LR4 Modifications for 80mm Closed Loop Steppers?

LowRider CNC Hardware Development
1

Some background: I’m currently troubleshooting my recent LR3 build as I start printing a LR4 out of PCCF. My LR3 is skipping steps (seems like on X mainly) randomly after cutting for > 1 hour (<0C temperature ambient) then Y will go out/derail as I then start cutting full depth. Y may be randomly skipping as well. This doesn’t seem like a torque/grub slipping issue but just the motors giving up for a few cm of steps intermittently.

Now I probably just need to just tune this, I don’t think my speeds are too aggressive. Slotting 7mm DOC, 1/8th bit, 10mm/s in Pine plywood isn’t that much load right? I’m using chip loads and RPM from a calculator. But as I rechecked all my wiring I got the 5V mixed up on the end stop wiring messed up on my SKR board and burnt it up :face_with_symbols_over_mouth:

My next attempt is going to be with fresh set of drivers and a new control board, plus adding some 40mm fans to the enclosure, so that may sort it out.

As I wait for the replacement board and to try and tune things it got me thinking. The X axis has 1/2 the torque as Y due to the single vs. dual steppers right? Ignoring rigidity concerns on Y and derailing with too much torque on the X axis for now, 3/4" oak plywood isn’t cheap these days.

StepperOnline has a longer closed-loop stepper https://www.omc-stepperonline.com/nema-17-closed-loop-stepper-motor-80ncm-113-29oz-in-with-encoder-1000ppr-4000cpr-17hs24-2004d-e1k and driver board that seems like it might work with the Jackpot passthrough stepper controls

Their stepper driver appears compatible with 5V signals:

https://www.omc-stepperonline.com/closed-loop-stepper-driver-v4-1-0-8-0a-24-48vdc-for-nema-17-23-24-stepper-motor-cl57t-v41

The biggest challenge here is the 80mm motor length for the added torque and rotary encoder.

Looking over the LR4 parts, it seems X stepper length is possibly unnecessarily constrained by the part layout? This seems easy enough to just cut out and possibly the biggest benefit?

Y seems a little trickier, the Z endstop mounting would need to be modified significantly and possibly rotating the stop. But it seems like this should be feasible to accommodate an 80mm motor, there’s not a lot of clearance issues intruding inboard in this area to add bracing?

Z seems less interesting to make closed loop or make stronger. Z rapids don’t seem AS interesting (I’m wanting to do some 3D carving eventually, but even then X/Y clearing should take most of the time), and much more challenging to adapt the design for. And assuming Z doesn’t collapse under the weight of the tool I would assume there’s not much load induced by the motion.

BOM cost seems reasonable, <200 for 3 motors and drivers + cables possibly and if the fault signal works to indicate repeated lost steps (Seems a little unclear) then that can halt a job with minimal material damage which is even better. I recall reading the Jackpot has a fault input although understanding how the output triggers from the drivers may be challenging.

I’ve seen a thread on using a closed loop stepper for fast Z on the MPCNC, but not much else. Has anyone else tried this?

1 Like
2

What do you have motor current set to?

TMC stepper drivers will cut out if they overheat. This is separate from the motor overheating, and quite a bit hotter than the motors should be allowed to get.

The motors should be kept under 50°C. The drivers can get hotter, but if they do, they will just shut off until they cool.

Stepper motors skipping steps makes a grinding noise (actually the motor just jumping quickly to the previous step, no gears or grimding involved.)

1 Like
3

This is unlikely to help you. Closed loop stepper control seems like it would solve all problems but in reality it is not a working implementation in FluidNC (or most control software we use).

This is much more likely to be the issue, or other mechanical limits.

If, for example, you’re trying to move more rapidly than whatever your spindle/cutter can do in your hardwood material, then you’ll still have an unsuitable result.

I’d encourage you to work on troubleshooting the setup you have and abandon attempts to fix it with closed loop steppper control.

4

Everything is still at the 0.9A defaults in the firmware. I know that can be changed, but I was going to add cooling fans to the driver enclosure first and otherwise try and get it stable, then play with currents to speed things up later.

And, the garage I’m running this in is very cold this time of year. Motors are definitely staying cool to the touch at-least shortly after starting a job and when sitting with motors powered but otherwise idle. Is there a debug command that shows if a driver overheat has happened or the driver temperature I can probe periodically in the SKR?

To be clear, I’m 100% wanting to get a stable LR4 with the normal v1e kit steppers and drivers and see how far I can push that. I’ve seen some others get impressive results. So I don’t think I’m anywhere near the limits of what should be possible.

My understanding here is with the external stepper drivers, you can treat these as a normal open loop control and the external stepper driver handles the difficult part. You just don’t get real-time feedback to the controller once you start losing steps or when the driver is nearing its limits. But FluidNC supports a fault pin: Control (Inputs) | Wiki.js which might work with the alarm output of the driver. But the documentation of that in the StepperOnline drivers is pretty vague as to what conditions trigger the fault LEDs vs the fault output, would have to experiment there.

5

The difficult part is having a working machine, not steppers that try to recover lost steps. You can’t solve a machine problem by upgrading one part that likely isn’t the reason the system isn’t working correctly. You then just move the failure point closer to the actual problem.

Said differently, it doesn’t matter if the steppers eventually recover if you’ve already ruined your part (due to other machine dynamics.)

6

Again, to be clear, I’m going to troubleshoot and get a working machine first without using closed loop steppers. I’m not trying to solve my current problems with a closed loop stepper. It’s the same reason I haven’t turned up past 0.9A yet.

I think I have my skipping issues sorted today, some combination of new SKR board, fans added for stepper cooling, or extra electrical tape around connectors (ESD from the vacuum hose?) seems to have made it stable for 3 hours of cuts. Of course then the collet damage form the previous cuts then ran and I have to wait for a new 1/8th collet.

“Lost steps” in the terms of the steppers with microstepping are going to be generally small, 0.1mm or less if detected within a few microsteps or encoder steps? I think most projects in wood can tolerate this as a defect if the job stops and machine deflection under heavy load should be worse.

Now that I have a better working machine, it’s more obvious the rigidity of the LR3 doesn’t really need stronger steppers than what I could drive with ~1.7A so going to a full 80mm long stepper is unnecessary.

The fault detection still seems very useful if it works. Only minor Y mount modifications may be needed for a comparable motor with encoder, i.e. https://www.omc-stepperonline.com/nema-17-closed-loop-stepper-motor-56ncm-79-3oz-in-with-magnetic-encoder-1000ppr-4000cpr-17hs19-2004-me1k and being able to detect skipped steps and halt seems like a nice capability in addition to running a little closer to the stepper limits.

I have a single driver and motor on order to experiment and see how the fault detection and failure modes work on the X axis with these drivers and to compare how fast they can be driven3.