Make my own EtherCAT device. 1 Digital IO

[@billstrahan4791]

I'm trying to learn as much as I can about EtherCAT, so the videos are most welcome!

[@DesertShedStudio]

Nice, I have a Mikroe EtherCAT Click board (also LAN9252 based) I picked up on Rod's recommendation. I need to get back to working with it, should be a good way to add SPI sensors and I/O to LinuxCNC running EtherCAT.

[@kubabuda]

Do you plan to program the microcontroller? These black boards are OK for start, but STM32 has no pins broken out to do anything useful

[@darkobul1]

Well I didnt think to go that far but considered using two ethernet modules with breakout board. Only advantage beside linking ethercat really has is galvanic isolation which is provided by ethernet it self. Each pin has isolation coils. Or I am missing something more about it? I did not get any ti try brcause of price and was thinking I can accompish something by using modules. I didnt watch whole video yet. Looking forward to see.

[@darkobul1]

Main advantage of EtherCAT is removed portion of bytes used in standard network making its packets smaller and dus faster to repeat. But if you consider that same can be achived with faster network if you want to chase nanoseconds. But you lose so many advantages of ethernet/ip which are never pointed out.

[@kubabuda]

Main advantage of EtherCAT is its timing consistency. Reading updates from every node in network will take the same amount of nano/microseconds, every time. You also get distributed clocks which AFAIK make it best protocol for timing-sensitive things like hi-perfromance motion control. For these things, faster network means nothing when its not repeatable. Cost is extra 10-20$ in BOM for every device in network, decreased network throughput, using separate network card (you dont want to mix it with normal ethernet comms) on master node. This thing is more specialized tool than ethernet/ip

[@darkobul1]

@@kubabuda i had no clue yesterday about it. Thanks to researching this I learned bunch of stuff.

[@darkobul1]

@@kubabuda i see you can get cheaper but imagine my shock whrn I saw MicroChip sells dev board for 150$ which in reality dont cost even 10$. I understand its a business thing but I doubt this protocol will live long unless those chips became pennies.

[@darkobul1]

Does ethercat support ordinary network switch to split multiple devices?. On redit someone spoke unceranly about it while it seems uses same network layers. I see those modules and devices have 2 ports for linking. But if switch could be used then single port and many MCU that has PHY could be used even at different speeds.

[@darkobul1]

Seems all EtherCAT devices work on 100Mbs. Not a single example on youtube except specialized devices and networks. So answer is likely no. All they could do is just shave few bytes from ethernet/ip header and for that reason is so much expensive and closed. Because its different inplementation.

[@darkobul1]

One guy on redit said something that makes it possible and there are cheap switches that provide vlan linking all ports. So it perhaps may be possible with some switches. "EtherCAT cannot coexist on the same network as regular TCP/IP stuff, including EIP. Again, a snazzy enough switch can probably VLAN it's way out of this and keep the EIP garbage from getting in EtherCAT's "

[@darkobul1]

Accirding to ai EtherCAT rides directly on Ethernet frame removing ipv4 20 bytes and its own protocol header is 8 bytes long like UDP.

[@kubabuda]

You are getting at it from wrong end. When you want multiple EtherCAT devices in network (which you usually do), you just daisy chain them: connect OUT port of previous device to IN on next one All Ethercat interface chips have 2 ports. Some can have 3 (like LAN9252), so you can branch your network if you really need it, but daisy chain is usually what you want

[@darkobul1]

@@kubabuda i did not know when I asked. I was looking to find a way how to test EtherCat(without buying thousands of dollars equioment and if will suit to use it as extending protocol for LinuxCNC. But now I know that many of my assumptions where wrong. I was thinking you may use different network speeds and mix slow and fast but that seems is not the case. They all have to be same speed of 100M and as you said they have to be linked in a row. While that may give some advantage in certain ways it loses other.

[@sebaschtl9710]

very greate! thanks alot

[@DasMooorhuhn]

That is super cool

[@darkobul1]

So really they have a special chip for ethercat? Wow...

[@darkobul1]

EtherCAT uses the standard IEEE 802.3 physical layer. No special hardware is required to implement an EtherCAT network.

[@darkobul1]

Github is full of ooensource code for it. I had no clue. I was thinking its something proprietary.

[@darkobul1]

And it was proprietary but made as standard but still lot of unplesnt things linger behind it.

[@darkobul1]

Some make chips with hardware implementation of EtherCAT protocol so thats how.

[@kubabuda]

@@darkobul1 you dont need special HW on master end, each device needs ESC (EtherCAT Slave Chip). You _could_ implement this in firmware if your microcontroller has 2 Ethernet ports, but this will be so slow that its pointless