Duet 2 Pro 4 U

[Edgars Batna]

Is a Duet 2 with more oomph remotely plausible? I mean any sort of oomph, including, but not limited to, CPU, connectivity, stepper drivers etc. I didn't see anything being actively discussed anywhere.

Just a thought, if Duet 2 had more CPU and RAM and supported interpolation from any microstepping, that would halve the magnitude of my own usual headaches. it's not a complaint, just there are limiting factors.

I do like the self-contained nature of Duet 2. I could go Duet 3, but it's adding way more variables than advantages from my point of view.

[droftarts]

@Edgars-Batna Maybe Duet 3?!

Ian

[Edgars Batna]

@droftarts said in Duet 2 Pro 4 U:

@Edgars-Batna Maybe Duet 3?!

Ian

That's too obvious, but would mean throwing everything currently in my printer electronics-wise upside down. There's 0 use cases for CAN extensions, Raspberries and tool boards on my printer.

[droftarts]

@Edgars-Batna You don't have to use CAN or a Raspberry Pi on Duet 3, you can run it in 'standalone' mode, pretty much like a Duet 2 Ethernet. It gives more CPU, much more connectivity and more (and better) stepper drivers (with interpolation at every microstepping) than Duet 2. It seems to be exactly what you want!

There are plans to update Duet 2 line at some point in the future, but I doubt it will be a big change from the existing Duet 2. I think we've talked about having a proper connection for Raspberry Pi (like Duet 3), and possibly improved stepper drivers. But it's a vague plan and a while away while we sort other things out.

Ian

[oliof]

@droftarts raspberry pi support for duet2 is on the implementation list for RRF3.2 (and marked as done on the list).

[droftarts]

@oliof Yes, I am aware of this! Though while the firmware implementation is working, the hardware still has to be modified at present, so an updated Duet 2 may have a dedicated header.

But using an SBC is not on @Edgars-Batna 's wishlist! I was really pointing out that any updated Duet 2 might be closer to Duet 3, and have features that he would not use. So perhaps Duet 3 is not as great a jump as he was thinking it might be, especially as Duet 3 does not NEED an SBC to run.

Ian

[Edgars Batna]

@droftarts @oliof

I've got 8 steppers connected, running Duet 2 Wifi + Duex 5 currently.

Switching the printer to Duet 3 would mean £185.00 + £85.00 and I've got 2 leftover boards and loads of cabling changes.

If there was a Duet 2 Pro, then I suppose that'd be closer to Duet 2: £120.00 and I'd just swap the main board, staying compatible.

For me the SBC does not add that much. It's an interesting addition, for sure, but it theoretically just adds cost.

[bot]

If there was a board made as a direct "drop in" replacement for the Duet 2, but which had better specifications to allow faster print speeds/pulse generation/etc., I would buy it, even if it cost twice what the duet 2 does now.

[A Former User]

@bot said in Duet 2 Pro 4 U:

allow faster print speeds/pulse generation/etc

no rumours about the cpu yet, but seems unlikey to be a major upgrade as it might require dc42 to release yet another build config. maybe the man himself has something to add?

[dc42]

I can release a little more information about the new board.

When we designed the Duet WiFi and Duet Ethernet, our aim was to make one board that would be suitable for nearly everyone. So it needed a lot of expandability and high current stepper drivers.

We now have Duet 3, which covers situations requiring high expandability and/or high current stepper drivers and/or increased processing power. So the new board does not need to offer those things. Instead it will provide features that the Duet WiFi/Ethernet lack, such as more modern stepper drivers, simplified wiring for BLTouch, a 5V output for laser devices and VFDs, on-board 12V supply for fans, Neopixel support, and a Raspberry Pi connector. The processor will have about the same power as Duet 2 but more flash memory and RAM to make room for firmware improvements. Like Duet 3 it will be capable of running either with a Raspberry Pi, or standalone using on-board Ethernet, with a WiFi version too. There will be no expansion connector like the one on the Duet 2, but there will be an optional daughter board to carry additional stepper drivers, similar to the one for the Maestro. We will likely offer an alternative daughter board to provide 5V signals for up to three external stepper drivers. Other expansion will be via I/O connectors similar to the 5-pin ones that we use on Duet 3.

[NitroFreak]

@dc42 said in Duet 2 Pro 4 U:

I can release a little more information about the new board.

When we designed the Duet WiFi and Duet Ethernet, our aim was to make one board that would be suitable for nearly everyone. So it needed a lot of expandability and high current stepper drivers.

We now have Duet 3, which covers situations requiring high expandability and/or high current stepper drivers and/or increased processing power. So the new board does not need to offer those things. Instead it will provide features that the Duet WiFi/Ethernet lack, such as more modern stepper drivers, simplified wiring for BLTouch, a 5V output for laser devices and VFDs, on-board 12V supply for fans, Neopixel support, and a Raspberry Pi connector. The processor will have about the same power as Duet 2 but more flash memory and RAM to make room for firmware improvements. Like Duet 3 it will be capable of running either with a Raspberry Pi, or standalone using on-board Ethernet, with a WiFi version too. There will be no expansion connector like the one on the Duet 2, but there will be an optional daughter board to carry additional stepper drivers, similar to the one for the Maestro. We will likely offer an alternative daughter board to provide 5V signals for up to three external stepper drivers. Other expansion will be via I/O connectors similar to the 5-pin ones that we use on Duet 3.

Please just make a high CPU powered version with no internal stepper drivers and integrated 5V signals for external servo (or stepper) drivers. They can make use of the CPU speed. If that board is cheaper it can aid in the adoption of servos.

[A Former User]

@NitroFreak said in Duet 2 Pro 4 U:

Please just make a high CPU powered version with no internal stepper drivers and integrated 5V signals for external servo (or stepper) drivers.

that sounds like the cnc version of the duet3 (not sure about the 5v)?

[Edgars Batna]

@dc42 said in Duet 2 Pro 4 U:

The processor will have about the same power as Duet 2 but more flash memory and RAM to make room for firmware improvements.

Is there a significant reason to not upgrade the CPU further? It's better to have the power and not need it. Also, you know this better than I do, but more features / bigger code usually means longer critical path.

I've also seen Atmel offering dual core SAMs. That would be quite fun having some tasks not blocking each other so much.

Just brainstorm/-farting here.

[dc42]

@Edgars-Batna said in Duet 2 Pro 4 U:

Is there a significant reason to not upgrade the CPU further? It's better to have the power and not need it. Also, you know this better than I do, but more features / bigger code usually means longer critical path.

It's a matter of getting the price vs performance right. The faster processors from Microchip are a lot more expensive; whereas we are aiming to price this board at significantly less than the Duet WiFi and Duet Ethernet (although if the £ plummets in December then that may not be possible). For this board it would have been possible to switch to a non-Microchip processor because it doesn't need to support CAN-FD, but that would have increased the software development cost and time substantially.

The additional features we have added to RRF recently do not put a significant additional load on the CPU. They are mostly alternative options (e.g. different kinematics, and 12864 LCD support as an alternative to PanelDue support) or features that are used only occasionally (e.g. conditional GCode, and probing in directions other than Z). So we don't need increased processor power yet.

[bot]

@dc42 Would faster CPU clock speeds and/or increased memory allow for faster printing of GCode with many small segments?

I think Edgars is interested in having very fine mesh/segment/toolpath resolution while printing at increased speeds. I'm interested in this too, but I'm happy enough with the current Duet that I wouldn't be bothered if it never comes. Just a casual wish of mine to experiment and push the envelope.

That said, I know you have ideas about motion planning, electronics, etc. that are far beyond my understanding. So, I'm just trying to convey the reason why we think we want faster CPUs. Perhaps you have a better way of achieving it which is even potentially currently possible on Duet hardware.

[pixelpieper]

@dc42 why go the extension board route for the extra steppers? I would prefer to have them all on the board, gives a better thermal performance and you do not have to stock an extra board. I assume you look into 2209s, the additional cost for three more is not that high. You still could have a lite version of the same board where you only populate 5...

[lepetitnicolas]

@dc42
Hello,
When do you release this new card ?
Currently, i used raspberry with klipper and the support for raspberry like duet3 is a good thing
Thanks

[A Former User]

@lepetitnicolas said in Duet 2 Pro 4 U:

When do you release this new card ?

this is the latest news https://forum.duet3d.com/topic/17409/duet-3-mini-5-initial-announcement/1?_=1593996217004

not sure if that was the board in question?

[dc42]

@pixelpieper said in Duet 2 Pro 4 U:

@dc42 why go the extension board route for the extra steppers? I would prefer to have them all on the board, gives a better thermal performance and you do not have to stock an extra board. I assume you look into 2209s, the additional cost for three more is not that high. You still could have a lite version of the same board where you only populate 5...

A board with 8 drivers populated would need to be significantly larger. It would no longer be possible to keep it the same form factor as the existing Duet WiFi/Ethernet/Maestro, and the large PCB would add to the cost. Also, having an 8-driver board and offering a lite version with only 5 drivers populated is not a good option either for us (because we would have to get the production balance right and hold stocks of both types) or our users (because the 5-driver version would not be easily upgradable).

[dc42]

Here are the prototypes, both WiFi and Ethernet versions.