Still working on the cabling standard, and I think I really like this one. In this version. It still has pairs of pins on the ends of the hot end connector that can be omitted if they are not needed for current capacity, but the remaining pins for +V and heater are moved to one side so that I can interleave them with other pins. This is better for thermal management (and should actually increase the ribbon cable’s current capacity a bit, since there are fewer adjacent pins carrying current) and interleaving the i2C and thermistor lines with V+ helps protect them from cross-talk. Another advantage is that there’s now only one pin that must either cross between pins or route around the whole connector if it is located on the edge of a board.
On your extruder connector if someone uses IDC connectors that do not have the key and they connect it reversed they will short v_supply and ground.
Andre is it better to get power to ground, or power to signal. In either case a keyed connector should be mandated.
If it were me I would ground all unused pins for 2 reasons… 1 noise, 2 mechanical support. The ground pads have more metal connecting them to the PCB.
That’s not the worst of what could happen, and designing for that contingency would require some huge sacrifices. Keyed connectors need to be mandatory, and it might also be a good idea to recommend a circuit that will detect that short and cut the power before microcontroller pins can damaged by such a mis-wiring.
@Camerin_hahn but grounding the pins, or even connecting them to each other, limits what future purposes we can assign to those without risking damage from mixing old/new versions.
0 ohm resistors 
Fyi, I love 0 ohms if I am not hand populating
Why not make motor wires in row (2,4,6,8) so if worst case you can just plug a normal 4 pin into in?
Because interleaving the motor wires with the endstop wires would result in cross-talk, which would be likely to cause the endstops to be triggered erroneously.
What if they are not endstop wires, like ground or something else. Also, have we actually test this theory? Curious to see how bad or not it maybe.
Yes, if you used all of the odd-numbered wires for ground (wasting half of the cable), you could avoid the cross-talk problem. A better solution would be to use two conductors for each of the four motor wires (which would at least increase your current capacity), but either way, you’re making your cable and connectors unnecessarily large.
I’m starting to get curious on this. If the signal wire is in between “the loop”, wouldn’t that cancel things out? I don’t know details on this, but old parallel connected hard drives had big ribbon cables, so how did they avoid the crosstalk with such sensitive signals?
They did a couple of things to avoid crosstalk. First, they didn’t put pins with high, oscillating currents next to sensitive signal pins. Second, the read and write clock pins (the most sensitive to cross-talk) were isolated by putting a ground pin on either side of them. Third, all of the data adjacent data pins were switched at once, between clock pulses, so that when the capacitive coupling did create spikes in adjacent lines, it would not happen while the something was trying to read the signal on that pin.
Interesting, good info. It’s all about the details. I still want to measure it someday.
Yeah, measurement is great, but you need to design with the worst-case scenario in mind, not just the “this case” scenario.
I’m also wondering if there’s a “self-induction” effect of the coil pairs and if that causes moire pattern issues to be more pronounced…might be a good excuse to fire up those old oscilloscopes we have.
You could use a common mode choke to cut back noise… But that seems more expensive than needed. I don’t know what your max power switching frequency is, but I don’t think and act could do math and clock out high enough to make much noise… But anything is possible.
Another option is filtering on the lines noise sensitive lines.
Fyi. A spectrum analyzer would be better for peaking at noise. I doubt you would keep one hanging around.
So in summary for X, Y and Z we have a supply rail, end stop microswitch to Gnd or magnetic/inductive sensor with V+.
For the Extruder we have interleaved V+, possibility to switch on break out or on controller board + probe + thermistor. Fan switching on board with pwm and i2c with regulator on breakout, simetric for a hotend only without stepper.