Thank you! The motor you specified has a longer body length (60mm vs 48mm), which wouldn't have been great for comparison in this video.

@@engineerbo will there be a Test with 60mm Body lenght? Would be excited to see it. Really liked your test setup. 👍

I have lots of different tests planned, but not one for 60mm motors (I don't have any!). But if there's a large enough demand for this test, I can arrange for it.

All these motors are hybrid stepper motors. Are you referring to something else?

@@engineerbo Hmm, I know them just as 'hybrid steppers'. They are just normal 4-wire steppers, but they are in round aluminum cans, I got a box of them in an auction years ago (they were buried in a pallet), and I've used them for various projects over the years. The first picture for google images "EHY21512" or "23HY4604" shows the TYPE of stepper I'm talking about. The cases are round aluminum cans, not laminated steel sheets

It was the first build of the dyno, the breadboard certainly isn't the best. I'll be building a PCB to replace it for the next video. I needed custom functionality for the dyno that perhaps something like an SKR could do, but it seemed more straightforward to implement it myself.

Thank you! That's an interesting suggestion. How did you test for resonance in your build?

@@engineerbo I think you could just add simple ADXL sensor to your rp2040 setup, place sensor on the motor itself and run loop changing motor frequency at the same time recording readings from the ADXL, at least I would start from there. On 3Dprinter setup it's much easier as Klipper helps to run such task semi-automatically.

My first guess would be that every motor has its own resonance frequency, and there wouldn't be "more" or "less" resonance, so I'm not sure how I'd design a test for this.

@@engineerbo I would assume a well-made and balanced motor should have fewer frequencies it resonates at. At the same time, I think the higher the vibration level and the wider the spectrum, the harder it is for the algorithm to mitigate those vibrations, and it will end up affecting the print quality. I'd probably make two tests: the first one to see the overall noise and the second to check how high the peaks are getting.

Thank you! I've started planning to test for position accuracy, so there'll be a video of that in the future.

Thank you! There've been several comments talking about positional accuracy, so I'll look into make a video on this after my speed and torque series.

Thank you! The brake calliper is mounted onto an arm that pivots about the same axis as the disc, and the other end of the arm is attached to the load cell. This makes the measured force "downwards".

@@engineerbo Aah! That makes sense. Fantastic, thank you!

Yes there are three capacitors between VM and GND. The caps are rated to 50V though, so they have a better than surviving than the TMC2209 chip.

Yes! Fake drives are incredibly prevalent in some areas. The sellers have probably decided that non-savvy buyers would purchase a 2TB drive, and end up only using a couple GB.

Thank you! The bracket design is not ideal (I've designed it to make it easier to switch motors out for testing), but I don't think the torque readings will necessarily be unreliable. If the motor is mounted on a flexible, springy bracket, the bracket will flex/deflect depending on the torque applied by the motor. The higher the torque, the greater the deflection. Yet the torque deflecting the bracket is still going to be equal the torque applied to the brake. This is going to be the same for a stiffer material. In reality, there might be a slight power loss by the motor flexing and "de-flexing" the bracket, but I think it's not a major problem.

You're right, the breadboard is rated for 2A, so there would have been some power loss there. That would be fixed for the next video!

Can't beat unlimited bang per buck.

The gate is pulled up by R7, so it only needs to be left floating for the MOSFET to be switched off. The problem is nHDR, which is connected to the MCU, cannot be pulled up to VPP. The MCU's GPIO protection diodes clamp the pin voltage to between GND and VCC (with a margin of error due to the forward voltage of the diodes). That means nHDR is clamped to ~VCC (and VCC < VPP), stopping the MOSFET from being switched off.

Thank you for the feedback. Could you please elaborate on what you mean by "the twist will not be linear"? I'm guessing there's little "work done" when the motor wiggles in the mount, but yes, it's not perfect. When you say my measurements are affected by the BEMF, do you mean the supply voltage is affected i.e. not perfectly stable at 12V or 24V? For measuring phase currents, I'm reluctant to add additional current shunts. I'm considered this, and may use a hall effect current sensor, though I'd still need to calculate losses. Overall, it might be better to keep things simple. Regarding wiring, yes there're losses involved. The breadboard is only rated to 2A, and the voltage drops could be significant at e.g. 2.5A. These are some things I'll be fixing before I move on to the TMC5160.

@@engineerbo For the mount, there are two things going on, fist is the natural hysteresis as the entire drivetrain comes under tension. every part in the drive train will have a different rate, the mount, the belt, the wheel, etc. These all sum together. Second is the way the mount itself winds up as it comes under load. First it compresses around the screws, then the plastic compresses, then it twists 90 degrees to the axis of the motor. once the rest of the drive train overcomes all the sticktion in the system, the mount will actually unwind, and then oscillate (over damped system with non-constant acceleration). All this happens at the beginning of the test where torque is at it's highest. yes, the back EMF as one winding is released is helping to power the next winding that is being powered. Each motor or driver should have a capacitor of a specific type and rating to absorb and drain the BAMF. The other issue is that BEMF can cause real havoc to the control board unless properly managed. Bigtree stuff seems to be pretty solid in general, but I would still take the time to make sure it is not a problem. The nice thing about shunts is that you measure voltage and calculate amperage, which is something that the RPi is happy to do. On the other hand, they need to be accurately characterized if you want true values. There are probably cheap inductive meters for the RPi available. Last time I dealt with it, I ended up just using a shunt tho since I did not trust any of the cheap boards I could find. It is a common enough problem that people want easy solutions to. I have both cheap and very "high quality" bread boards, and have found they all are pretty bad at gripping anything after they have been used. IC boards can be really bad about stretching them out. I think your overall approach is fairly solid.

Thanks for elaborating. The dyno only measures the speed and torque that gets applied to the shaft, and the brake also only affects the shaft, so if the motor is shifting about in its mount, the torque applied by the motor to the dyno would be lower. For example, of the motor is not mounted at all, the body of the motor will just be spinning freely, and it's not actually applying any torque to anything (except the air), and the dyno's measurements should reflect this. The speed will also read as zero, which wouldn't match the step frequency, so we'll know there's a problem.

Yes the TMC2299 can be configured to have different running and holding torques, and that's supported in Klipper too.

I'd be more than happy to test the fancy stuff if I could afford them!

Yes you're right, the user's needs are briefly mentioned in the preliminary conclusions. I just couldn't fit it into the title.

Thanks for subscribing! To measure holding torque, the motor is usually made to hold it's position, and increasing torque is applied to it (from an external source, e.g. weights) until it position holding fails. With my dynamometer, the torque comes from the stepper motor itself, so strictly speaking, it cannot be at zero speed and increasing the torque at the same time.

You might be right about the stall torque. The reason I've added belts to the dynamometer is so it has a little compliance, which makes it possible to measure the torque when the motor is holding its position. But I don't think it'll actually be able to reach the actual holding torque, because the motor still needs to move, even if just a little bit, to move through its torque range.

@@engineerbo Ah. I thought you meant stall torque not hold torque. Yeah, hold torque is trickier, especially if you want it to test in the same setup. One approach that can work is to replace the brake with a larger motor with a decent controller. In practice, especially if you have a bit of gear/pulley ratio between the two. Still do the actual measurement with the load cell, but replace the gradual application of brake with the gradual application of PWM (instead of open-circuit). You might want a braking resistor, depending.

Yes using an external source of torque would solve the problem. But I'm not sure the actual holding torque is that critical to know, since an extremely slow speed gets me a number somewhat close enough.

@@engineerbo Agreed, at least for this application. Pull-in torque can be substantially different than pull-out torque even at zero speed, but I don't think you care in this case. One other thing that might be nice to measure - although is somewhat terrible to measure to be fair - is the stepper motor resonant frequency and behavior around said resonant frequency. Note this depends on the inertia of your load! Doing a slide test at (near) your resonant frequency is somewhat of an interesting worst case. (Essentially: a stepper motor 'snaps' to the next pole, especially when not microstepping. But in practice it'll oscillate around said next pole somewhat before settling down. Interesting things can happen when you do another step at the quarter/half/three-quarter/whole period of this oscillation.)

Thank you! The tests will be done in the near future!

Thank you!

Glad you enjoyed it!

Thank you!

Some people say it's perfectly fine to "overclock" their motors and I totally believe it, but it's also difficult to test whether the motors end up being unreliable.

It'd be amazing if they did!

Thank you! I've got Moons' on my to-do list now.

I looked it up, and the drylin E stepper motors don't seem to have that much higher rated torques and are pretty pricey. Are these the same ones you're referring to?

@@engineerbo depends. they have many similar products, including very pricey "industry grade" stuff. I'll try to post a link, hopefully YT will let me do it.

​@@engineerbo​ okay apparently that didn't work. You gotta try to sort by price on their website, and just pick the cheapest one. Also to clarify: The rated holding torque isn't higher, but at least according to the first party graph they retain much more torque at higher speeds. (0.3 NM at 700 RPM / 24V)

Ok thanks, I'll have a look!

Power is torque × angular velocity, so actually the power curve would give you the same information, just in a different form. If you use mechanical advantage to increase torque, the tradeoff is speed. If you assume a perfectly efficient system, the speed-torque curve's axes are simply scaled up/down, but will keep its shape.

I'll add this to my todo list, thank you!

Thank you!

Yes I too interpreted it as step frequency, which is the same as angular speed assuming the motor isn't stalled. E.g. for this motor, 200 (full) steps per minute is 1 RPM.

Interesting, what is the design flaw? A couple other comments mention the lack of capacitance (presumably near the driver), which probably means voltage spikes killed the drivers.

@@engineerbo On page 16 of the TMC5160 datasheet there's a section specifically saying that you should not provide more than 40V on the VSA pin, it's the pin that supplies voltage for the internal 5V and 11.5V regulators. On common stepstick designs VSA is tied to VMOT, which will supply whatever voltage you are giving to the motors. So using them with 48V is basically silicon lottery. That's the flaw the standalone 5160 boards solves. Don't get me wrong, the capacitor is also needed, a good rule of thumb is 100µF for every amp of current that's given to the motor. For these ones you tested a 330µF electrolytic capacitor of 50V or more should be enough for all test cases.

Thanks for the heads up! It'll be pretty annoying if I had to modify the boards to test at 48V.

@vinnycordeiro I just found some time to have a look at the TMC5160 datasheet. On page 16, the note I think you're referring to doesn't actually say not to connect VSA to >40V. Instead, it's simply recommending VSA < 40V if certain conditions are met e.g. MOSFET total gate charge > 50nC. This note is related to the text preceding it (Chapter 3.2), which talks about high power dissipation of the internal linear voltage regulators at high input voltages. So the designs by BTT etc are actually fine, at least in this regard.

@@engineerbo I oversimplified my answer but yes, you are right. It's just easier to stay under 40V than having to double check the MOSFET choice of manufacturers. BTT did that on their TMC5160T Plus V1.0 board, connecting VSA pin to 12V directly instead of VMOT.