Это видео недоступно.
Сожалеем об этом.
300 Watt BLDC motors with ODrive project
HTML-код
- Опубликовано: 17 авг 2024
- Ordered 200watt, 8mm shaft from RobottDigg, but got the wrong version, and they sent me those big ones. I had to test them. And the learning experience was great, even though print quality is not. It will help me re-tune my ODrive 270kv motors better.
ODrive instructions: • ODrive 3D Printer inst...
Motors: m.robotdigg.co...
If you wish to support me: ko-fi.com/vez3d
Discord: / discord
Thank you Vez3D for your devotion in sharing the results of your hard work. Eventhough the outcome of the test is not the one expected, the data you share is very valuable and we thank you for that! oh, and by the way... clearpath, cough cough clearpath lol!
Yeah maybe one day when I have extra 💰
@@Vez3D ruclips.net/video/clgpZJFoqIQ/видео.html Contains everything you need to know
Vez is back In town gather round kids
@ 3:20 that's called a "key" . The slot it goes into is called the "keyway". This is used in engine crankshaft pulleys for example.
Yes the key goes in a keyway like a door goes in a door jamb genius
Yeah you definitely need to build a CNC now
Haha yeah.. I been playing with longmill its fun
Still hoping 🥺
Exactly what I was thinking when I saw those motors lmao. Those and a beefy spindle would be helpful for cutting metals!
Love those longer videos, pretty interesting to see how those chonkers behave
One of the best 3D printing sites on RUclips. We all learn a lot from your endeavors. Keep up the good work!
Those servos are way too big for the printer. There is a relation between the inertia of the rotor and the mass you want to move with it.
There's a reason manufacturers and good vendors state the rotor inertia on the product page for their servos and steppers.
When the servo is so op that it doesn't feel the weight of the gantry then there's not enough mechanical dampening and proper PID tuning becomes impossible.
Servo drivers are usually set to 300-350% for peak torque/current. So on acceleration those two servos can pull up to 2KW.
Your gantry weighs what, 350gr? They'd be happy little servos in a big router with a 10Kg gantry.
Get 140 or 180W JMCs with integrated drivers and your printer will fly.
I know lol... they shipped the wrong model... so i had to try them. Thats not what I ordered.
@@Vez3D u can build a house printer for concrete with those motors !
strap some bricks to ur gantry
Thanks for posting this. Always wondered what would happen with beefier BLDC :) and while this is a sample of one (or two!), it does make a lot of sense.
we cant really say bigger is worse.. because there are a lot of different motors out there, and maybe one will be perfect. I just cant test them all hehe
The wobble on the standing motors from digital camera stabilization is quite funny :D
Thanks for the learn it was very informative no time lost it told much more
Are you still using ODrive's step/dir mode? I really think you should try and switch to native (probably over UART) if so. Seems like you're hitting the limits of the ODrive's basic mode and need something with more resolution.
I thought ODrive was the way forward...
What happened?
@@eukaliptal ODrive has multiple modes of operation - step/dir is basically a way of emulating a stepper motor (which typically have a fixed number of steps). However since he is trying to maximize acceleration, step/dir resolution isn't really high enough (in a fraction of a second, think of how many steps it will have to move).
@@megamac125 I see.
If the goal is smooth motion, stepping seems like a fundamentaly bad way to do it right. So you say UART is the way to go?
@@eukaliptal well odrive is just a motion driver platform, like tmc2209, so it's more that odrive needs to be used in the proper way in order to get the most out of it. Step/dir is for compatibility, not for performance
@@megamac125 odrive and some klipper folks are working to implement it natively. Im not sure where it stands though.
You could also remove the key and flat side the shaft to save meat on the printed pulleys
You could make a filament rewinder with them too…thanks vez
I would take the standoffs down a bit try and save some threads just in case if room permits
Print the pulleys with the key in place!
“Motors? They’re too big” said no 3D printer builder ever 😂
I'm going to guess the resolution (at least in that mode) of the 3 phase system isn't as precise as that of a stepper and that's what's causing the wobbly sidewalls. It looks to me like an oscillation between 2 positions in the motor kinematics. The encoder may have a crazy-fine resolution, but the driver is going to fight with the magnetic/mechanical side to hold it to the precision we expect.
I could be completely out in left field doing the armchair thing too.
3 phase is actually better than 2 phase for both precision and torque at speed, thus handling dynamic loads better. 2 phase is better for low speed and holding torque.
so use 2 phase for z and 3 phase for corexy?
@tuffenough4u
Belive it or no 3 phase servos are way stronger than any stepper motor and more prisice
That is ac servo motor right?? It odrive work with step/dir?
😂 They look heavy, they feel heavy.... They *are* heavy!
You can use a 20T gear on those motors if you put the gear *above* the shaft instead of *around* the shaft...
True! Good point. But there is more grip with 40t.
Thanks for making all of these videos on 3D Printing with ODrive! I am going to watch several hours of your videos on repeat until I understand how to do it.
Do you know if there are any NEMA 17 style motors that could plug straight into an ender 3, but are BLDC and suitable for using with ODrive?
Probably are but dont know for sure
Never mind your idea seems good roll with it
Nice work, brushless CNC is the way forward ;-). Subscribed with the bell.
Ty. You got mine as well
@@Vez3DThank you.
I am not that sure... Depends on what you make but the only thing I see you doing with this is break bits. Unless you put a LOT of margin in your feeds and speeds which ... kinda defeats the purpose.
Looks like you need way more reduction from the motor to the motion system
3:15 Woodruff key
11:10 Flange
love the channel
Woodruff key works too same as keyway excellent
Yes rounded style so he is correct keyway
It’s a fringing key way a key comes with a key way there is not one without the other so it’s a keyway
@@cambouiscom listen here captain you have no clue who I am if you only knew first you misunderstood the plot this thing Vez was talking about was a part and is a keyway I don’t care what kind I never said woodruff I corrected to say my original keyway claim to help those calling it different watch whom your talking to be nice or I will clear your ice stop tolling for glory you never deserved
@@cambouiscom also calling it a key alone is not for people who don’t know this is yt remember
I got the odrive system myself for my own printer, and played around with it for several days now.. the board came with an old version of firmware not compatible with newer python odrivetool, so I had to hook up swIO and swCLK to my raspberry and flash the odrive board with latest dev version. then got anticogging to work, but had to adjust vel_gain and those settings to avoid motors to stall/jitter out. Then I looked at the code for odrive, and realized its really not calibrating at all, its just trying to adjust one setting doing one turn, thats it.. a calibration will have to adjust several settings like the vel_gain and those settings, so I figured Ill just implement my own anti-cogging algorithm in a fork of the latest odrive branch on github.
Wow man. You might want to contact them with your work. It might be interesting for both side
@@Vez3D I tried to dig in a little with odrive guys to see if they had something figured out for anti cogging algorithm, and Im not sure anymore if they do. I see no code for odrive v4, just a board with stm32f767 chip but no code. and they claim to completely rewrite the anti cogging algorithm, but since theres no code at this point Im not sure anymore. Which is why im gonna make my own algorithm. Basically the velocity gains needs to be specified in are from - to for the algorithm function to test jitter/stutter from the motor, and find the highest velocity gains settings for lowest cogging.
@@sveinjohansen6271 after speaking with ODrive, you are right about it. The anticogging in V3.6 is not a real anti-cogging. so If you ever code something better, please let me kknow :
What material are the motor mounts made from?
Nice job even though the motors did not work out. :)
Abs carbon fiber.
Isnt this because the encoder is only fixed with 1 screw? the encode move very little on this accelerations and might cause the ripple.... those motors can do much much more
I'm assuming English isn't your first language but you speak it very well. It is quite humorous when you can't find the right words though. Not a bad thing it adds to the entertainment value of your videos
You can remove key from shaft and print pulley with key not grove for key do you know what i mind ?
You mentioned you use clipper; are you using input shaping, and did you re-tune it for these motors? I can imagine the defaults are not very attuned to higher inertia higher torque motors but id imagine they could perform quite well if the controllers were attuned to them.
Input shaping only makes sense for open-loop controlled systems, i.e., steppers. It is basically just trying to learn how the system is going to vibrate, and calculate a way of blindly compensating for those vibrations. Servo motors actively compare the requested motor position vs. the actual motor position given by the encoder. If something like resonance causes the actual position to devite from the requested position, the closed-loop control system (PID) actively compensates for that. Adding that blind input shaping to a servo would probably not be beneficial, at least for fairly simple setups like this.
I do know that for more complex systems, input shaping and PID can be combined. An example of such a system is a crane which moves a mass using a cable of varying length as it lifts the mass up and lowers it down. Without any control the cable would swing like a pendulum, making it difficult to safely lift heavy objects. With a PID controller, you can tune out the swinging for one single cable lenght, but as you make the cable longer or shorter, the PID controller is not tuned well anymore. To fix this, you can make a simple mathematical model (input shaper) of a pendulum, whose frequency depends on the lenght of the cable. Then you control the movement of the crane using that model. Then you add a PID control (or some other closed-loop control) on top of that mathematical model to account for more random things like wind.
@@viljamip it's only a partially closed loop system; it's closed from motor driver to motor output shaft; not from driver to nozzle position. A stepper is passively closed in the same way if you will, as long as it's not losing steps; and input shaping should work the same and be equally relevant to both.
That’s also true. So this is a somewhat more complex system after all which probably justifies input shaping.
Instead of grinding those down I would instead either drill a few holes in the plastic bracket, or print a new bracket to hold it and actually use those holes.
Alternatively, extend the shaft by putting a bus around it
Uh you won't be able to get the tolerances right on a printed bracket to make it clip into the encoder. The brackets that come with it barely do.
@@forbiddenera Why's that? It doesn't appear that difficult. But alternatively he could use the threads on the printed bracket to clamp the encoder as needed after aligning it.
I have all the same stuff.. odrive motors odrive 3.6 and those exact encoders. I struggled for about 2 hours to snap them on, out of three encoders only one snapped on easily, one broke and the other took a lot of effort. The clip is very thin, like 0.2mm or something..one layer isn't gonna work heh
is there a reason why you are using physical touch encoders as opposed to something that is no contact such as an AS5048B. that's what we use on our brushless motors.
no particular reason. That is just what I got from ODrive, and what they are selling. The model you reference is a hall sensor I think. wont fit ODrive... I think
@@Vez3D they are full encoders not hall sensors. its what we use on vescs for custom motor applications. on the Odrive site in the encoder guide, it says you can use them so it might be worth a shot.
while the noise is slightly higher, the resolution is double, which generally results in less cogging and such.
I have herd this before Vez he is correct this is what I call a real servo instead of assisted type
That’s nurreled on it will come off but not be a tight as once was as first installed
3:19 Le chemin clé, is a Keyway in English. lol :P
Loctite!
One thing that bugs me. It might be due to something about the ODrive that I don’t know but, why didn’t you just use the Hall effect encoder on the motor? If it’s the same type (I.e. both quadrature or both absolute) then I don’t see why it shouldn’t work. Somehow input the pulse count and you are done. Please correct me if I’m wrong but that just seemed odd to me.
I honestly dont know. I was under the impression that it was not possible. But im way out of my knowledge range here. Hehe. But if it works with hall sensor and its precise then its awesome
@@Vez3D Honestly I've never used ODrive, so don't know for a fact but what I do know is, as long as it is a quadrature or ablsolute encoder output (i.e. VCC, GND, A, B) then I don't see any reason why it wouldn't work considering that the encoder you use is of the same output type. I personally have experience in building robots for competition and I haven't seen any issues with one incremental encoder over the other as far as the output signal goes. I would like to suggest that you try it, but then again that would take time and I guess a little resource so I don't know if you'd want to try it. No real risk of harm to your equipment if you do tough, unless you plug power into signal but that's just common sense :P Any how, enough rambling on my part. Love your work, keep it up and if you try those old encoders the motor came with, please make a video, I'd like to see it XD
@@berkakarcaloglu1915 i have broken the little center part.. ferrite? Anyway..its gone so wont be able to use them. Unless I buy some
Vez3D sad. İ'm pretty sure that someone somewhere is selling one but considering everything I'm not sure if it'd be worth it today you. You know, since you have a working setup right now 😁, don't wanna mesaj it up😂😅. Me joking aside I guess now you know a little more about encoders and hopefully you might have less work next time bu using the pre-existing encoders. P.S. if you still want to try and find those magnets Ford the Hall effect encoders, I'd love to help, but agin don't think it's worth it. Also fun fact those encoders you ended up using are also Hall effect encoders if my memory serves me right. 😜
why not design the "key" to the inner diameter of the belt gear?
it would be too deep in the plastic... and be there were the teeth are. With a 60t it would be ok
@@Vez3D I mean design the key part as part of the gear rather than a slot. The plastic would therefore be thicker at that point, not thinner.
@@jneilliii aahhhh yes very good idea. I dont know why I didnt think about it. Though I would still need some kind of m3 bolts to secure it, but that with the slot would be perfect.
My thoughts exactly! Also wondering why not design a new mounting plate for the encoder with insets to accommodate for the screw studs, would give a much better mount and no need to grind down the studs.
@@BlackHeartScyther shaft is not long enough i think. Right now its limit.
Have you tried the Clearpath motors?
No but I wish. My buddy MirageC has them and he loves them
did you try to use the Arc Welder reduce G-Code ?
The motors look like they are jiggling on the table. Some weird picture stabilization going on 😂
Welcome to RUclips
That guy is called a shaft key.
It’s a key way
keyway but I'm sure anyone would understand if you called it a shaft key nonetheless.
@@patprop74 unless your a millwright or a mechanic or a teacher or a rocket scientist allll day it’s a keyway but ya I get ya I was only trying to help your terminology my bad call it what you please shit
I didn't know you can use brushless motors for a 3d printer
THANK YOU
Ic you wrapped up the magnum is that Teflon tape? Or did you make a sock
I'm curious too. Looks like RTV silicone to me but its way too clean. Might be worth doing a short video on that!
@@etle113 Looks a bit like silicone hose that you'd find in a car intake/radiator etc
Bog maybe onto something or tis rtv
Vezzzzz we know you know we know you can hear us lol
Those are threaded standoffs
What power supply are you using?
380watt 48v.
Thought you said you were nowhere near maxing the odrive motors lol
Heck no! Odrive are powerful motors. I was trying to see if there was a better option to reduce cogging at low speed. But odrive motors performs way better than these... and I will retune them with what i have learned with this project. Im pretty sure I will see improvement
3:50 if the key is too deep into the pulley, why don't do the opposite and model the key on the pulley.
Any chance that the fact that the top of the Gt pulley shafts isn’t braced is introducing flex into the entire system? It looks like the shafts are only held by a bearing(s) at the bottom? Isn’t corexy more sensitive to this kind of thing?
its braced it he back. you just dont see it. is that what you mean ?
I think I saw a NEMA 24 motor once and though "how peculiar" 😂
I know this is old, but i guess you didn't know odrive can interface with those halls aswell
yeah I knew. But its not good enough for precise movement I think
@@Vez3D idk how much resolution your encoders have, but ODrive can read each hall sensor at up to 4096 different values, and i'm sure it's combined together so idk how accurate that would exactly be, you'd probly lose 1000 steps maximum because hall sensors have a min and max voltage output, but it should still be very accurate, and would allow the motor to stay original i guess
edit: the encoder would have 2048, so in theory hall can be even more accurate BUT im not sure if it would improve anything or make it worse
Could you not just use the hall sensors instead of ripping it apart? Would they not work with the current electronic set up? Those motors are massive! Also on the spindle those are keyslots and keys.
This was more than a year ago. Hehe. The hall sensor is not precise enough i think and not working for odrive
@@Vez3D ah okay, I’ve never messed with the hall sensor, I know the encoders you use have a fairly high resolution. Still a fun video! Besides the fact that you got the wrong product lol.
It's called a keyed shift some keyed shafts can be knocked out could be helpful
Havw you printed in an enclosure? Did you have issues with the differences in expansion rates of steel and carbon?
My vzbot is all enclosed and no issue. But i doesnt heat more than 40-45c
@@Vez3D hmm I do take mine to 60c. I heard from e3d that theirs bent like a banana (maybe exaggeration) but they also used a flat CNCee Peice not a tube
Where are those encoders from?
How are 3 printers not CNC machines?
Did you thought about trying JMC Servo's?
I did. One person in my discord server has tried them. I think he likes them
have u not tested any anticogging algorythm for odrive? it would reduce the cogging torque in those odrive bldcs, idk if that would solve your problem, cause low cogging servos/bldcs tend to be too expensive
Yup it was tuned with anticogging
@@Vez3D have u ever considered the idea of using 3 phase linear permanent magnet motors in our printer? Idk if that solution would be too expensive, but in therms of performance, linear motors are the best, they're fast and precise, and there are low cogging version available on the market. I think for a high speed 3d printer it would fit perfect
@@leonardoaugusto5491 yes i have. Linear is another thing on my list to try
@@Vez3D that's great, can't wait to see it
Please use ODrive and those motors in some other invent! It was very interesting, not boring!
Yeah I have a plan when for next summer on my kid electric car ;)
They are called a keyway
30$ refund? How much were they to begin with?
the motors are listed for 58$ each on robotdigg, the ones he intended for 48$
"woodruff key"
Given that these motors can operate at much higher speeds, would changing the gear ratio have an impact on quality?
I tried with 2:3 ratio and with 500mms..and it didnt help
try like a 40:120 or something dude
@@v1Broadcaster I tried 1:3 and that was too much. with high accel the belts were just slipping on the gears even with high tension
@@Vez3D apologies if it’s already been suggested but might a planetary gearbox be a good addition with belts on the output of the gearbox? Perhaps that doesn’t get around the belt skipping issue though
@@connerknox5645 that would fix the belt skipping issue...but might transfer the problem to gears getting beaten.
Interesting so you conclude that other than original odrive having cog-gin issue speed wise the o drive superior whole stepper motor are balanced.
Well the 270kv from ODrive were performing better.. at least with what I found. Im not the tuning expert..so thats just based on my experience
@@Vez3D I understand, I’m learning too so I really appreciate you videos experimenting with these motors.
Such a high powered motor cannot be driven by a belt. The force is greater than the tensile strength of the rubber. You must use lubricated metal gears only.
I probably use 10% of the available power haha
So not true - we drive 40kw ac servos with belts that move 100kg loads. speeds around 6-8m/s. And they are precise as fck...entering plastic molds for example. But belts are with metal reinforcements.
@@wkbrd
Hold up! 40kw??? Good Lord!
Why did you move away from the ODrive motors?
Im switching back to them. But wanted to see if I could find better ones. Odrive ones are good but i have a bit of cogging issue at slow speed. But what I have learned in doing this new motor project will certainly help me tune the odrive motors better
@@Vez3D Are you having cogging issues while the odrive anti cogging is enabled? I heard pretty good things about it previously.
@@eelcohoogendoorn8044 yes its on but at low speed there are still bit. It doeant need much to chause wall artifacts on a straight wall on a 3d printer
It was funny to see you remove a hall effect encoder and replace with a optical encoder. I would have thought that encoder should have been able to output usable position data. I think no matter what motor you go with you need to gear down the motor to reduce this cogging artifact with odrive.
You are just dumping money into a system where the RND should have been done by ODRIVE them selves and there should be a known solution to this problem. It would require a very spefic brushless motor with way more poles like a steppermotor having 200 to 400. This is why you need higher gear ratios to smooth out the movement and microstepping.
I dont know if that hall sensor would be odrive ready? No clue. As for the rnd..im happy to do it. ODrive will also do some rnd on their side as well I think.
@@Vez3D yeah the unknown out put of the hall effect probbaly isn't worth the 1 to 4hrs trouble shooting expecially if your encoder happened to fit on like it did in the end.
I would have went with the tapped m3 hole option as this wouldn't affect returning the motors to a stock configuration.
I doubt there would be a easy fit option with that shaft size for any planetary geared option. Then you are also reducing the top speed of the motor. Most brushless motors can spin very fast not so much of a issue like a steppermotor. That would limit your top speed using a large reduction ratio. 60 lt to 40t is only 1:1.5 ratio i would suspect you need much higher before seeing a knoticable difference deppening on the motors magnetic poles and per step degree rotation.
@@Side85Winder 40 to 60t. And I tried it as well with no significant improvement. I also went up to 500mms and no improvement in this case. While odrive motors, when you go fast it fixes cogging
Odrive has had tons of rnd and there's maaany different ways to use the controller, don't blame them if one method doesn't work perfectly.
@@forbiddenera i don't blame them for anything. It just seems its not ideal setup for a 3d printer. You deffently can get it to work on a 3d printer but its not a good use case for its setup it seems. I am sure in future there will (are) be better motors for odrive system to use for a 3d printer that will produce way less 'cogging' it seems a lot of work to figure it out and $$$ (hence the RND).
They seem way more ideal for pick and place machines. In this case the system can get to point A precisly and it can get to point B precisly. The movement inbertween can have 'cogging', it doesn't matter and it can make that move a hell of a lot faster than other setups.
Find Marco Reps on youtube, see why Odrive is not good for cnc control.
Now I'm seasick. Your image stabilizer is pure horror. The motors looks like gummy bears. But nice project as always...
I assume it's premier/AE post process stabilization. It works but it makes everything look like it's made of pudding
do you really need such big motor for a light corexy axis?
No haha. Unless you wanna do crazy things
We do……. Lol soon will need a dampening table too
Actually smaller brushless hi pole motors maybe worth a try you will also kind 5 mm shafts
👍
1.8 KG!!! OMG
Why not to use the hall sensor? It can also work with odrive 🙂
You sure? If so then..wow ..all that trouble haha. But do you know if its as precise as my 8192cpr encoder?
I thought so but was not sure enough just sayin
Yes i'm sure for my hall encoder that i bought on ali. I tested it yesterday and it works good. But as i can see on your video, you're encoder seems to be compatible to odrive.It have the 5 correct pin (5v,ground, A, B, C : index). It's an incremental encoder with index signal. The encoder resolution must to be verifiy on the constructor website (I'm sure that we can find this same encoder on ali). For mine, the resolution was double compare to the AMT 102v that i use on other mechanism....
@@djwillx974 wow ok.. thanks man..good to know
Threadlock
But did you tuned the rigidity of motors? Common servos are much better than odrive at any speeds when they tuned.
Im sorry ..i dont know what rigidity tune is. I have tuned them to the best of my knowledge..but im no expert here.
@@Vez3D by defaults many servos have low rigidity set up that means the "rubber effect" between exact position and desirable position. It makes problems at accelerations, corners etc.
How manny Belts you need per Print??? hehehe
Such a shame after all that hard work.
Why not just use the Hal sensor?????
It doesn't work in step for mode for odrive
@@Vez3D ok... maybe I am missing something or as usual I came late to the party...but why would you want to use odrive for 3D printing when you want repeatability.
@Ldah Hall becaue not so long ago steppers and drivers were not able to achieve high speed high accel. But now its a different story :)
@@Vez3D ohh...I did come to the party late then...ok...well what's the latest with those beefy envious steppers you have...have you switched back to Hal and idk maybe upgrade a board to get faster speeds. ?? Thank you for the explanation BTW...I appreciate it.
lets just skip all the possibilities and go straight to linear motors. they are pricey but godspeed
for example a PI v857 does up to 5000mm/s at 50000mm²/s at 400mm travel length and built in encoder but no words on affordability (lol probably not)
That would be crazy cool! One potential problem will be the weight of the linear motor for the y axis.
You need a reduction with these kind of motors(1/20 to 1/100) and they are mostly used with a matching driver from the same brand.
I don't know what kind of controller you're using for this thing, but those motors shouldn't have step discrepancies set to 3600-5000 microsteps, especially on a 2:1 reduction.
Assuming you're using a GRBL controller or something of the sort, that is your problem. A true serial comm 16bit DSP is required with real S-Curve algorithm capability if you want those to deal with dynamic load accelerations reliably. I am personally involved with Trocen Automation with the 78 touch screen series controllers for lasers. It uses a 32 bit third gen Texas Instruments DSP with parallel comm that far surpases most other Chinese controllers with smoothing accelerations. Open source has yet to catch up with DSP systems like this, but hopefully soon.
It's using ODrive
@@Vez3D Ah, I wasn't aware they made a motion controller, I just thought they made motor drivers.
@@Tuffenough4u the motion controller is either a BTT 1.4 SKR board or a spider board running STEP/DIR towards the Odrive board which powers the BLDC motors.
so they are nema 24? another common standard.
not realy.. more like nema23 size.. but just a tiny bit bigger
@@Vez3D i love your work, i am new to FFF and tinker and test everything like you! Just wanted to say thanks for sharing all the good and bad
Not Nema get what I give u that’s what servos
@@CC-kc5lb its the same motors you suggested, exept the 300watt instead of 200. They are not nema
@@Vez3D holy cow the stratasys ones are not like that they are smaller but the servo idea was great I had no idea I had influence on this I now feel bad it never worked lol
you need T5 pulleys and T5 belts.
i cant watch this with that wobbly effect that you added for some reason...
$30 refund for sending the entirely wrong product, that says it all. Wouldn't do further business with them.
I ordered from Robotdigg a few times in the past, never had a problem with them, however, they seem to only be responsive if you ask about your order on Skype, messaging them from their website is useless.
No nema 23 fit they should have sent them back these company’s man…
:(
Your image stabilization is atrocious.