Timeline for Is it possible (with reasonable effort) to use ethernet hardware to provide point-to-point digital I/O bridge?
Current License: CC BY-SA 4.0
18 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 3, 2021 at 9:39 | vote | accept | M Szil | ||
| May 2, 2021 at 2:04 | comment | added | Bruce Abbott | "8-bits in each direction... and up to 30m distance and ... <1us latency" - for what purpose? | |
| May 1, 2021 at 23:18 | comment | added | jcaron | @Justme I must be missing something. The smallest Ethernet frame is 64 bytes, or 80 bytes including preamble and interframe gap. That 640 bits, which at Gigabit speeds is 640 nanoseconds, just under a microsecond. Granted, with the overhead at both ends this may jump to above a microsecond, but on the wire, sub-microsecond is definitely possible. | |
| May 1, 2021 at 22:59 | comment | added | user57037 | SPI could also work for you. You can SPI into a serial in parallel out shift register. Add a latch signal, and you have a great little IO expander. There are also I2C IO expanders. | |
| May 1, 2021 at 22:58 | comment | added | user57037 | Maybe what you are really looking for is a serdes. But UDP packets are also very lightweight fire and forget packets. | |
| May 1, 2021 at 3:21 | answer | added | Jasen Слава Україні | timeline score: 0 | |
| May 1, 2021 at 3:14 | comment | added | AnalogKid | AND - is the remote system powered out at its location, or from the base station? | |
| May 1, 2021 at 3:10 | comment | added | AnalogKid | You don't say what your limits are on the size of the bridge. For example, 34 wires would be 8 differential pairs in each direction plus power and GND. Assuming you are aiming for something between that and just 2 wires, what range of wire counts can you handle? and please don't say "as few as possible". | |
| May 1, 2021 at 2:46 | answer | added | Kartman | timeline score: 1 | |
| May 1, 2021 at 1:04 | answer | added | crasic | timeline score: 2 | |
| May 1, 2021 at 0:58 | comment | added | crasic | EtherCAT IO modules process packets "on the wire" and theoretically 15us response is possible, but 1ms is more typical. Very deterministic, if desired (includes synchronization mechanism) , is simple for master to control (specialty crafted ethernet frame). | |
| May 1, 2021 at 0:50 | comment | added | Math Keeps Me Busy | Can it be done? Yes. Would I recommend it? No. Ethernet is a heavy-weight solution for reading/writing remote GPIO pins. If you want to go that route, Raspberry Pi is the solution. But "something" over RS485, such as CAN, Modbus, or (I2C over RS485) will be much simpler. RS485 is industry standard for remote bit-banging, and is quite reliable. If you love Cat-5/6 cabling, you are welcome to use it. If you need galvanic isolation, that can be arranged. Need some power? That can be added. | |
| May 1, 2021 at 0:09 | answer | added | hacktastical | timeline score: 2 | |
| May 1, 2021 at 0:05 | comment | added | Justme | Sub-microsecond latency and Ethernet is not possible. Smallest Ethernet packet is multiple bytes so even at gigabit speed that's multiple microseconds and with overhead that causes non-deterministic behaviour. Unless you would care to specify what exactly you mean with "Ethernet hardware here". Do you mean cabling, connectors, other hardware? | |
| Apr 30, 2021 at 23:59 | comment | added | brhans | This TI AppNote might be of interest: Hundreds of Megabits @ Hundreds of Meters: Extending the Transmission Length for LVDS | |
| Apr 30, 2021 at 23:57 | comment | added | user76844 | you are right about using ethernet for this application, but don't cut corners. it's not too difficult to have a microcontroller that will simply take UDP/IP packets and toggle gpio. even take a second rapbery pi, it's probably cheaper then developing a new board. | |
| Apr 30, 2021 at 23:51 | review | First posts | |||
| May 7, 2021 at 7:40 | |||||
| Apr 30, 2021 at 23:44 | history | asked | M Szil | CC BY-SA 4.0 |