Motor replacement. I am ANGRY

I’ve just been informed that I have to send my intermittent X motor back somewhere before they will ship me a replacement…

WTF?? I’ve been struggling with this literally since I got the machine, I informed them IMMEDIATELY that there were issues. I’ve been struggling with this for many weeks… Now this??

Plus the defective filament runout sensor

Plus the other motor currently on Z which has been totally inop at times but is working at the moment.

Color me seriously pissed off.

Sorry for the problem you ran into. Could you kindly PM the ticket ID you received and I can look into this case.

Sorry again.

Edwin

I think it’s this number: 78664 Been corresponding with Zero on this motor since literally the day I got the machine.

It may not be all in one thread.

Initially I thought I’d done something wrong, then the problem seemed to resolve when I traded this motor from right side Z to X.
That went ok for a while, then I started having what looked like a slipping magnetic bed, except the displacement was always exactly along the X axis, and the bed showed no indication of ever moving.

Now I’m getting a slip of a few mm at sort of random points during a given print, or it prints ok, just seems random.

I am printing ABS in the enclosure, so the temps are elevated. I can’t say for sure if the problem tracks with temperature. Certainly the initial total fail of this motor when it was in Z was anything but elevated temp. It literally failed to work the first time I turned on the machine, before I even took the enclosure out of the packing. No amount of cable swapping or plug checking would get it to work, but when I swapped it to the X position, it worked fine, at least for a while. Then I got distracted by the filament switch problem.

As upsetting as this is at this moment, I’m willing to spend some time helping to find the root cause.
I’m the lead hardware engineer for backcountry access, and I have a few years experience in motion control with Hunter Douglas (expensive motorized window coverings)
I have some 40+ years of embedded systems design experience.

With the BCA products, it’s quite literally true that if my products don’t work, someone’s going to die.
I did offer a design review under NDA.

That said, my “spare time” is limited, and I didn’t buy this machine to spread it out on my workbench and diagnose it.
I’m trying to get projects done, and when this thing works, it’s glorious, but those moments are not common at this point.
I have prints that look something inbetween machined and molded quality, better than I’ve ever seen from a filament printer, including ones done by printing companies prototyping for BCA.

From what I see on the forums, you’ve got a significant issue with the motors. Many of the “bed leveling” problems may be one side of the Z failing intermittently.

Y rails may be failing to move the print for a while, causing these massive printhead messes. Similar with X I suppose, though in my case it either goes properly, or it misses a few steps and then works ok for a while again.

I might suggest a diagnostic mode that returns a given axis to the limit switch position say once per layer. You believe you are N steps away from the switch detect, so stepping back in that direction by N steps should result in a switch detect again (plus or minus some small slop) If you don’t detect the switch when you should, that would indicate that some steps were missed. Back at Hunter Douglas I created an automated test system for problems like this that logged the errors, and emailed me automatically when a problem was detected, taking a picture with a web camera, and giving me a history plot. Very revealing. I was able to improve their positioning accuracy significantly and shake out some very obscure faults that would occur maybe once a year in a real install by running those motors 24/7 and monitoring every move.

Anyway, I hope you can understand my frustration. Occasional periods of wonderful performance from the machine, marred by a string of problems.
Intermittent faults are always challenging.

Hi David,

I have informed Zero of this case. He will reply to you soon.

Sorry for the inconvenience.

Edwin

Ok. I will be able to spend some time with the machine in about 8 or 9 hours. Please advise

The Austrian company ams makes a line of magnetic rotary encoders. A magnet is mounted on the end of the shaft and the chip uses the Hall effect to sense its rotary position. They have a number of chips available; for example the AS5048 has 14-bit angular resolution, plenty accurate enough to detect missed steps even with some derating. This particular part might be overkill for closed-loop sensing in the SM 2.0, given that they’re not even microstepping, but they’ve got cheaper parts at lower resolution as well. Adding such parts into the linear modules would have added $25-$30 per linear module to final cost, which in my opinion would have been well worth paying.

I have looked into replacing one of the end caps in the linear module with a part that would allow a sensor based on this chip to be mounted permanently. It’s mechanically not all that difficult, assuming basic machining capability.

I was trying to propose a solution using what they already have in place.

I’ve seen those sensors, but they are not inexpensive. I don’t think we need closed loop control here, (though it would be nice) but a single hall sensor and a magnet sending a pulse on every full rotation would work. All we really need to detect is that suddenly we are not where we should be.

Making a last test print before dismounting my X rail and sending it in. They now require this (per email from Zero) before they will ship a replacement rail, and I’ll be totally down till the new one arrives, whenever that might be.

Not good.

I figured, but the ams sensors seem to be the cheapest way of retrofitting a position sensor onto the linear modules.

The 14-bit ones are about $10 single unit cost. The cheap ones are around $3. Not popcorn parts, to be sure, but not very much money if you’re going to spend even an hour of your time on getting one working.

I didn’t say it explicitly, but in your situation you’re trying to isolate the defect in your linear module. In that case what you need is a data acquisition system, for which this part is ideally suited. Sure it could also be integrated with a closed-loop controller, but it’s useful even without it.

I’m proposing this as a diagnostic mode for all printers. You wouldn’t enable it unless you suspect a problem, but it would help sm figure out what’s going on I think. It seems like there are a lot of people with motor problems.

It would do something, but couldn’t function reliably. The problem is that, in general, if a stepper motor is missing steps during ordinary operation it will also miss steps in diagnostic operation. Take the simplest model where there’s a uniform probability that a step is missed. The accumulated error is a random walk around zero. In one dimension the walker returns to the origin infinitely often, so accumulated error remains clustered at zero. The best you can do is to make the duration of ordinary operation much longer than the duration of diagnostic operation. In that case you’ve got a decent chance of showing the error.

In the best of circumstances, you’d have false negatives. But it’s worse than that. The limit switches in the linear modules are very inexpensive and are not designed for repeatable triggering at the same position. They will certainly trigger, so they’re fine for a limit switch to protect the machine from itself. But they’re not good as a zero-detection switch. An optical interrupter is quite repeatable, but that’s not what’s in the machines.

What’s needed for self-diagnostics is some kind of redundancy, so that you’re not comparing a potentially-faulty component to its own behavior. A rotation sensor on the lead screw would accomplish that, as I mentioned above. If you’re looking for an easier add-on diagnostic, a more traditional DRO linear scale that followed the carriage would also work. There are some available now at 300 mm travel at less than $100. They’d be far easier to retrofit on the side of a linear module for diagnostic purposes than a rotation sensor.

I’m aware of the limitations, I did a lot of years at Hunter Douglas designing motorized window coverings and documenting various issues. But an onboard diagnostic could be done and it would help people document issues and help SM figure out what’s happening in the field.

If I were doing this as a project like I used to, I’d use a high rez stringpot with quadrature encoder and a microcontroller to command the motor and record errors. But I’m sure SM has that capability.

It’s folks in the field like me with one machine and not wanting to spend 2 months to convince SM that there is a real problem.

Well, a week later the replacement is installed. Waiting for the bed to heat up now.