Linear Stage Rebuild

It is now time to tackle the more serious problem - the vertical banding that is showing up in my test prints.  This will be more challenging than the extruder problem I dealt with earlier because it has no obvious cause and will require some investigation.

The pattern visible on the test print does leave some hints.  If you look closely you will see that along with the pronounced line at the center, there additional regularly spaced lines on each side.  Repeating patterns in a print are helpful, because you can measure their spacing and immediately narrow down possible causes.

In this case the MX80L datasheet contains a rather suspicious number.

Knowing that the print artifacts have the same spacing as the MX80L's electrical pitch narrows down the range of possible problems quite a bit.

A very easy test, and therefore the first test I performed, is to check the resistance of each phase of the servo.  We know from the above table that it should be 4.3 ohms.

Two of the three phases are very close to 4.3 ohms which is good, but the third phase measures 6.2 ohms.  Motor windings almost always fail partially shorted or open, so seeing one phase of the motor showing 2 ohms more resistance than expected is extremely weird, and more or less confirms that the print artifact is originating from the x axis servo.

I will confess that this is the first time I checked the winding resistances.  The servos seemed to operate perfectly, so I assumed they were fine.  Its really quite impressive that the servo looked so good despite a failed winding.

This is an annoying problem but it was easy to track down, which is a relief.  In the worst case I can simply buy a replacement servo and swap it in.  I've always wanted to take apart am MX80L and this is a good excuse to do so.

Before proceeding any further I should make it clear that you should never disassemble a working cross-roller stage.  As soon as the two halves slide apart the bearings will likely be knocked out of alignment which may permanently degrade its performance.

The cables are fully modular and disconnect from the stage so it is quite easy to pull it off for further examination.

Complete disassembly of the linear stage starts by removing the cross-roller endstops.

The stage also has a mechanical stop that can be disengaged by backing out the center screw of the linear rail.

With these two steps complete the linear stage is free to slide apart.  There is actually enough clearance to access all the internal components of the stage without disengaging the bearing, but I've always wanted to take one apart so I decided to fully separate the stage into its two halves.

You can see the roller element assemblies contain a small gear that keeps them correctly aligned.  Under ordinary circumstances the anti-creep pinion is not necessary, but for extremely high performance applications the rollers may slip.  Anti-creep forces the cylinders to remain in alignment and extends the lifespan of the bearing.

The servo is really quite beautiful when separated.  I think I'll put it on display if it turns out to be beyond repair.

Disconnecting the cables allows further disassembly of the stage.

The stator assembly is held in place by two very small screws; nothing more is necessary because when assembled the stator is held place by the attractive force of the table magnets.

The stator assembly includes a PCB that handles the windings, hall, and thermal sensors.
The dab of white goop is thermal paste that connects the temperature sensor to the windings.

Pulling away the PCB reveals how the windings are arranged and makes the problem fairly obvious.  Each phase of the motor is split into two parallel coils and arranged in a wye configuration.  One of the six coils in this motor has failed open.

I desoldered the problematic winding and confirmed that one coil is indeed broken.  I was hoping the problem would be a poor connection between the magnet wire and the PCB, but the winding has simply failed open somewhere within the stator.

Unfortunately this is where the journey ends.  I am not prepared to rewind the motor, and the windings show no indication of physical damage or hints of where the break might be.

The new plan is to reassemble the stage and see if it still functions after being taken apart.  I would also like to continue to use this stage for testing since I don't have a replacement immediately available.

Reinstalling the stator is quite straightforward.  The only thing to be cautious about is hall sensor alignment since if they are too far out of position it will cause problems.

After the cables are reconnected and tucked into place the two halves of the stage can be reunited.  As far as I am aware there is no formal method for reassembling an anti-creep cross roller table in this manner, it just isn't done.

It might seem like the two halves should simply slide together just as easily as they slide apart, but in practice the procedure was quite challenging.

There are two main problems that must be overcome.  First, the stage is somewhat preloaded so pushing the cylinders back into their tracks requires some force.  Secondly, the anti-creep pinion must be perfectly aligned or you will lose travel.

Start by placing the cylinder carriers half-engaged with each side of the stage.  Its easier to see what you are doing with the stage placed on its side.

Next comes the trick.  The pinion gears which center the carrier actually mesh with a plastic ladder that can slide independently of the bearing.

Pull the plastic guides back so neither side is engaged with the pinion.

Then slide both guides back in at the same time.  This will ensure that the pinion engages both sides at the same time.  Once the pinion has correctly meshed with the plastic teeth the bearing will no longer lose its alignment.

Finally you can  reinstall the end stops and carefully push the two halves of the stage together.  It will take force to seat the cylinders, but the bearing should be smooth once assembled.

If it feels tight or starts to lock the chances are the pinion is not correctly aligned - take it apart and try again.

I finally applied some new oil to the bearings and stuck the stage back on the printer for testing.

After confirming that the stage worked, I re-aligned the printer, mounted the extruder, and reset machine zero so I could begin printing again.

Now for the real mystery - even though I made no repairs, the servo no longer produces vertical artifacts.  I suspect that my rebuild messed with the rotor/stator alignment which lowered its peak force in exchange for reduced cogging, but that is just speculation.

You will notice that the print still has a hint of waviness but I am very confident that this is a result of overly aggressive servo gains (or the lack of a layer cooling fan) and can be fixed by tweaking the printer's configuration.

This leaves me in a somewhat awkward position because I'm not entirely comfortable leaving this stage in the printer, but cannot justify the cost of replacing a "working" component.  In the end I decided to delay the decision and will consider purchasing a replacement stage once the printer is complete.

1 comment:

jbeck said...

Thanks for sharing!
I want to point out, that if you find replacement roller bearings (come up for cheap on ebay every now and then) you can extend the reach of the stage. Most of the ones you come across have 150mm of travel, but I have seen as long as 270mm. As you can see, the magnet track goes along the whole length of the top...



Email *

Message *