Ender 3V2 dual Z axis upgrade

Seems simple enough but what I don’t know is will the dual steppers stay in sync. If they don’t turn together then the gantry will be out of level. Also unless you get the same type they may not increment the same amount.

Has anyone made this upgrade.

I’ve never done the upgrade, but yes, you absolutely need matching stepper motors to ensure that the degrees per step are the same, the step direction (obviously) is the same and that the current consumption is roughly equal. The only potential problem I see is that the stepper drivers on the motherboard should be able to handle double the current. Some motherboards get around this by having a separate driver for each stepper, but controlled from a common output on the microcontroller. The common output is important so that the steppers move in sync and don’t require an upgrade to firmware. If they don’t use a common output, then every time you upgrade the firmware, you’re going to have to remember to re-enable dual Z-stepper outputs.

I have a cr-20 and an Ender 3 v2, both of which have dual z.

I don’t know if my thought process is correct but I did have concerns about keeping the steppers in time with each other. I’ll just ramble off my decision making process for you to take as food for thought.

I have had my Cr-20 pro for a bit over a year, I did dual z on it because I made it direct drive and wanted more support for the added weight. The Cr-20 has a fixed bed, as in no bed levelling knobs.
I put the Fysetc dual z system on it that is belt driven with only one stepper. It has worked great since I installed it, but I changed it for reasons below.

I just got the Ender as part of the college program I’m doing so I decided to upgrade it in a similar fashion to my cr-20 by doing direct drive and dual z. They are basically the same printer with the difference being that the Ender has an adjustable levelling bed.

When I ordered the parts for the Ender from 3dpc the Fysetc belt system was out of stock so I ordered the Creality dual stepper kit instead.

So we now have a printer with a fixed bed and one that is adjustable and three ways to do dual z on each

  1. One stepper driving two lead screws synced with a belt
  2. Two steppers in parallel sharing one driver
  3. Two steppers on individual drivers for individual control

As far as I’m concerned, option 2 is not an option. The problem is, as you were initially concerned with, the motors aren’t synced. When you power down the printer the steppers can move independently of eachother. Or they can miss steps since they are two steppers sharing the current from one driver and grow out of sync. It may only be a tiny amount but it leads to a variable that causes inconsistency. You would regularly need to manually measure to tram the gantry to the bed to make sure things stay parallel.

So now we have option 1 or 3.
On my current setup I am using option 3 on the Cr-20 with the fixed bed and option 1 on the Ender.

Let’s start with my Cr-20. It is using two steppers plugged into an SKR 2 board with 5 stepper drivers. Each z motor has independent control and the printer has a bltouch. With this setup I can use G35 in Marlin which does auto gantry levelling. Before each print the carriage goes back and forth between each side of the bed and probes to see how far out of parallel the gantry is to the bed. Each stepper can move independently so it is able to auto correct and it will get within 0.01mm parallel across the width of the bed. This works well because the bed is fixed so it is a constant reference point to probe off of. As long as my bed is square to the uprights everything is happy.

On the Ender I have the belt driven system that used to be on the Cr-20. In this case the gantry was set up level to the frame of the printer, square to the uprights and the bed is the moving target. Instead of the gantry getting trammed to the bed, the bed gets levelled to the gantry. The belt keeps both sides in sync with eachother so if you manually move the gantry up or bump something with the printer off both sides stay together so long as the belt doesn’t slip (which is very unlikely).

So long story short, on an Ender I recommend the belt system. It is easier to install, doesn’t require a board upgrade for more drivers, doesn’t need a bltouch. If you’re careful setting it up it keeps everything moving together nicely and doesn’t really need any maintenance.
If however you are feeling adventurous the independent two stepper option does have potential for really good accuracy and piece of mind knowing that every time it is only out of alignment by 0.01mm at most.

I would happily do the belt system on both printers and the only reason I went down this rabbit hole was because 3dpc only had the dual stepper version in stock at the time which forced me into a board upgrade and a whole convoluted upgrade path.

Is there any problem with backlash on the belt system IE: slop introduced when the belt is reversed, it might create inaccuracy?

What about the belt stretching over time, will this change the step relationship between the two screws?

I have a CR touch on my printer which runs a level check before every print. If I had option two, both steppers running off of one driver and if there is an inaccuracy introduced because of that, will the level check not take that into account when it runs and compensate for that. I’m not opposed to the belt system but they do seem to much more expensive with fewer parts.

Another problem with option two is it appears that with only one driver the torque sent to each stepper is half of what a single steeper would receive. Reading reviews of these systems the posters don’t seem to be bothered by it but my “always looking for a conspiracy mind” thinks it could be a problem.

There doesn’t seem to be any backlash issue with the belt. It is the same as using belts for the X and Y axis. The belt systems have a tensioner, I never once had to adjust it after the initial install. Your Y axis belt sees way more strain with the inertia of the bed whipping back and forth and they don’t stretch that much so on the Z it is fine.

On a printer there needs to be several things that are perpendicular or parallel.
Your frame is fixed and should be square when you build the printer
The X gantry should be perpendicular to the frame uprights. This is somewhat fixed when you assemble things but should be checked periodically.
The bed on an Ender is not fixed so it gets adjusted with the levelling knobs to make it parallel to the gantry which was supposedly set up correctly to begin with and gauged off of the frame for alignment.

The problem with doing two steppers and trusting a touch probe on a printer where the bed is adjustable is that the probe is assuming your bed plane is perfectly perpendicular to the uprights. You are now using a probe that is trying to create a reference for where it is in space by measuring off of something that is floating (an adjustable bed on springs).

Your bed probe should not be used to compensate for the tilt of your bed, it should only be used to map the deviation in flatness so that the printer can follow the contours. If things are set up properly you should not have tilt in your bed. You can’t check your bed tilt if your X gantry is also changing tilt. You could get them parallel to each other but how do you know they aren’t both equally out of square to the uprights.

I am using two steppers on two drivers only on my printer with a fixed bed. Theoretically my bed is nicely perpendicular to my uprights (it isn’t exactly and I will try to improve it when my 0.1mm shims show up) but either way in this case the bed is moving back and is coplanar to the plane that carriage moves on and stays square to the uprights.
My X gantry in this case is the moving target but when it probes it can sort out where it should be because it is probing a bed that always stays on a fixed plane.

The whole point of this is that you need to have only one thing that has potential for getting out of alignment. In your case that is the bed.
Using a timing belt for the two lead screws keeps both sides pretty well locked together so you aren’t introducing a second moving target like you would with two steppers.
None of these solutions are perfect and all of them require care in initial setup. The game is just to gain consistency and remove places that require your input and can introduce human error.
If you print a cube you want it to be square, not a parallelogram.

I agree that it is hard to trust reviews. I only found one comment when I was researching that shared my suspicion that dual steppers shouldn’t be used on a bed with springs. Was he a nut job that just fuelled my dive into a rabbit hole? Maybe. But this makes sense in my head so I’m going with it.

I’m sure between my two posts I have written way more than you were looking for so hopefully it makes sense and if somebody else wants to jump in to tell me I’m dumb, go nuts. I’d love to know if I’m missing something.

1 Like

wouldn’t it be better to have 2 steppers, 2 drivers, and home both individually? Seems to be splitting them from one driver or trying to mechanically link them could introduce problems that might not be apparent. anti-backlash measures only work well when the parts are new. I see now BTT has feedback options to add to the steppers to provide closed-loop operation so you’re sure they are synced and homed properly. The SKR V2 board I just put on my printer already had headers on it for this to work

My Cr-20 has an SKR 2 with two steppers that “home” independently using G34 in Marlin. They get within 0.01mm of being parallel to the bed which is definitely far more accurate than I can set up with my belt system and a pair of digital calipers.

My point in my previous post was that this doesn’t work if your bed is a moving target since if your bed is tilted to one side using G34 will cause the gantry to match the tilt of your bed. If you printed a cube it would cause the top and bottom to not be square to the sides.
On an Ender you would first need to make sure that your bed is a perpendicular plane to the uprights before using G34. In that case you would probably have to level your bed using a dial gauge off of the frame.
If your gantry is ‘locked’ for alignment then you can use assisted tramming to probe the four corners and adjust the bed levelling nuts

My CR-20 the bed is fixed. An ender has an adjustable bed. I did dual stepper on the CR-20 with G34 and belt synced on the Ender and use assisted tramming.
G34 aligns the gantry to the bed.
Assisted tramming aligns the bed to the gantry. It tells you how out each corner is and you manually turn the nut and remeasure with the probe, it’s just more accurate than the piece of paper and a feeling method.

After having the belt system on my CR-20 for a year I did notice that the gantry was a bit out of square. I had never checked it again after the initial installation so I’m not sure if it went out of adjustment over time or if I was just sloppy with the installation. I did also move that printer around a ton in the final week before doing the dual stepper and SKR 2 swap so that may have knocked things out a bit. It was only out a small amount.

I was thinking about those BTT closed loop steppers and might look further into them down the road for the sake of having a check system to make sure they are staying in step.

As far as backlash is concerned I feel like there is more backlash in the lead screw than what is introduced by the timing belt. I’m not sure that the belt introduces an appreciable amount of slop in the system, it felt pretty directly linked in my very unscientific testing. Nuances of backlash in mechanical designs is not an area I have any expertise in though.

Thanks for the help.
I have been do some research aka: watching YouTube videos, of which there are many and Ivan at My Tech Fun has installed both on Ender3 V2’s with some comparisons. Basically in his opinion and he seems fairly knowledgeable, that the stepper systems are a bit more accurate while the belts being less accurate are more stable. The steppers work well under power but don’t touch actually don’t even look at them when turned off or they will become out of sync.

Making everything square is part of the installation since some parts will be knocked out of alignment while being worked on. I would think that periodic checks should be regular maintenance.

As far as using the G34 code on an Ender 3 series I don’t think that is possible since there is only the one driver and it drives both steppers. Replacing the motherboard with dual z axis drivers isn’t high on my agenda right now.

There seems to be trade offs with both systems, with the majority of installs being the stepper version, maybe because of the extra cost of the belt type. I am leaning towards the belt system. I can’t find either at 3D Printing Canada. Maybe discontinued???

The only thing they all have in common is that the hardware is crap. Poor quality screws. More research to follow.

I think accuracy with the belt mostly comes down to how much care you take setting it up but once it’s on it seems to stay synced quite well.
It does seem to add more mechanical resistance so the stepper motor might have to work harder but the amount that Z moves compared to X and Y is so small I have never noticed the stepper getting hot, even in a 50* enclosure. The resistance might be offset or cancelled since you’re now lifting from both sides instead of only lifting something at the end of a lever, but that’s just speculation.

My Tech Fun is great, that guy always seems to have solid information.

3DPC had the Fysetc kit in stock a few months back, I’m not sure it’s discontinued it’s more likely just difficult to restock like a lot of products are these days.

There is this one from TH3D. It has good reviews and is in stock.

When you do get something installed let us know what your ended up going with and what your thoughts are on it.

This is my stupid question for the day.

All the kits are marketed for the Ender 3 but as far as I know the Ender 3/ 3 Pro/ and 3V2 all have the same frame dimensions and these kits SHOULD fit on any of them. The Ender 3 Max is a different sized frame and they won’t fit on them. Am I right on this or have I figured this all wrong. I’m planning on ordering a belt system tomorrow if it will fit my Ender 3V2 model.

@Loosenut it should be fine. I have the Fysetc one installed on an Ender 3 v2 even though it doesn’t specifically list that printer and that TH3D one should fit the same.

Well it’s done and will arrive in 3 weeks to eternity with the current shipping model.

I was temped to just make my own since I can make all the brackets and bearing mounts. I have the bearings and threaded nut, all I would need is the rods, belt and timing wheels plus the hardware. That is the real kicker here, collecting all the right bolts, nuts and t nuts ETC, it’s not like I can just go to a local hardware store and find them. I finally decided that I wouldn’t save enough to make it worth while.

1 Like