Extruder Gears out of Round, TPU/IR sensor poor performance
I've been *mostly* happy with the upgraded IR sensor watching the idler deflect mechanism.
But I have to have it adjusted JUST perfect, and it needs adjustment with each filament type, temperature, day.
A big part of the problem is the gears seem to be out of round, off center, something. While waiting for filament to enter the extruder, you can watch the IR flag move back and forth with each rotation of the gears, and this movement is large on the scale of the deflection when the filament is loaded, this noise makes the adjusting VERY critical.
I've used longer-thrown arm extensions to get more movement, and this helps detect soft filaments like TPU where the idler deflects minimally, but as the movement described before is about the same as the movement when filament is loaded, it's pretty hard to make reliable.
Does anyone else see this motion?
I've done some measurements - both on the door and on the idler gear itself. These are all done while loading shore 95a TPU (the cheap, hard stuff). Presumably it is worse with NinjaFlex.
| Min Postion | Max Pos | Min um | Max um | Range um | |
| Empty | -2 | 2.75 | -50.8 | 69.85 | 120.65 |
| Loading | 5.25 | 8.25 | 133.35 | 209.55 | 76.2 |
| Unloading | 3 | 6.75 | 76.2 | 171.45 | 95.25 |
| Loaded | 3 | 8.25 | 76.2 | 209.55 | 133.35 |
Min is the lowest value on the dial indicator as extruder turns (units 0.001") Max is the highest value.
Min/Max um are in microns (1/1000th mm)
Range is the difference between the last two numbers.
Loading is while filament has just been fed into extruder.
Unloading is while reversing the filament.
Range is the lowest 'unloading' and the highest 'loading'.
As you can see, the difference between the highest "empty" value and the lower "loaded" value is a mere 6 microns. It seems unreasonable to expect a sensor to see that a single bacteria sitting on the edge of the flag would put you over the line. You can use a long arm to exaggerate this difference, but the signal to noise is not good.
I believe that at least some of it is coming from the extruder gear and not the idler - although I see peaks for each tooth, I also see a deflection of 0.0045" on the smooth part of the driven gear. Note the peak is not at the screw (one might expect a low point here), but ~30 degrees around from it.
The bare motor shaft deflects less than 0.0002". The idler gear deflects 0.0015", so I guess the deflection of the door due to the gears could vary from 0.003 - 0.006" depending on the gear's relative phasing. That's more than the filament being loaded (~0.005").
Setting them at extremes, I get 0.007" deflection, 0.0025" if out of phase.
The remaining deflection happens in a very small area, perhaps dirt in gear teeth.
Note: Repeat to check: I did these by reading the back of the tension screw instead of on the door. Slightly longer distances expected.
| Min Postion | Max Pos | Min um | Max um | Range um | |
| Empty | 0 | 5.75 | 0 | 146.05 | 146.05 |
| Loading | 6.75 | 12 | 171.45 | 304.8 | 133.35 |
| Unloading | 3.25 | 7.25 | 82.55 | 184.15 | 101.6 |
| Loaded | 3.25 | 12 | 82.55 | 304.8 | 222.25 |
There is clear overlap here between loaded and empty.
TL;DR
Out of round Bondtech gears cause more movement of the IR flag than TPU causes by being loaded.
I maintain an informal list of San Diego, CA 3D printing enthusiasts. PM me for details. If you include a contact email and I can add you to the informal mailing list.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
Extruder gears "out of round" suggests one or both of the following build errors. They actually are machined round, but you can cause issues during build because set screw will impinge on idler door gear if it isn't truly against flat of motor shaft..
1. Loosen the Bondtech set screw and be VERY careful to make it engage at right angles to flat of motor shaft. If you have it not quite at 90 degrees, it will protrude too far out and impinge with other gear once per revolution.
2. Idler door axle is not engaging support pillows on both ends of the axle. This tilts the idler gear and once more you get the once/revolution set screw impingement.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
also check for debris caught in either the driven or idler bondtech drive gears. a little debris ( or burr ) at the bottom of one the teeth will cause a bump of the door
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
Extruder gears "out of round" suggests one or both of the following build errors. They actually are machined round, but you can cause issues during build because set screw will impinge on idler door gear if it isn't truly against flat of motor shaft..
1. Loosen the Bondtech set screw and be VERY careful to make it engage at right angles to flat of motor shaft. If you have it not quite at 90 degrees, it will protrude too far out and impinge with other gear once per revolution.
2. Idler door axle is not engaging support pillows on both ends of the axle. This tilts the idler gear and once more you get the once/revolution set screw impingement.
Out of round may be poorly trying to capture their not centering correctly on the shaft. Similarly, the inner and outer holes may not be concentric.
I've shortened the set screw, it does not stick proud of the surface.
I really put a lot of effort into getting the gear square to the shaft, and putting it on and off repeatably doesn't seem to alter the result.
I thoroughly cleaned the inside of the gear and the shaft - they were nearly clean beforehand, and cleaner after.
Between the gear teeth there was a fair amount of grease and dust, which I cleaned, thoroughly, and inspected under a microscope before taking the above measurements. I tried multiple spots along the gear to ensure I wasn't hitting surface imperfections.
In short: I don't believe this is coming from something other than defects in the gears. The MMU gears went out by the thousands and were replaced by PR as out of round/differently sized and defective, so suspecting the extruder gears to potentially have the same issue is not unreasonable.
Perhaps a better way to put it:
If you want to make a measurement of X, you would like to have noise less than 1/2 X, something I do not believe to be achievable with gears this far off. A flat idler gear, for instance, might help here as I would expect greater deflection.
Has anyone else measured this? I would like to know if my numbers are atypical.
I maintain an informal list of San Diego, CA 3D printing enthusiasts. PM me for details. If you include a contact email and I can add you to the informal mailing list.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
I've got 4 pairs of bondtech gears with the MMUv1 and another one with the MK2.5 upgrade. I've selected carefully the couple that I currently use. Some of them are a lot eccentric while others are not. This has little to do with the assembly and it can be repeatably shown by just replacing the good ones with a bad couple.
I've seen this behavior on a skele MMU extruder, where on a full circle rotation of the gears a "no filament" condition was detected at a specific sector of the rotation. The solution was to switch to the good couple of bondtech gears.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
I've got 4 pairs of bondtech gears with the MMUv1 and another one with the MK2.5 upgrade. I've selected carefully the couple that I currently use. Some of them are a lot eccentric while others are not. ....The solution was to switch to the good couple of bondtech gears.
Do you have any sense of HOW out of round they were?
I maintain an informal list of San Diego, CA 3D printing enthusiasts. PM me for details. If you include a contact email and I can add you to the informal mailing list.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
I haven't measure it, but it is enough to make the idler door swinging about 1/2 to 3/4 of a mm
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
3/4 is quite a bit! I think I was getting ~1/8 mm minimum net deflection at the gear (measuring the bottom of the door would ~double that).
Setting the gears unfavorably, and measuring deflection at the tensioning screw, I get 0.008" (0.2 mm) deflection.
Shore 95a TPU:
Empty: 0-6.5 mils 0-165 um
Loaded: 3 - 15 mils 76-381 um
PLA:
Empty: -0.5 - 4.8 mils (130um)
Loaded: 26 - 30.5 mils (660-775um)
There's a significant overlap with TPU where if I tell you (or sense with IR sensor) that the idler door has deflected between 76 and 165 um where you can't differentiate between being loaded and being unloaded.
The best thing would be to get the noise down. The next best would be to make the minimum deflection when loaded larger. I'm already running a weaker spring. Obviously the PLA numbers show the deflection is reasonable for rigid filaments, with a non-troubling SNR of 660/130 = 5:1. TPU has a ratio of 76/165 = 0.78 :1
With those being the constraints, I'm not sure anything can be done other than to set sensor so that it trips at a known motor location, then homing the motor on each start up, and taking your measurements based on being near a switch point. That would drop the noise floor by probably an order of magnitude.
Hint: Use M302 S0 to let extruder run without preheating. 🙂
So yeah.... Call this open to community input, cause I'm stuck.
I maintain an informal list of San Diego, CA 3D printing enthusiasts. PM me for details. If you include a contact email and I can add you to the informal mailing list.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
Just to complete the dataset:
Idler gear 0.002" (maybe 1.9?) out of roundness. Measured on filament extrusion teeth was still within 0.003" at worst, and the gear-coupling teeth had min and max values that varied by the same 0.002".
I was concerned the outer teeth might be off-center from the outer smooth diameter, but they seem fine.
I maintain an informal list of San Diego, CA 3D printing enthusiasts. PM me for details. If you include a contact email and I can add you to the informal mailing list.
RE: Extruder Gears out of Round, TPU/IR sensor poor performance
Quickly drawing this up, I have the gear diameter, tooth depth (approx), outer tooth diameter, and it doesn't leave a lot of meat for 'sensing'. In fact, if the teeth sink all the way in you're barely moving the idler at all.
This effect could be halved by using a flat idler, and this might be a good way to go with flexible filaments.
Numbers from:
0.6 mm diameter difference from flat spot to top of teeth on gear, another 0.8 mm (diameter) from top to bottom of teeth. That makes 1.4 mm that you'd lose off the theoretical 1.75 mm. We're starting to get to a signal that's 0.3 mm tops for something soft. My PLA measurements show 0.7 mm deflection, which says the teeth don't go too deep into harder filaments - and checking under microscope it looks like teeth went in ~0.1mm or less.
I maintain an informal list of San Diego, CA 3D printing enthusiasts. PM me for details. If you include a contact email and I can add you to the informal mailing list.