Main Site Documentation

Wire Thickness Question


#1

What is the recommended wire gauge for UART lines? I am trying to pick out a ribbon cable that I can run all my COM lines, USB debug, and power lines through. One of my power lines is 24V so I need to have as thick of wires as possible while still maintaining signal integrity. I did see some ribbon cables that had a mix of signal and power lines but not for a reasonable price.

Depending on the cable thickness, I will run multiple 24V lines to distribute the current.


#2

I don’t recommend to use high voltages on the same connector as the signal-lines. If anything goes wrong, maybe the connector is reversed or a faulty cable is used) your whole thing will go up in smoke (if you’re unlucky also your debugger). Is there a good reason not to use two seperate connectors?


#3

We don’t want to use two connectors because this is going in mass production and that will add assembly time / increase the number of components. I am not worried about an assembly error as if we lose one unit out of hundreds it is of little significance. I am solely concerned about maintaining signal integrity while keeping large enough wire to run 24V through. My boss said instead of 2 connectors he would rather use ~4 of the wires in the ribbon cable to distribute the current drawn from the 24V. A lot of our design choices don’t make sense I just do what I am told.

So forgetting everything but the balance of wire gauge / signal integrity, what is the thickest wire that can be used on a data line? Specifically COM lines and the D+/- lines for debug.


#4

22AWG. Wire.


#5

Then you’ll need to do more than just worry about the choice of wire if you’re that concerned about signal integrity. Particularly with USB wires, that really are intended to be impedance matched. At a minimum, from info that’s been posted here before, you will need extra GND wires interspersed around the higher voltage lines and between the signal wires to provide them with a level of interference protection.


#6

What is the current drawn on the 24V line? That is the main question and then you can select the cable to suit. You can double up on wires to do this but also remember that the connector choice also dictates the current carrying capacity.

How long does the ribbon cable have to be?

Comms will be fine if you use RS232 or RS485 signal levels. With TTL you might run into issues.

USB, like Brett says, might be an issue over a long cable but adding a ground wire between, as he suggested might work.

You are going to have to do some experimentation and testing to make sure the USB works before you commit any design. Also make sure you have the USB tracks on your board run as differential pairs.


#7

Ya I am struggling on this part, what do you guys think of this set? This is biggest cable I could find with a matching set of connectors I could find (on DIGI-KEY). I could go with more wiires (13 will be occupied) but I didn’t really see much that wasn’t $100+. I am find with getting 5’ ribbon cables and cutting them. If this set isn’t could please throw some recommendations my way this has been a long search already.

https://www.digikey.com/products/en?keywords=AE14G-5-ND
https://www.digikey.com/products/en?keywords=1175-1647-ND
https://www.digikey.com/products/en?keywords=1175-1630-ND

Ribbon cable just has to be 6" or less. The signals are either 3.3V or 5V running through. The FT232 is running at 5V as well as my PIC that are sending the comms through the cable. Then also running 3.3V, 5V, 24V, GND, and USB debug through the lines.


#8

@Brett I am actually going to just move the USB debug circuit to the same board as the G30 so I will not need to run cables. However, what should the impedance on the traces be for D+ and D- from USB micro plug to the G30 as well as the D+ and D- lines going from a USB-A to and FT232? I have not seen anything in the data sheets about this. Neither set of USB lines will be going through the ribbon cables as of now to simply things, just need recommended trace width.


#9

Let me find out what the current is going to be. The entire device uses less than 1 A for all circuitry combined I’ll check how much is going through that part.

Should I put a ground wire between each COM line? Like:

COM1 TX
GND
COM1 RX
GND
COM 2 TX
GND
COM2 RX
GND
DETECT PIN1
GND
DETECT PIN 2
GND
GND
3.3V
24V
5V


#10

Hi,

I recently worked with a touchpad which connected with my GHI device through a very thing ribbon cable. The designers of the touchpad made their pinout so that the wiring would be like you mentioned: signal - gnd - signal - gnd - signal - gnd.

Then the version 2 came out and without changing the ribbon cable they removed all of the gnds except for one. The power signal was one edge of the ribbon cable, the ground was next to it, then every other wire was a signal. (They increased the types of busses available by removing the extra grounds).

So they must have determined the alternating gnd scheme was overkill. But they figured this out by trying it, not by theory. When electrical expertise is not available. Just try it! :slight_smile:


#11

just run it as a differential pair (which it is). That is the reason diff-pair trace routing exists


#12

If you want to apply reasonably good engineering practices here’s some suggestions. I size wires based on 1) the temperature rise due to current flowing through the wire 2) voltage drop (voltage drop = ohms/foot for your wire * current in the wire * round trip wire length 3) power/energy loss (power loss = current in the wire ^2 * ohms/foot * round trip wire length, energy loss is power loss * number of seconds you are actually running that current through the wire). You need to size the wire so the temperature rise is less than what the wire insulation is rated for. Teflon can go up to 100 degC or more, PVC is something like 85 degC. You need to check the spec on the wire you’re using. I don’t want anything that hot running around my system so I keep the temperature rise much lower. What an acceptable voltage drop is depends on a lot of things. Signals may meet the minimum voltage level at the TX side of the cable and drop below the minimum acceptable voltage on the RX side if there is too much voltage drop. Linear regulators, DC/DC converters, etc. also have minimum acceptable input voltages. Power/energy loss matters on a battery powered system. In most of my cases, this is the overriding concern. I don’t want to waste a lot of my battery resources heating up wires so I try to keep my power/energy losses less than 1% of the total energy in the battery pack.

Signal integrity is a whole separate and complex subject. One of the best thing you can do is use shielded twisted pairs. Shielding is expensive and not always necessary but twisted pairs can provide a controlled impedance path which is important for signal integrity. USB compliant cables are twisted to maintain roughly 100 ohm impedance. And you can get cables with multiple shielded and/or unshielded pairs of the same or different wire size.

Ribbon cables are the worst possible choice for signal integrity but obviously work in a lot of situations. The unfortunate thing about the signal integrity issue is if you just hack a system together, it may work in your factory but not in another, more noisy, factory and now you have unhappy customers.

Seems like a lot of stuff to worry about, right?. The good news is none of the above is hard to do and once you do it the first time, it becomes pretty routine.


#13

Thank you guys, great answers. The lines are just carrying motor commands which they did in the last model with a ribbon cable as well so hopefully I can replicate it. They used a separate connector for the 24V line but I am supposed to try one for the prototype PCB (left a hole for running a separate wire in case separating it with 2 GND wires doesn’t work or the wires melt… The largest wire I could find in a ribbon cable with matching connectors was 28 AWG (that wasn’t $50-$1000) so we will give that a shot - maybe things will get exciting and I will set something on fire again haha.


#14

100% with you there. We use shielded twisted pair cables (actually twisted quad configuration split into 2 pairs) in an ROV umbilical and run RS485 over 1500 - 2000M without an issue and that is bundled in with 3000Vac and 1000Vac wiring for the electrical power.

I saw this with an original design that used a short 15cm ribbon for an LCD interface from the main board to the LCD. It was so troublesome and you got flicker on the LCD. I later reworked the board to use an FPC cable and the difference was such that the display has been super stable since and I could use a longer cable too.


#15

What is FPC? I made a couple variations on the PCB to try different methods just in case. Arranged everything so my display, G30, EEPROM, etc are on the same board, I just have COM lines for my FT232 and power needing to go through the cables.


#16

#17

:joy: I’ve never seen LMGTFY before. Snarky, but hilarious