Duet 2 Hangprinter, 5-bar Scara, Polar, Rotary Delta kinematics

dc42

RepRapFirmware has grown to the point of filling up all the available flash memory (512kbytes) on Duet 2 WiFi/Ethernet. This means that after the 3.4 release we will need to either stop adding new features to Duet 2 firmware, or remove existing features to make room for new features.

One possibility is for us to remove support for some of the more exotic kinematics from the Duet 2 firmware build. So if you are using one of these kinematics with Duet 2, please speak up!

CNCModeller

@dc42 I'm using my old Duet 2 WiFi in a standard Cartesian machine now so have no requirement for polar kinematics on Duet 2.

I'll be using the duet 3 on my future builds.

Thanks for the heads up

Exerqtor

@dc42
Wouldn't an option be to split it up? Like one build with say corexy, cartesian and delta. And one build with the other types? If that's not a crapload of extra work?

dc42

@exerqtor said in Duet 2 Hangprinter, 5-bar Scara, Polar, Rotary Delta kinematics:

@dc42
Wouldn't an option be to split it up? Like one build with say corexy, cartesian and delta. And one build with the other types? If that's not a crapload of extra work?

Yes; but doing that would mean we had to test twice as many Duet 2 build configurations. Duet 2 is on the way to becoming a legacy platform, so we don't want to do that.

oliof

In my experience with writing the Tripteron kinematics, they're pretty well encapsulated and easy to to add in for a custom compile for those that need them.

That said, if possible I'd prefer Duet 2 to be feature frozen (probably with 3.4), maybe with backports for important fixes, instead of removing existing features to swap in new ones (this is not rational, just a preference). But if it would give another couple years of new releases and features, removing the rarely used kinematics from the Duet2 line up seems to be the most sensible step.

o_lampe

@dc42
I'm someone who likes to tinker with odd kinematics as a hobby, but I can't spend much money on it. Duet 2 is only one option to keep cost low. Gladly there are other low cost controllers which are capable to support RRFs new features.

JoergS5

@dc42 I developed the 5-bar Scara and currently develop robot kinematics.

I would prefer to have a core firmware and the possibility to link the additional kinematics, eg as separate binaries (analogue to dynamic dlls), if this is possible.
Something like M669 C"MyNewKinematics", the firmware checks for MyNewKinematics.bin and links and runs it.

This would allow

keep core small
develop and create the additional binaries
decoupling of releases

Disadvantage may be a slower execution of dynamically loaded bin than a static one.

The kinematics should then have a stable API for the functionality. I would propose some additional functionality for 3.5 like dynamic velocity of axes, which can be calculated inside the kinematics code.

oliof

@joergs5 that doesn't solve the space issue -- you would still need to build your own core binary that is small enough to allow for both linking and your new kinematics.

Also note that this only applies to Duet2, not Duet3 boards.

JoergS5

@oliof linking doesn't need much memory, and dynamic linking means the link addresses to run the separate binary are calculated on the fly. I mean, if you use Duet 2 without special kinematics, you only install the core bin without the special bins.

taconite

I talked to Tony on Formnext and he told me that support for Duet2 will be ongoing for years. Now you are thinking about a feature freeze. To be honest I don't think it's a good idea because then I guess more and more people are not going to use Duet2 anymore but prefer to run Klipper when the firmware for Duet2 gets more and more outdated. I can't say if the Duet3 boards are a viable replacement (e.g. I am running Duet2 and Duex5).

I prefer the option to remove the support for exotic kinematics from 3.4 onwards (so they could still run 3.4).

To make a real decision I guess it would be important to know what you would like to do the RRF in the future. Which features do you want to implement which are a USP for RRF

tekkydave

@dc42 I have duet2 wifi and duex2. Personally I don't plan an upgrade to Duet3 in the near future. I think removing the least used kinematics would be a good way forward. Are there any stats on what kinematics are in use?

JoergS5

@taconite said in Duet 2 Hangprinter, 5-bar Scara, Polar, Rotary Delta kinematics:

remove the support for exotic kinematics from 3.4 onwards (so they could still run 3.4).

RepRapFirmware is open source, so even if additional kinematics are not included in a release, they can be added and firmware compiled anytime, for every version.

vwegert

@dc42 said in Duet 2 Hangprinter, 5-bar Scara, Polar, Rotary Delta kinematics:

Duet 2 is on the way to becoming a legacy platform

Maybe it's just me, but that's a dangerous statement, unless you're also planning to drop the price of the Duet3 series components significantly. Don't get me wrong, I'm happy with my two Duet2s (Duets 2?) - but from what I can see, Duet has always been at the upper end of the price range anyway. The Duet3 ecosystem is very expensive to buy into. I can only speak for myself, but I would not have invested that much money (and won't for the forseeable future).

tekkydave

@joergs5 I'm OK with compiling my own binaries for 3.5 onwards if selecting which kinematics get excluded is simple enough. I'm not familiar with the RRF code.

jumpedwithbothfeet

@vwegert the price for the duet 3 mini is either cheaper or the same price as the duet 2 where I am, why would you choose the 2 over the 3 if that the case? is this not the same where you live?

oliof

@vwegert Duet 3 mini is a bit cheaper than Duet 2 (125EUR for the mini 3 vs 145EUR for the Duet 2) and compares pretty well for most projects that would have used a Duet 2 before. Duet 2 and Duex can be substituted by two Minis for the most part if you need Duex 5 capacity, and by the Expansion board for the mini or the rest of the CAN expansion boards if you only need one to three extra steppers.

Of course it would be great if the 8 years runtime of the Duet 2 platform can be extended, and removing rarely used kinematics seems to be a viable way to keep most owners happy.

dc42

For clarity: this change would affect the firmware binaries that we release. It would still be possible for users to compile the Duet 2 firmware with one of the exotic kinematics included, although they might need to remove support for e.g. delta kinematics to make space.

JoergS5

This post is deleted!

Diamondback

I personally think that removing some of the exotics is fine, upgrading from Duet 2 to Duet 3 is no trivial or cheap task, esp when dealing with more advanced printers that use more than one board. I'm afraid some people would be left behind if Duet 2 wouldn't be updated anymore...

In any case, can you please make sure that the Duet 3 platform offers all the planned upgrade paths before deprecating existing hardware (whenver that mught actually happen)? I'm mostly talking about the dual mainboard (2 Minis or a 6HC + a Mini) configurations that still have some restrictions (according to the documentation). I need to hold off my Duet 2 + Duex5 conversion to 6HC + Mini until that stuff is properly supported.

breed

@dc42 I vote get rid of them. Would much rather have more modern features on my production printers than a shorter path to an odd kinematic play toy printer.

tenaja

I, too, vote to get rid of the specialty kinematics for duet 2. I have around 6, and none need more than simple xyz or corexy.

JoePrint

In my opinion if you freeze features on the Duet 2, you're going to lose those owner's loyalty to the brand... When I can't have the next killer feature that comes out, I'm not going to invest in a Duet3, but will be looking at other cheaper boards that support RRF, or most likely, will be jumping ship to the Klipper side!!

I think removing the exotic kinematics with an option to swap them back in for a manual compile, or even having a pre-compiled firmware for both, is a good option.

@joergs5 said in Duet 2 Hangprinter, 5-bar Scara, Polar, Rotary Delta kinematics:

Maybe someone can even find good solutions for those problems in the future with additional hardware.

Along the same line of thought as what joergs5 mentioned, another option is to take a hybrid Klipper approach and offload some of the functions to another controller or RPi, like input shaping, DWC, heater tuning, etc. Or even a full Klipper approach. That will provide plenty of space for additional features and extend the life of the Duet 2 and 3 for a long time. Not sure how feasible that would be, but it was just a thought.

jumpedwithbothfeet

@dc42 I don't know about coding or the amount of work it would take but would it not be possible to have the kinematic made in a modular fashion kinda like a macro you can choose whilst setting up for the first time, imagine not many people change the kinematics once their printer is built, similar to what Exerqtor was suggesting, or are the kinematic to integrated?

vwegert

@oliof Interesting - for whatever reason I had the Duet3 mini as "fewer channels than Duet2" in my head. Thanks for clearing this up.

JoergS5

@dc42 if you decide to remove those kinematics from core RRF, I offer to take a lead for source and documentation for 5-bar scara, 5 axis robot, 6 axis robot and later this year for stewart/hexapod kinematics.

My proposal are dedicated github repositories on Duet3D (so the repositories can be found) for those kinematics with source and binaries, so users who want to use them can easily install them. I propose to offer binaries for Duet 3 based hardware (6HC CAN, Mini Eth, Mini WiFi, 6XD) for main releases and maybe for a part of the beta versions.