duet3 and e3d ToolChanger and Motion System
-
Toolboards are somewhat like a 3HC expander, except they are physically smaller, and have one of everything. One stepper driver, one heater mosfet, one thermal sensor input (actually, a couple), two fans (heatbreak and part), a few general I/O, and so forth. They are CAN attach.
Therefore, your tool changer can have two power wires, and 4 CAN wires, running to each tool. Instead of 10 or 12 wires.
Tool boards are currently in final beta, to be pre-order in the next couple of months, and general order soon after that.
-
@Danal said in duet3 and e3d ToolChanger and Motion System:
and have one of everything.
assumed so but was unable to find the link coooooool ... definetely no reason to rush with d3 order, d2 will get the job done more then adequate and then I can move to d3 ... knowing myself I'm going to now dump my major test system and make something between my cubebot and this toolchanger but so far I dislike the corexy.. we'll see with bit more time with toolchanger maybe it grows on me .. currently my cubebot is operating with 2 independent X axes but will see how corexy shows up in real usage and how this new head changer thingy shows up before I decide to move on to new build... I surely have another 7-8 months to play with it before I can start making anything new (on some huge project attm that's supposed to be done before september, so ideal time to use TC as is without "improving" it ) ...
anyhow what's much more interesting to me is the whole synchronous CAN movements, need to get that out of the firmware, as I had serious issue making something like that (over modbus but same thing) last year for some project, I had to introduce the "trigger" line and changed the "standard" modbus to some ugly custom monster I'm really not happy with.. and it was 2 axes only!!! so I'm stealing the ideas from RRF as soon as I .... (I did take a peek in the source and it's not "my way of doing things" so it will take bit more time to figure everything out but hey.. that's why we do share the source so we can all learn )
-
@Danal hm, who's designing this tool board, I'd love to make some changes :D... I was planning to make one "huge board" out of duet2 + duex5 + "some upgrades" but looking at this concept now I like it more with a ton of can slaves and fewer wires .. but I'd love some "minor" changes to make it more versatile
-
@smece said in duet3 and e3d ToolChanger and Motion System:
@Danal hm, who's designing this tool board, I'd love to make some changes :D... I was planning to make one "huge board" out of duet2 + duex5 + "some upgrades" but looking at this concept now I like it more with a ton of can slaves and fewer wires .. but I'd love some "minor" changes to make it more versatile
Like all of Duet's products, I'd imagine that the design is based on what end users want or have asked for. I'd also imagine that because Duet 3D is a business, their design decisions will be based on the potential sales volume of units. In which case, I'd also imagine that OEMs will have far more influence than us single users. There is of course nothing to stop you from contacting the Duet team with your design change requests. But if what you ask for is something that will only suit a single user, I doubt very much that it would be put into production.
-
@deckingman I think it would benefit all, but I'm more in the "lemme get my hands on kicad files and gimme few hours " (I do prefer altium, as that one I know how to use, kicad is still not my forte but, better than eagle anyhow ).. and it's rather small changes, nothing serious, but important for me (and possibly others)
-
The v1 tool board design is frozen and we are testing the first pre-production ones this week. We are responsive to user needs, so please us know what additional or changes features you would like. Bear in mind that it's impossible to produce a single design that's perfect for everyone. Also, a single-tool board must be small and reasonably priced.
-
@dc42 said in duet3 and e3d ToolChanger and Motion System:
The v1 tool board design is frozen
Ah, /me late again
perfect for everyone ... reasonably priced.
The changes I have in mind are mostly on the pcb side that can probably be implemented without changing pcb dimensions and should not change the price of the board .. but I don't see any schematic/pcb on the https://github.com/T3P3/Duet.git repo? am I looking at the wrong place? otherwise, I'd first make changes and sent them as a suggestion
most of the "changes" I'd make would be adding few headers, I really hate that none of the boards use the 8 external diodes, with the cheap 8 diode protection board in pololu header format, I'd for e.g. love to have a pololu header on the board over the existing driver, this way I can easily add 8 diode protection as add on, also if driver blows I can easily remove it and push in another on in to the header ... as while removing the no-legs chips is super simple with hotair and some minor experience, replacing it is a hassle so that header would be useful...
second header I'd add is a jtag for the arm, not a full .1" header but something small, for e.g. non tented vias that I can push pogo pins to to access the arm directly if need be..
probbly few more things but without schematic and board design impossible to say, all I managed to see is the grayed image showing available connectors and dimensions
-
@smece, I have seen no evidence that the 8 external diodes do anything useful at all. Seeing that the stepper drivers already have similar diodes integrated, I very much doubt that they do anything useful, apart from increase the output capacitance. That might just possibly be useful when using Pololu drivers, but not on Duets, which have output capacitors built in.
The board already has a SWD connector, so it doesn't need JTAG.
-
@dc42 said in duet3 and e3d ToolChanger and Motion System:
@smece, I have seen no evidence that the 8 external diodes do anything useful at all.
I did :D, first with Toshiba drivers, then with TI drivers. Tested TMC2130, also helps ... first EMI, tested with old nema23 motors (different ones, I have 100+ of those that I collected over the years), measure interference around the printer and those 8 diodes lower it significantly .. 3dB at least with TMC and over 5dB for Toshiba .. second, kill the motor wire while spinning, 8 out of 10 times driver dies without those diodes, with diodes I'm yet to see dead driver from motor connector shaking lose or pinched motor wire .. so they do help.. major thing I noticed is they help immensely to stabilize ADC input (not a big deal on 3d printers but I use steppers for some chemistry equipment and that noise on adc is super problematic) ...
again, I would not add the diodes on board, just add header so you can push them in if you have use case for it...
also, step/dir/enable + spi pins for driver configuration, on "any" header so one can hook external stepper driver... I don't have a problem to desolder the TMC from the board and hook my gecko stepper or siemens or even xinje servo driver .. I'm talking headers only, nothing else
The board already has a SWD connector, so it doesn't need JTAG.
that's cool, I haven't seen that on the images I found ..
-
@smece said in duet3 and e3d ToolChanger and Motion System:
and this toolchanger but so far I dislike the corexy.. we'll see with bit more time with toolchanger maybe it grows on me
Take a look at Jubilee. It is E3D toolchanger compatible (spacing of kinematic lock, etc) and designed to use plastic parts in many places that E3D uses metal. Because of wear, it uses a rather subtle and clever torque sensor on the tool lock. It also has a good implementation of CoreXY, with all the places that need to be precise in that geometry correctly engineered to be precise. Anyway, take a look. Right now, it is my tool changer of choice.
@smece said in duet3 and e3d ToolChanger and Motion System:
what's much more interesting to me is the whole synchronous CAN movements
Duet 3 movement commands are not synchronous. That would be a train wreck on a shared bus.
Moves are sent approximately 1/2 second before they are scheduled. Each phase of the move (start, accel, const speed, ramp down, etc) is sent in 'ticks' and periodic messages on the bus keep all boards in tick sync.
Much, much, simpler and more effective.
-
@Danal said in duet3 and e3d ToolChanger and Motion System:
Take a look at Jubilee.
Already did, I think I heard about it from you, def. on this forum
Thing is I'm in some huge project attm so pre September I'm not building myself anything.. That's why getting toolchanger & motion system from e3d was imo a good move as I can study them before I move to a next build... I already bult many printers (2x2x1meter was the biggest one) .. joined dr. Adrian 10 years ago in his quest to populate world with reprap .. but I'm too stubborn about stuff I don't try myself and even more stubborn about stuff I did try so.. "overengineered printer from e3d" sounded like a good idea for a start (first dissapointment are these acrylic sides, I was expecting aluminium, second dissapointment the thin bottom sheet, I was expecting at least 8mm botto .. but .. I get why.. shipping cost is already terrible)designed to use plastic parts in many places that E3D uses metal.
I like metal parts, have few cnc milling machines and mini lathe.. my first printer was rapman, if you remember those, made by uk company bitsfrombytes (that 3dsystem acquired for peanuts) and it took me less than a month to replace all acrylic pieces with 5mm aluminium ones ...
Duet 3 movement commands are not synchronous. That would be a train wreck on a shared bus.
sorry for not being clear enough I did not assume "commands" are synchronous, I'm talking about synchronous movement and the way that's achieved. yes, dc already told me he's filling the move table on the slave and is sending sync "ticks" to keep them in sync... thing is I tried to do exactly the same over modbus and xinje controller and I could not make it work, then I made my own slave and again my axes are going out of sync, not immediately but... and since the implementation is actually on very sensitive and dangerous device where 5kW servo's move around 6 metric tonns with moments of force going over 10000Nā m so I had to give up and introduced the separate "sync line" to sync the drivers... terrible solution but works.. I know these synchronous movements are possible, I recently repaired a huge 7 axes cnc for cutting glass where all the drivers are on the twisted pair going into plc so it does work, it's just I never managed to do it myself so seeing a source of a working solution helps waaaaaaaaaaay better than just empty theory
-
@smece said in duet3 and e3d ToolChanger and Motion System:
I'm too stubborn about stuff I don't try myself and even more stubborn about stuff I did try so.
Man, I don't know ANYONE else like that...
[bus based ] synchronous movement... ...a working solution
Also take a look at Klipper.
-
@Danal said in duet3 and e3d ToolChanger and Motion System:
Man, I don't know ANYONE else like that...
Also take a look at Klipper.
checked it out the day it was published ... I'm heavy octoprint user and they are heavily dependant on the octoprint so I kinda heard about it immediately,m checked it out, very similar concept as replicape with 2 pru's executing stepper stuff only better as does not require specific hardware, they made firmware for almost all boards used in 3d printing word to behave as slave ... I haven't tried it for printing yet (no time) but I did use it for some quick adaptation of release table for glass bending and quenching (if that's proper english term for glass) as I needed to turn 12 steppers, read bunch of sensors and had to be done "in a day" so I used klipper and it was perfect for the job ... it works similar to mesa cards with linuxcnc