much rewiring you would need to do, CAN BUS would be so much easier if you don't run the Catalyst board.
Silicone isn't too great in places where you need to have some wire on wire movement. proper PTFE insulation on wires and slightly loose wire sleeve ( even PET one ) is great. On negative side wires with proper PTFE insulation are not cheap , on plus side they are kinda "soft". But i don't like it either and i will move it to CAN with proper cable for that , which is really nice and again - expensive.
Friends don't let friends Fysetc. Dissapointing to see they still have the same issues in v0.2 kits as the early v0 kits.