Sovol SV08 CoreXY Fix

A few things first:

• This mod will very likely improve your print quality but it also will most definitely void your printers warranty.
• This is still a work in progress. Stuff might change depending on the input i get
• sorry for the wall of text
• Please do not simply download and print the parts they are mostly for illustrative purposes. Measure the offsets on the belts of your printer and generate both sets with the correct offsets for your printer. You may make it even worse otherwise
• This is not a simple task. You have to partially disassemble your gantry and replace some important  parts. Before you tear your printer apart make sure you have tools, material and knowledge to put your printer back together.
• If you're running stable OpenSCAD (version 2021.01) this will take some time to render. Even in the preview. Using a current nightly (I ran it with 2024.09.10  from the snap store on linux) will cut this down to mere seconds. I hope that there will be a new stable release soon but even the nightly has currently a much better performance. I had no issues rendering this model with both versions
• Make sure you have the filament calibrated as good as reasonably possible. This is a structural print. Good layer adhesion is a must. If you are unsure, print a temp tower, look at the quality but more importantly break it apart. At what temperature does it get harder or impossible to break by hand? That's what you want to print with. If it doesn't look as good as 10°C cooler? Doesn't matter! Calibrate linear advance. Calibrate your top surface flow rate. You need an even surface. Do all the tests with the filament you're going to use.
   

I recently bought my SV08. After switching it to mainline Klipper and replacing that awful inductive probe with a Biqu microprobe, I started optimizing the printer. On functional parts with x or y dimensions over 150mm i noticed that the precision got worse. increasingly the larger the part. Calibration prints for belt tension helped to reduce this error significantly but i never manged to get it acceptable. I started digging into coreXY and finally found that the routing of the belts was not parallel to the moved axis. It's not that much. The most significant is the routing from the tension pulley to the tool head. But it is enough to cause slight deviations and also the belt tension will change slightly.  This is nothing the skew correction can fix. It needs to be fixed mechanically. what a bummer. I was thinking about shipping it back. In the end I decided against it and made a fix. The alternative would have been building a Voron 2.4 350. It takes longer and I don't have so much time to spare.  This took already way to long

How to calibrate hole sizes and the alignment guide: 

• If you have the standard (longer) M4 inserts from Ruthex or compatible ones (5.6mm hole, 8.1mm long) at hand, you can use carriage_test_demo. Otherwise change the dimensions in the settings section of the customizer, check generateTest and you can generate a new one with the required dimensions. 
• The ridge on top of the test piece must fit snugly in the underside of the 2020 extrusion of the x axis. If you can twist that part more than a tiny bit, increase parameter baseWProfilePin, if you need to force it, make it smaller
• the holes inside the ridge are M4 with an insert on the bottom. Check the fitment of the insert. You don't have to actually melt them in but many inserts have a small guide at one end. that should fit snugly but without force
• the larger hole on one side is for a 5mm axle. This should be a slight press fit. If you don't use carbon or glass fiber filled material, you can print the hole 0.1 to 0.2mm to small and use a 5mm reamer or high quality drill bit. With filled filament you don't want to do that without tungsten carbide tools. Steel won't last for much more than one hole
• the small hole is for M3 and it should be a loose fit.   
• If all holes have the same dimensional error, consider trying the hole expansion in your slicer.

How to measure the offsets:

• Move your gantry up from the build plate for better access. Between 100 to 200mm should work well.
• Turn your printer off or at least disable the steppers
• Best unplug your steppers on the gantry. Moving the tool head by hand to fast can damage your stepper drivers
• You there are 8 measurements to take. All the belt sections along the x axis will define your y offsets, the belt sections along the y-rails that go around the pulleys in the carriages will define your x offsets. Take your time 
• left side of the printer standing in front of it are also the left side offsets (x = 0)
• the rear pulley in the y-carriage is a little bit further back and also closer to the outside. The belt goes around it to the back of the printer
• for the measurement itself spread the pulleys as far apart as possible. For the measurements on the left side of your tool head move it all the way to the right. Now take your caliper with a depth gauge and measure the distance between the rail behind the belts left next to the tool head. As soon as you touch the belt with your caliper and have the pin against the rail, zero it. Then go to the other end and measure for the upper and lower belt individually. The relative measurement tells you your left y offsets. on the left side the upper belt is the rear pulley, the lower ist the front pulley. On the right side it's opposite. Repeat for the right side of the tool head
• I measured the x offsets against the 2020 extrusions. should be close enough. The sections between from the rear to the tension pulley are irrelevant as they don't cause a movement of the axis. Move your tool head all the way to the back, zero your caliper at the tensioner against the belt farther away from the 2020 and measure your offset in front of the y carriage on both sides. Here you have to invert your value. A negative measurement is a positive offset and vise versa. These are the x offsets for  the front pulleys.
• move your tool head all the way to the back. Measure the offsets of the left upper and right lower belts. Set zero at the rear close to the pulleys of the motor. These offsets are also inverted and for the rear pulleys
• you should have all eight offsets. Don't forget to plug in your steppers!
• enter the measured offsets in the customizer of OpenSCAD - they are all the way on top
• uncheck generateTest and confirm your other options
• render (F6) and export the stl (F7)
• toggle generateRight and export the second set.

Material and print settings:

I used glass fiber reinforced PETG from Tinmorry. It's affordable, prints really well but you'll need at least a hardened steel nozzle and it often needs a good bit of drying. It also required quite unexpected high nozzle temperatures. After a temp tower test i set it up for 285°C. But on the other hand they state 260 to 290°C so it's alright. Below 275 the layer adhesion was fairly poor. If you use that stuff put some pva based adhesive (glue stick, 3d milk…) on your build plate.

These were my parameters in Orca slicer:

• 0.6mm nozzle (brass with tungsten carbide tip)
• 4 perimeters
• 2.4mm top/bottom
• 25% gyroid
• 1 concentric top/bottom layer
• adaptive layer height between 0.2 and 0.35mm (you can temporarily set your extruder limits to these values) - split the model in objects and set it automatically for each piece with a quality setting of 0.1. This is mainly for better dimensional accuracy in z
• Supports are only required for to holes of the pins - if you are using the pinWithHead option. Everything else prints entirely without it
• there is no actual need for first layer expansion as everything should be equipped with chamfers.
• I printed both sets individually in the center of the bed. The error with stock hardware was there small enough to work

Hardware:

• 6 heat set inserts M4 around 8mm long
• 6 M4x40 in with the default top thickness - best 6 with a low head
• 4 5mm pins/rods 40mm long for the provided models. 35 to 45mm will most likely work but it's untested. You shouldn't use the old ones. they are to short and will not have enough contact with the printed parts

Assembly and installation:

• check fitment of your pins in the top and bottom parts (straight holes on top, standoffs on the bottom). If they slide out on their own, then they are to loose - check your calibration and best print again sorry. Otherwise you can leave them in the bigger lower part
• install inserts in the bottom of the biggest part of each set (3 holes in straight line)
• go to your printer and move the gantry about halfway up
• have someone to help you or something to temporarily support the x axis ready (adjust z accordingly)
•  turn your printer off and disconnect your steppers
• cover your hotbed with cardboard, paper towel or something similar to avoid grease from the rail on the bed
• fix your toolhead securely approx. in the center of the x axis. use tape, zip ties, magnets on top of the rail left and right to the toolhead. This prevents the x-carriage from sliding of the rail in the next steps. Trust me that will ruin your day.
• release the belt tension. losen the two screws on the side of each tensioner and then turn the screw between the belts all the way clockwise. (I know tightening a screw to release tension…)
• disconnect the belts from the toolhead (2 M3 screws on each side)
• Remove the pulleys from the y-carriage. there are M4 Insert in the stock parts below the pins. You can use one of the long M4 screws to gently push them out after removing the tiny screws on top. Careful above the lower pulleys are some tiny spacers. 
• put the loose belt out of the way.
• Remove the Screws holding the x-axis 2020 extrusion on top
• remove the for m3 screws on each side underneath the linear rails. after the first side you'll need to support it or have someone holding it up
• lower the assembly, pull the old carriages off to the side and lower that extrusion with your toolhead gently on your covered bed. If you didn't take any measures to prevent the carriage from sliding off the rail then there is nothing from stopping that now.
• install the lower part of the left and right carriage to the linear rails with the same M3 screws. don't tighten them all the way yet.
• put the rail on the carriages (much more convenient right?)
• put one screw temporary through the extrusion to prevent it from dropping
• put the pulleys in. the ones with teeth are in the front
• put the printed spacers in. the short ones go on top of the upper pulley. the small ring on one end first (sits against the inner bearing race of the pulley). You can also use that narrow spacer but they tend to bite into the cover and loose their function
• put the belt in the carriages, remove the temporary screw and put the top cover on and place all the screws loosely. Don't worry at this point it felt quite wobbly
• attach the belts to the toolhead. I hat to shorten mines by one tooth on one side only.
• move the x-axis all the way back against the motor mounts. start tightening all the screws on the carriages. pay attention that both side touch the motor mounts. if one side doesn't touch after tightening readjust. A tiny amount can be fixed with belt tension. It should be less than a millimeter. The better you get this alignment, the better for your printer
•  start tensioning your belts. Turn the screws back in equal amounts. Three to four full turns is a good start. If you're halfway out on the tensioner, you'll have have to shorten the belt. Tighten the tension screws all the way (springs fully compressed), remove the belts from one side, slide the belts out to the side and slide them back in one tooth further away from the end and try again
• after putting proper tension back on the belts, take a look at the sections affected by this change. If everything went alright the belt now run parallel to the moved axes. when you look at the belts along the x axis from the top they now run at the same distance. If there is a visible offset in any of these sections then there is something wrong. Either wrong offsets or the belt came off a pulley.
• Move your toolhead in all four corners my hand. Is there anything in the may? Does the extruder motor clear the new mount all the way into the mechanical limit?
• Plug your steppers back in
• congrats the installation is complete

post installation

• confirm your steppers are connected
• home your printer.
• Don't be shocked when the gantry runs much smoother and more quiet. 
• move y with the steppers back until it's just about to collide. If your offsets are close to mine the 10mm is a good starting point
• move the x axis towards zero. I started with 15mm im 1mm steps
• open your printer.cfg and adjust the homing positions of x and y by the amount left in the previous test
• disable nozzle cleaning for now. all the positions in there need to be offset by the change.
• If you use that script for z offset calibration with the pressure sensor then disable that too
• check that QGL and bed Leveling remain within the new limits 
• With mainline klipper you'll measure resonances again
• Calibrate eqal belt tension

Pros:

• For more rigid
• better access the belt drive in future
• less intense resonances - amplitude reduced by almost 50% in my case (see y results before [20241222] and after [20241230])
• fully parametric
• printed in less than one hour per side with a really slow profile

Cons:

• Heavy - approx 36g added weight
• decreases the x and y travel slightly (x approx sum of front pulley offsets and y the bigger of the front pulleys) in my case x is decreased to 350mm and y to 361mm
• needs additional hardware
• voids useless warranty
• unsuitable materials like PLA might fail unexpected mid print
• complex error prone measuring of offsets
   

Updates:

• V1.1 
  • improved pin hole generation. Now you can use the pin head option also for a smaller hole below a regular pin so it can't drop down and there is a way to push a pin out if needed. With the default settings the hole will be sized for a M3 Voron style insert. That resembles the same function as on the original parts
  • New parameter holeDia5mmShaftTop. You can make the holes in the top parts a bit bigger so you won't pull out the pins when removing the top 
  • removed the reference to the furniture bolts as they didn't work as pins. they started to buckle