Can I use the SimpleFOC library to drive syncronized axes in closed-loop?

I’m sorry to hear that, been there too. :confused:

I really like it’s front-page :slight_smile: GitHub - cbruiz/printhor: A 3dprinter/cnc firmware framework powered by rust embassy

Yeah, I have this little dream of a 2x2 linear motor 3d printer running on rust. I’m in the process of putting together the rust equivilent of sfoc at the moment, hoping that’s used.

What exactly do you mean by 2x2? I hate Rust, so we could only collaborate on hardware, but I have a dream of a printer using the Ultimaker style crossed rail gantry with two linear motors driving each axis. Ideally inverted from the Peopoly Magneto X, so the permanent magnets are on the moving parts and coils on the stationary rails so the motor wires don’t have to move. This is one of the potential uses for my stepstick-compatible board, along with making ultra-lightweight spiral extruders.

I’m pretty sure I could make the linear motors, but I don’t know what to do about encoders. It seems stupid to use belts to turn rotary encoders, but I don’t know of any low-cost linear encoders aside from vernier calipers, which may not be precise enough even if you can figure out how to get the reading from them in digital format.

check out the diy linear motor project: GitHub - cmore839/DIY-Linear-Motor.

The project owner has the same setup in mind as you regarding placement: Cross gantry, 2 lin-mot per axis.

As for inverting the coils/magnets, that’s a thinker. It would significantly inclease the moving mass, and you wold need a way to set which phase triplets to activate. You wouldn’t have enough default gpio-pins available, so you would probably need a multiplexer of some sort. I could see heat disapation being a massive advantage, assuming you can share the coil-energisiation across more than just 6 coils, but the wire advantage, I don’t see as being much of an advantage. With reasonable cable management, it should be a near non-issue. I would be keen to hear your thinking on it.

Making linear motors, as per the DIY linear motor I linked, seems quite straight forward, they’ve even got a coil supply setup if that part of the project doesn’t appeal. As for encoding, there’s linear encoder options:

  • HEDS: similar to what’s typically used in printers. affordable, suitable accuracy, works great with pulse counters. Has issues with environmental exposure (dust build up).
  • glass rod linear encoders. TTL lines, found them on aliexpress for about twice the price. don’t know much about them.
  • industrial style: great, but massively expensive.

The rotary encoders are probably more than suitable for prototyping, at least.

I hate Rust

I love the language for a lot of reasons. I love klipper, but the project owner is, in my experience, and it seems quite common among those that I’ve spoken with, is driven by ego, and filters the community such that he is surrounded by sycophants. The community and industry deserve better, and if something else comes along, I feel that it deserves to be in a modern language.

1 Like

Ah, that looks more realistic than what I had in mind. The rails would be like an unrolled outrunner stator, with coils wound around teeth. The moving part would be two rectangle magnets stuck to a piece of iron, so moving mass would be extremely low. It’s true that wires and cable chain aren’t that heavy, but it lacks the elegance of my pie-in-the-sky machine :slight_smile:

Thanks for the encoder recommendations, but I forgot the main reason it’s such a problem is that it also needs to be inverted so the sensors are on the rails, else the carriage will still have to drag wires around. That will have the advantage of giving an absolute position reading, but I doubt there are any commercially available solutions for it. I feel like the answer will involve a vernier pattern, using capacitance, magnetic field, light, or electrical contact.

Regarding selectively activating coils, I had considered it but I don’t think it’s really necessary. It will be producing a lot of unnecessary magnetic field, but since nothing will be pushing against it the only energy consumption should be in the resistance of the copper wire, and since that won’t be moving you can use a lot of it to minimize resistance. Although cost could be an issue.

Something I recently have been looking into is trying to replicate and improve the work on this project, which does not have fantastic resolution but depending on your number of pole pairs, could possibly work : Project | DIY Digital Caliper | Hackaday.io

2 Likes

Magnificent! I found about it last week, I really want it to move forward.

The only thing about capacitive encoders that I’m not too sure about is their sampling rate, and if they are any good measuring while in (“fast”) motion.

There is also the cheap-o-self-mixing laser interferometer: https://www.youtube.com/watch?v=MUdro-6u2Zg

There is also optical strips and magnetic tape for linear encoders:

I’m just starting to explore these options for a stick-slip piezoelectric nanopositioner (i.e. not recommending those products or suppliers).

Haha! We are living the same life on two separate halves of the planet… I have been obsessed with trying to build a << 1um resolution encoder for use with a nanopositioner also… I watched this video and the Sam Zeloof one recently. I think they are actually maybe too impractical for the stage you/me are talking about.

At my work we use the optical tape with very high pitch for closed loop motor control, but even this tape is like $100 per inch or soomething like that.

Did you read the paper in HardwareX also? I just ordered the parts to try and build it, I think that using some of those 28-bit inductive sensors might be the best method to set up a closed loop. I will dm you :slight_smile:

1 Like

Interesting, can I ask for a link to the specs?

I was thinking about using this kind of things as “optical tape”:

Or these:

Would they be useful?

I’m also not sure what sensor can be used to read from them. I imagine that either a very fine laser point (like the ones in OPUs) could work, or maybe the sensor from a high-DPI computer mouse (e.g. this thesis).

Hoping the rough ideas are useful. Implementation is out of my reach ATM.

FYI Magnetic encoders from RLS can get to 250nm/count for around $100/ea for tape and readhead, this is the cheapest “high” resolution encoder I know of from a non-ali type source.

2 Likes

I laid out an adapter board for a nucleo-144 that connects 4x ali drv8301 driver boards, and has SE/differential encoder inputs, along with some SPI/UART/gpio broken out to connectors. It’s laid out around the H723 peripherals, but I think nucleos are designed to be pretty interchangeable so hopefully the F7 and U5 I also have will at least partly work.

2 Likes