search

fellesverkstedet / fabricatable-machines

Motion systems for flexible digital fabrication and research. Easy to fabricate and customize. Documentation: https://github.com/fellesverkstedet/fabricatable-machines/wiki
163 stars 35 forks source link

Faster electronics wiring with next gen HRBL shield #21

Closed JensDyvik closed 5 years ago

JensDyvik commented 5 years ago

Our HRBL shield has helped speed up the process of wiring motors and control boards. But i think there are financial and accessibility benefits from making it even faster. Our current luxury socket plugs are costly and the 0.5mm2 8pin ribbon cable is not easy to source.

I propose a shift to a common IDC 16pin 1.27mm pitch 0.127mm2 cable type with common female 16pin 2.54mm pitch IDC connectors. I also propose the following pinout (this pinout will not short if the cable or connector is flipped during assembly):

hrbl-connector-proposal-with-orientation-check

As we move towards providing ready made motor shields for integrated motors in combination HRBL shields we can have IDC connectors on both sides, bringing us quite close to plug and play motion control.

Advantages over current connectors and cables:

Disadvantages:

Jaknil commented 5 years ago

Interesting suggestion! I really like the idea of using the IDC to save time connecting things and I agree that the price of the connectors we use today is annoyingly high.

Some wire calculations: Your link says IDC 16pin 1.27mm pitch 0.127mm2 cable, this is not correct. The 0,127mm is the diameter of each copper strand in each wire. The copper cross section diameter of the linked ribbon cable is 0,081mm^2 (correct for AWG28).

I find that using the AWG values is the best way to get the actual mm^2 copper in the wire. Just a reminder that a smaller AWG = thicker cable.)

I see that the total copper area when using 4 of AWG28 (0,081mm^2) wires for each polarity is tot 0,32 mm^2, this can be compared to the AWG20 (rainbow ribbon) we used before, it has 0,52mm^2/wire. So even using 8 of the 16 wires for power delivery gets us only to 63% of the copper area we had with the rainbow cables.

Funnily enough it might not matter so much. The stepper driver chops the voltage until it gets the current it wants through the motor coils, so it will not "notice" the resistance in the wires. As long as the wires don't heat up too much or we choose a too small power supply it might be just fine using those wires. Regarding wire heating, they power wires are on the sides, that's good, and it's 3W of heat in total over a 1 meter ribbon and that sounds like very little. So in conclusion I think it will work. But we will loose ca 12W in heat to all the ribbons. It is unconventional, which might lead to people not trusting it, but it should work and would be a nice all in one package.

Other research: I also checked the 16pin connector and it should be able to take the current as well.

I tried to research a 10 pin IDC connector that was large enough to take this 20kr/m AWG20 10 pin ribbon cable https://no.farnell.com/pro-power/pp001508/ribbon-cable-10core-20awg-300v/dp/2776633 But I could not find such a connector, for that cable I think we are best of with the ones we have used before. And I agree that these ribbon AWG20 cables should be considered exotic.

However, I would like to raise an alternative that we can compare your suggestion to. Call it the conventional way with a separate power cable and a signal ribbon:

Power / Signal separate 8pin Signal AWG28 Ribbon and IDC connector (costs similar to the 16pin version) https://no.farnell.com/amphenol/t812108a100ceu/socket-idc-2-54mm-8way/dp/2215231 https://no.farnell.com/3m/3365-08-100/ribbon-cable-8-way-30-5m/dp/2064465 image

2pin Power AWG18-26 Cable and IDC connector (cost very little and we need them for the PSU to board connection anyway) https://no.farnell.com/amp-te-connectivity/3-640426-2/housing-18awg-2way/dp/1580081 https://no.farnell.com/amp-te-connectivity/640445-2/header-vertical-0-156-2way/dp/589044?MER=sy-me-pd-mi-acce https://no.farnell.com/jsc-wire-cable/2440-0100s-b-r/no-of-conductors-2conductor/dp/2836435 image image image

Comparison

Price-wise they are similar and inexpensive: (Circa prices)

What price unit Quantity Tot
16 core 8,97 kr/M 4 35,86666667
16 pin connector 4,5 kr st 8 36
TOT 16Pin       71,86666667 kr
8 pin connect 3 kr st 4 12
8 core wire 7,4 kr/M 4 29,6
2 pin connector 1,48 st 8 11,84
2 core wire 2,51 kr/m 4 10,04
TOT 8+2pin       63,48 kr

The main difference in the end is that the user needs to route either a 16 wire cable and 2 connectors or a 8 and 2 wire with total 4 connectors which I feel is kind of similar. Otherwise I think they both fall under the same pros and cons as above.

When used with plain integrated motors the 8+2pin alternative will make it nearly impossible to mix up signal and power and is much easier to connect compared to bunching up 4 wires into the screw connector. That makes me want to go for that option for the cards we order now, and then reevaluate later when we have more progress on motor mounted shields.

JensDyvik commented 5 years ago

I agree that 8+2pin is the better option for now. Fast and safe wiring on the driver side + safer power lines (larger cross section ) are good arguments.

The cnc milling version of the board could maybe have power and signal connectors on opposing sides of the board. This will reduce the amount of vias needing manual fabrication.

On a side note, it could be interesting to include Tx and Rx pins in the wire harness in the future. Octoprint has recently gotten support for action commands: http://docs.octoprint.org/en/master/features/action_commands.html

With futere shields for intergrated steppers we could highjack the serial communication to send messages that will stop the machine and trigger popup dialogues like: "Heavy vibration on X axis. Please check milling bit for wear, your settings or fixturing before continuing" or "X motor is not able to move freely, click here for troubleshooting help"

Jaknil commented 5 years ago

Review with Jon: