I guess its almost impossible to have a 1 fits all implementation. But really, does having the MCP23017 16 channel just on the autosteer board under the wasted space of the cytron maybe just a good use of space, whether used or not?
Maybe having an I2C driver/repeater chip for external connection should be included as well.
It also would need the ethernet card as well. If going to completely redesign the board, would using an ESP32 be better? Or another new 32 bit design? Teensy? etc
TOO many optionsā¦
Built in ethernet port would be cool, although the one I use just stacks between nano and PCB anyway, so looks pretty clean / fineā¦
I dont really know if ESP would be betterā¦Matt seems to use them for most of his projects?
I really think the autosteer board should be just that, autosteer, UDP connection, job doneā¦ Each tractor would have its autosteer board and or machine control board, and implements would get āimplement boardsā which I guess is essentially a section boardā¦ also UDPā¦wether it be ethernet (my preference) or WiFi makes little oddsā¦
The problem as always, is so many use cases I guessā¦as I guess to some (or maybe many) in a single tractor / virtually single application use for AOG, having autosteer, section control and all functions on a single board would make for an easier install, would never need to expand outside of that and would do everything they need toā¦its swings and roundaboutsā¦
I have built my autosteer box around a PCBV2 now, so probably wouldnt really want to change that unless I had to, and I was hoping to add section boards to my sprayer, fert spinner, and other implements, and have a single RJ45 ethernet connection for each implement to hook up to the the tractorā¦
I guess having it on the PCB dosnt do any harm, and just document that if you want to run with no, or with external section control, just dont bother with some components on the PCBā¦ if its wasted space anyway, its not doing any harm / might be of benefit to some peopleā¦just as an optional thingā¦
Definitively my prefered option is prebuilt relay boards. This is what we are using with Cerea. They are optoisolated, easy to find, cheap solution, easier to integrate, easier to replace for the case that is broken.
So it might be better using thinner wires? I use one twisted pair for SDA and one twisted pair for SCL of a cat 6 cable. Do you think, using just one wire (instead of the pair) might reduce the capacity? Another question might be what would be the minimum frequency for mma and bno application.
I am surpriced that you have to reduce I2C frequency in the video.
Because I have the nano on the right door side, bno at the roof, the mma about 1,2 m away from the nano. At nano side and bno side I have a screw connector and at mma side I use a plug. I have had one fail yesterday, but it could have been another problem (it doesnāt repeat today and it didnāt happen the day before). And on another tractor I have almost the same setup, just the mma on the pcb - here I have not detected any problems yet.
I vote for a I2C driver as you suggested.
ESP32 is up to 2 orders of magnitude faster than the arduino in some circumstances. ADC isnāt very linear but we donāt use that anyway. They also have numerous comms options. ESP-Now etc.
The best way is a twisted pair of SDA and SCL. Just those 2 twiste togther. Same as inside an ethernet
cable
Continuing the discussion from Autosteer/Relay Board:
Perhaps redesign to support nano 33 boardsā¦
just thinking perhaps Xbee socket integration for communication options? Avoid building another case/enclosure for gps, etc
beau boulot
expanded auto steerboard: You need here a lot of cables to the relayboard, which might be in another box, so I prefer the solution with the I2C bus going to relay board. But in this case you are restricted again to the free pins/ports of the steering nano. If you need e.g. some input pins this might be the āshow stopperā (except there is an i2C - IC for this also). So I vote for i2c relay board and for separate machine controll board.
I suppose the ideal scenario is don;t limit any particular design. Even though there is 16 pins of relay out on the autosteer board, with the i2c connector you can still add anything you want, our just drive a simple opto realy board
i2c relay board: Range is restricted, I guess you can not put it on the big air seeder 4 - 5m away. This votes for the separate machine control board, e.g. concted with LAN, again. So maybe it is not worth to spend time in this project, except for learning? (I have learned that there is a MCP23017 usable with arduino, great!) On the other hand, 16 or 20 wires to the relay board into the cabine is also doable. You donāt need even a plug for it, if the bus cable could made be plugable? However, a I2C repeater/driver then is a need. (This I learned, also.) 
Well, it really is about flexibility. If the relay requirement is close to the autosteer then I2C is the way to go. OR if distance needed, use the machine port, or ethernet, or wifi 
I guess there is no single best way at all
Or something like this and the i2C relay board can be put to the end of the seeder (if it works):
Ya, you need two chips, one at the master, the cable, then the receiver
Yes. Would it be a option for V4 autosteerboard? Will increase flexibility for sure.
Well not on the board, but you could build your own and attach. But why do that, just use the machine module part and use UDP or wifi.
So you have been coming to a conclusion at this topic - no MCP23017 or something similar on the autosteer board?
As others have said, I also like the possibility to add some relays for use mainly in cabin (lift and maybe a few sections) And I think easy to set up for beginners.
So I vote for the MCP on the autosteer board
Looked at your picture of board, which I like with the fuses, but if you have not already spotted it yourself : A0 A1 and A2 did not move down together with the 3 connectors.
Yes thank you, part of the cleanup.