SparkFun Forums

[geome]

I'm currently designing a power distribution board, and certain traces need to be able to handle 5-15A of current. I'm not especially knowledgeable in the area of trace widths. What thickness/width should I be considering for this amount of current?

Edit: Should mention that I plan on having these lines on an external layer, although info for if they end up being internal is nice also. Thanks!

[Blackfin]

Try checking out an online trace width calculator such as:

http://circuitcalculator.com/wordpress/ ... alculator/

[geome]

I did try that, but there are some things that bothered me. Some people I work with have a board that is rated at 30 amps, but there aren't any traces that are ~430 mil as the calculator suggests would be necessary for that (at a fairly standard 3 oz/ft^2 thickness). Is the issue that I need to consider a much thicker sort of trace (beyond that of standard fab house specs) for this?

That being said, I don't suppose ~165 mil is too big to work with, I just want to make sure I'm in the right range to do it properly. (Feel better having some expert input rather than just the calculator to go on, that is.)

[NleahciM]

I did try that, but there are some things that bothered me. Some people I work with have a board that is rated at 30 amps, but there aren't any traces that are ~430 mil as the calculator suggests would be necessary for that (at a fairly standard 3 oz/ft^2 thickness). Is the issue that I need to consider a much thicker sort of trace (beyond that of standard fab house specs) for this?

That being said, I don't suppose ~165 mil is too big to work with, I just want to make sure I'm in the right range to do it properly. (Feel better having some expert input rather than just the calculator to go on, that is.)

3oz copper is far from standard for most PCBs. That's very heavy.

Remember that these calculators are choosing a current based on a set temperature rise. So perhaps the 30A board that you are referencing is actually designed to have a much larger temperature rise.

[geome]

I kind of meant standard in a "most fab houses seem to support it" sense.

I don't want a massive temperature rise, because minimal power loss is pretty important. According the the calculator I'd be looking at ~4.2mm trace for 15 amps with a 20 degree max temperature rise (at 3 oz/ft^2 thickness). Does that seem reasonable/practical?

[slagment]

Not best practices, but:

What about multiple traces, or pours?

You could do half the trace width on top and half on the bottom.

[geome]

The problem with that, slag, is I plan on measuring measuring current through the trace, also. Multiple traces complicates that. =/

[NleahciM]

The problem with that, slag, is I plan on measuring measuring current through the trace, also. Multiple traces complicates that. =/

Are you planning on using a shunt resistor or using the actual trace as your resistor?

[n1ist]

Don't forget to take the trace temperature into account. Copper resistance is dependent on temp; it will go up at 0.393% per degree C.
/mike

[geome]

The problem with that, slag, is I plan on measuring measuring current through the trace, also. Multiple traces complicates that. =/

Are you planning on using a shunt resistor or using the actual trace as your resistor?

Using a low resistance current sense resister.

[n1ist]

That should work better. I'd use a Kelvin (4-wire) connection to the resistor; that way, the sense part of the circuit won't be affected by voltage drop in the force traces.
http://websrv.mece.ualberta.ca/electrow ... _Connected
/mike

[george graves]

Well, correct me if I'm wrong, but your traces should be kept short as possible - and have as low of resistance as you can. Your shunt should be doing all the voltage dropping.

Anyways - why not pull the solder mask on those traces, and then tin them with solder by hand? A nice thing 2 mm coating of solder would handle a lot. Is that not good to do?

[leon_heller]

When I needed high-current tracks on a home-made PCB I soldered thick copper wire on top of normal size tracks. It worked very well.