Anyone seen this?

I think there will also be a place for Mesa Hardware as its an amazing ecosystem and fantastic for retrofits.

For a new build designed for resale, Ethercat is very attractive but more costly. Well planned, there is significatly less wiring and its far more flexible. Say on a gantry table, why not use AC servos and place a IO module for home and limits with the servo driver right beside the motor? Or put another up on the gantry head  and/or tool changer?

There are all kinds of cables available today. Why not use one with 3 phase power for servos and ethernet combined?

Don't loose sight of the fact that CNC is a small corner case with Ethercat technology and pales in comparison with factory automation. Your market is much larger than CNC.

rodw wrote:

I think there will also be a place for Mesa Hardware as its an amazing ecosystem and fantastic for retrofits.

For a new build designed for resale, Ethercat is very attractive but more costly. Well planned, there is significatly less wiring and its far more flexible. Say on a gantry table, why not use AC servos and place a IO module for home and limits with the servo driver right beside the motor? Or put another up on the gantry head  and/or tool changer?

There are all kinds of cables available today. Why not use one with 3 phase power for servos and ethernet combined?

Don't loose sight of the fact that CNC is a small corner case with Ethercat technology and pales in comparison with factory automation. Your market is much larger than CNC.
 

Well I don't think anyone's trying to replace Mesa, it's still generally the best bang for the buck. I'm just sort of mentioning that I think they can actually offer EtherCAT based versions of cards to integrate into the big picture of a machine just running EtherCAT hardware. It's always more complicated than I think it is, but I think those Ethernet Cards don't actually need to be physically changed at all to do so.

Not sure I'd want to stick servo drives right next to motors or on gantry's or use AC servo cable that's combined with Ethernet. Those all seem like very bad ideas from a EMI and caps falling off of PCBs perspective. The point about the IO module is actually understated tho, most of the newer EtherCAT drives I've seen have a few GPIO right on the drive and since it's EtherCAT you should just be able to read/write it just the same. The limit and home switches can just be connected to the drive itself without the need for an IO module.

It's all still very expensive though, I just looked at the price of the Leadshine EtherCAT servo drives and they're up there which is why the Mesa stuff will still be appealing for a long time. As for the whole factory automation thing, I've been looking at dusty old Siemens PLCs for years....actually I wish it was just Siemens PLCs but that's a different story. No small scale manufacturer can deal with that level of scrutiny, which goes to the point of the reason it costs so much but I don't see Siemens PLCs in stuff that's not at least on the level of a Haas so I don't know that people would use that level of stuff around here given a choice.

I am keen for more affordable ethercat devices and would be happy to help out. I have no formal coding or electronics experience and just kludge things together. Bank rolling prototypes / production runs is something I could help with.

I have been tossing around ideas for an ecosystem.

Unmet needs:
Low cost IO
- example: ethercat IO module on gantry for limit switches, home switches, touch probe. Just need Ethernet cable and power from control box to the gantry
- multiple ethercat esc can do 16 GPIO without any MCU / FPGA

Lots of IO at reasonable cost - even cheap Chinese undocumented IO are $100+ for 16 IO, compare to Mesa where you get 48 IO
We should be able to have lots of IO for reasonable cost

Low cost HMI

Step generator. To link existing step/dir servos/steppers into ethercat

Sorry I misunderstood what you wrote.

Maybe there is enough interest in the community to make a couple different designs and open source them?

LAN9252 would be the ”easy” option as you could pair it with any mcu but something like the AX58100 would be pretty cool. You could do tons of stuff with the chip alone (ofcourse you need supporting components).

About the FPGA ethercat IP core, is the license tied to the chip like it is for example LAN9252, so that you have ”limited” number of FPGAs that you can work on? Or you need to buy a software license to integrate it to an FPGA?

AX58100 looks good, but reliable suppliers an issue.

AX58400 looks amazing, but haven't found a source. I have not contacted ASIX to enquire.

I can't see creating an ESC in FPGA being worthwhile. LAN9252 is about $10.
www.reddit.com/r/FPGA/comments/10ydm1i/ethercat_slave_in_fpga/

Cheap MCU like RP2040 ($1) are plenty for most use cases. And easily programmed / reconfigured.
ESP32-P4 also looks like it would be a good option for MCU.

Mikroe EtherCat Click already offers LAN9252 -> SPI
So one simple first module might be SPI -> MCU -> GPIO
(This module could then also be used directly with RPI / SBC with SPI)

RDA wrote:

Sorry I misunderstood what you wrote.

Maybe there is enough interest in the community to make a couple different designs and open source them?

LAN9252 would be the ”easy” option as you could pair it with any mcu but something like the AX58100 would be pretty cool. You could do tons of stuff with the chip alone (ofcourse you need supporting components).

About the FPGA ethercat IP core, is the license tied to the chip like it is for example LAN9252, so that you have ”limited” number of FPGAs that you can work on? Or you need to buy a software license to integrate it to an FPGA?
 

The link I posted in the first post was open source from the "MetalMustangs" Youtube channel. He posts here, cant remember his screen name....something with starts with an H

I looked for info on that AX.... chip as MetalMustangs switched to using it for a 2nd PCB....he didn't explain why. I couldn't find good info and I never heard of ASIX so It's not something I want to touch. I'm more worried about the code personally, the code for the above doesn't look very friendly. If I can get my friend involved he'll get that straightened out....if he takes it seriously.

As for the FPGA core I have no idea, I'd imagine the cost is less than implimenting a physical chip. But yeah I think it's generally charged by the number of devices produced since it's IP.

pippin88 wrote:

AX58100 looks good, but reliable suppliers an issue.

AX58400 looks amazing, but haven't found a source. I have not contacted ASIX to enquire.

I can't see creating an ESC in FPGA being worthwhile. LAN9252 is about $10.
www.reddit.com/r/FPGA/comments/10ydm1i/ethercat_slave_in_fpga/

Cheap MCU like RP2040 ($1) are plenty for most use cases. And easily programmed / reconfigured.
ESP32-P4 also looks like it would be a good option for MCU.

Mikroe EtherCat Click already offers LAN9252 -> SPI
So one simple first module might be SPI -> MCU -> GPIO
(This module could then also be used directly with RPI / SBC with SPI)
 

The ESC IP core already exists, you just buy the license. The whole point was that we already have tons of FPGA cards on the market, the Ethernet core can be replaced with the EtherCAT core. I'm sure it costs less than the actual chips in the long run.....doesn't mean it's cheap and it's something Mesa themselves would have to offer.

Those Dev boards aren't useful for anything I'm doing. I took the "easy route" and made carrier PCBs for dev boards for my current project and got burned by bad software projects. I'm just going to create a proper board this time and stick an STM32 next to a LAN9252 and let somebody go to town on the software. I don't see any good reason to use an RP2040 or anything else over an STM32 for what I'm doing, I have alot more options with STM32.

looks like metalmusing is hakan

hackaday.io/project/181058-ethercat-servodrive

notes on lan9252 and ax58100 with stm32

but yeah all the dev boards are too much money for what im doing, it would defeat the purpose of my project. if we can make something like that ethercat motion controller work with 3 axis, then it would be worth it for me, and i would make a shitty 3 axis stepstick board using one of those stm dev boards

If you look at the reddit link, a couple of years ago Ethercat licence (IP Core) for FPGA ethercat was "23k USD for ET1815 or 6k USD plus 3k per 1000 devices for the ET1816"
ET1816 is for small business.
For the development of an EtherCAT device, the ET1816 one-time kick-off charge is required, plus ET1816-1000 royalty for 1000 devices. The royalties for 1000 devices must be paid in advance each time.

So even for 1000 devices, we are looking at $9 USD per device, PLUS cost of the FPGA.

Does not really compete with cost of already available ESC, unless making very large volumes.

For most uses we don't need an FPGA.

RP2040 is attractive because of:
PIO for step generation
Low cost
Availability has been good
(all in theory, I know very little really)

pippin88 wrote:

Unmet needs:
Low cost IO
- example: ethercat IO module on gantry for limit switches, home switches, touch probe. Just need Ethernet cable and power from control box to the gantry
- multiple ethercat esc can do 16 GPIO without any MCU / FPGA
 

That could be done, just depends on what the definition of low cost is. If you made 100 of them and gave them to friends it'd cost you $20ea. If I complied with the ETG stuff, best case is $40 as it would still cost me $20 to make them and I'd have to offload quite a few just to recoup the cost of the CTT stuff.

Lots of IO at reasonable cost - even cheap Chinese undocumented IO are $100+ for 16 IO, compare to Mesa where you get 48 IO
We should be able to have lots of IO for reasonable cost

For "lots of IO" it's a bit trickier. I'd have to think about it You can obviously use an MCU over SPI to the ESC to get the MCU's IO. Maybe those exotic ESCs have something exotic, not sure. Another possible option is to use multiplexors off the ESCs IO, I think that would be a reasonable way to do it.

Low cost HMI

www.bausano.net/images/arduino-easycat/E...FT_display_AN003.pdf

You can do something like that just like that. Wouldn't be too difficult to come up with a better SPI touchscreen and have it all in some enclosure. The problem with that is the "UI" is hardcoded basically and I'm not into writing embedded UIs so not sure how that would work out but hardware wise it's possible....not sure how "low cost" tho.

Step generator. To link existing step/dir servos/steppers into ethercat

The link in the first post the guy has that. They work but he's having issues. hardware wise it's not difficult. Not quite sure what issues he's having or what's wrong with the code. Remora targets STM32 and I think they have good stepgens, not sure if that codes portable or not. Either way If he get the code issues resolved it shouldn't be an issue to use that code in any ethercat project.