We say the same again, I think. 2 mm wavelength, Spatial frequency k is = 0.5 mm^-1.
Then the ringing spatial frequency vs speed is k = f / v. It seems silly, but this is useful later graphing the resonance versus speed.
Ah, now I am with you. You think in frequency, I tend to always think in wavelength because I can measure the latter 
1 Like
Yes! My brain has both after measuring resonances in Hz but also mm but also speed and… Yeah. 
1 Like
I found the broken stepper (from the x axis), took it apart and compared it with an older stepper that I still had lying around.
While the basic design is probably the same for all chinese NEMA 17 steppers, the Mocotech seems quite cheaply made in comparison to the other one:
- wires on stator coils are wrapped untidily
- rotor has less but thicker electric sheets (unsure if I translated that correctly - I mean that thin iron rotors of electric motors are made of…)
- rotor is not potted at all
- magnet glued in place inside rotor is about a millimeter off-centre
- rotor notches are only overturned very roughly and full of burrs, if you rotate it outside of the stator you even see the spiral of the tool cut moving.
All in all, numerous chances to get some kind of repetitive error.
1 Like
Maybe. The Moons motors I ordered came quite recommended on the Prusa forums. Granted they’re 0.9 deg, but I didn’t see a big difference.
Apologies for not posting the data yet. The holidays are upon us, and some attention needed by the family besides. I have not forgotten though!
No problem at all - same here X-D
When it comes to the steppers: since
- you excluded stuff behind the steppers by swapping the gear
- I excluded stuff in front of the steppers by replacing my whole controller including the stepper drivers to another model
- and the only thing that at least partially fits to the 0.8mm ripple wavelength I have on my parts is the stepper fullstep distance multiplied by 4, while I do not have any 2mm ripples or muliples of that
- and the Y stepper data lying around somewhere in this forum also does says the Y stepper has a rather high inductance which the Duet team says may cause issues
I actually do put some hope into the thought that replacing the Y stepper might do some good, even if it means some additional work since the Y stepper is anything but standard with its max. 40mm body length, 8mm dual shaft and 197mm long primary side.
Luckily, I had bought a full set of replacement steppers when the X axis stepper failed, so I do have a stepper that I can play around with. The only question is : which dual shaft stepper should I try?
Edit: since there are not too many biaxial steppers out there which fit and are easily available, I ordered an OMC 17HS16-2004D now. Let’s see how that one compares to the original one - the torque curve calculator I have says it should be a bit better for the roughly 2kg of Y gantry mass the J1 has, and the OMCs had proven to be nicely silent on my old printer
Yes! The x stepper is not actually so bad it’s an obvious problem. The artifacts from the Y motion are much more noticeable to me.
I actually have most of the hardware I need to modify the Y axis with 0.9 steppers. Special couplers and all.
Just, after seeing no dramatic improvement in the x data, I decided to stop and think about the bigger undertaking. I did most testing on x simply because it was easier, aside from firmware tweaks for microstepping and spread cycle hybrid mode.
I’m eager to hear your results. I found biaxial stepper and ordered a stiff coupler and some additional shaft to extend and cut as needed.
Not tried yet though!
EDIT to add, my J1s is working normally again at least! Replaced the main board a week ago. I’m still laughing that I did all this work on my printer, then restored it, THEN it breaks by my clumsy hands.
I went for two (radially rather stiff) flexible shaft couplers - my stomach feeling that a bit of angular flex might not harm, the Y drive of the J1 is mechanically overdefined. Let’s see what that does 
Those flex couplings can have a lot of give. May not be very precise. And possible the two sides don’t stay in sync. If they’re the spiral type.
But on second thought… One thing I was also going to try was adding vibration dampener to the motor mount. Granted it’s stiffer than I expected, but I wonder if flex couplers would achieve similar.
You are right, the cheap spiral type is not that good and might have not enough radial stiffness as well for the J1. I went for one with a… somewhat zick-zaggy cut (don’t know the correct english name for those, it is the GMT FACMS20-5-8) instead.
You will need to remove the stiffening rib next to the stepper to do that, which is why I dropped that idea for now.
Okay, new stepper is in and TMC settings are updated. Draws more current, sounds very slightly less noisy, has more power and slightly reduces the artifact depth on the “worst-case edge” of my test part - but it does not have a really significant impact.
Since the original J1 board already seems to have heat issues with the TMCs (otherwise Snapmaker would not glue heat sinks on them even if TMC says that cooling their stepper drivers must be done from the PCB and adding sinks onto the top of the chips not make sense), I would vote for not switching the stepper on an original J1, even if I will leave mine in the printer now since it will make my life easier if I ever want to replace that one.
I noted however that if I put my finger on the stepper, I do not feel any vibrations any more, while I well feel them if I put my finger onto the side of the top frame of the printer, directly above the place where the linear rod is screwed on.
For what it’s worth (although you’ve completely disassembled yours also), Snapmaker also sinks their drivers from below. There’s a thermal pad under the board, and below it the aluminum chassis has a polished extrusion that rises up to turn the aluminum body into a big heat sink for the drivers.
Pretty clever touch actually.
Hm… the basic idea behind that is indeed clever, but the way it is done is a typical case of “…did not think this through…” :
- first of all, they did not seem to get their manufacturing tolerances right, requiring them to use a much too thick thermal pad (the thicker those things get, the worse their heat conductivity).
- second, they did not take the flexibility of the PCB into account. The PCB has not enough screws and not in the right places to reliably make sure all the TMC’s are pressed against their heat sink reliably.
- And third, what is the sense of adding a passive heat sink to the stepper drivers if you add a noisy fan for the power supply and an even noisier fan to get the hot air out of the printer compartment? Not to speak of the absence of any kind of controlled air flow down there.
(Additionally, my stomach feeling says they probably made the aluminium down there too thin for a decent heat sink, and the complete lack of air flow due to warm air getting caught in the dent in the printer bottom probably does not help either. But that is something I did not check in detail, therefore it is not in the above list.)
My solution when replacing the whole controller was to throw out the noisy small fan in the back, add a larger but silent (and really temperature controlled - not only on-off) radial fan blowing outwards right next to the controller board and a printed air guide right below the PCB that makes sure all the air drawn by the fan flows along the underside of the PCB where it is needed. Result: same amount of fans, less noise, and the need for both the “case heat sink” and the added heat sinks on top of the steppers is eliminated.
Hmm. On point 1, I did not think it was too thick when I changed my board. It actually took me a second to recognize what it was. Since my replacement board didn’t include new padding I peeled the old one and reused it, but tore it twice trying due to its thinness. Fortunately it was sculptable enough I was able to shape and close the small gaps just fine.
I do want to replace that little chassis fan though. It’s hard to believe that little thing moves much air for all it’s noise. But also, there’s quite some thermal mass to the air and the metal chassis, so I don’t worry about this much.
I have certainly seen at least one, far worse setup for driver thermals that never had trouble. (A Klipperized Ender)
Hm, the one on mine was at least 2mm thick. It still tore, but definitely not due to its thickness but rather because it stuck like hell to the PCB.
Those things are/were added to some Raspi 3 aluminium cases that use the case for passive cooling. And they usually take great care to use max. 1mm (better 0.5mm) thick thermal pads since doubling the thickness only lets half the thermal energy pass through. That together with the relatively bad thermal conductivity these pads have (around 6…12W/mK compared to 400W/mK for copper) and the small size of the TMCs results in a rather bad heat transfer.