# Running large current through vias: large vias or many small vias?

• posted

Hi there - I'm working on the layout of a PCB that will be carrying ~30A or so. I'm running some parallel traces to increase the current carrying capacity of the PCB. I would like these traces to be connected with vias. My question is this: Is it better to have lots of really small vias, or just a few really large vias? Also, should these traces be connected just at either end of the trace, or should they have vias connecting them throughout the trace?

Thanks!

-Michael

Loading thread data ...
• posted

A large via will have quite a big surface are but so will many small ones. You can use either approache.

Personally I would go for one large via so there arent so many tracks running around to confuse matters.

• posted

You can do relately simple resistance estimations on printed wiring traces.

Copper Resistance: Rc= r . L/HW

r = 1.7241WcmE-6 or 6.786WinE-7

Tempco Tc= +.039%/°C

R = K L/W

K @ 1oz (H=.0014 or .035mm) = 4.84E-4 @ 2oz (H=.0028 or .070mm) = 2.43E-4 @ 3oz (H=.0042 or .105mm) = 1.61E-4 @ 4oz (H=.0056 or .140mm) = 1.21E-4 @ foil .01 = 6.79E-5 @ foil .02 = 3.39E-5 @ foil .05 = 1.37E-5 @ bar .10 = 6.8E-6

Note: K is traditionally expressed as ohms/square, as L/W of the 'printed' square is 1. The resistivity of 1oz copper can therefore be considered as 0.48 milliohms/sq.

L/W 2oz 3oz 4oz

1 .243 .16 .121 mohms 10 2.43 1.61 1.21 mohms 100 24.3 16.1 12.1 mohms

- and a 1oz Plated-Through-Hole .062L .0017H

dia. 0.02 W=0.0613 R=0.48mohms dia. 0.04 W=0.1257 R=0.24mohms dia. 0.06 W=0.1885 R=0.159mohms neglecting solder fill.

Individual squares (with individual L/W ratios) along the length of a complicated trace-width are simply additive, like series resistors would be. Parallel traces can be treated as parallel resistors.

Self-heating of the trace and the surface temperatue rise resulting, in free air, can still be first order approximated as 1deg rise for every mW dissipated from a cm^2 surface area.

So depending on what you're trying to do, there's no reason why you shouldn't be able to check for effectiveness, potential hotspots, efficiency effects and overkill.

RL

• posted

Or simply 500 uohms/square for 1 oz copper.

I think it's 3900 ppm/k, which is +0.39%

John

• posted

If there are AC currents involved, multiple vias works better and you should join your traces here and there along the length.

With DC currents, conducting area is about all that matters.

A cute trick is to put a through hold test point pin in the design to do the via. The soldered pin lowers the resistance of the hole.

• posted

I wouldn't put vias "here and there along the length" as the hole reduces the width of the trace. Along the line of the test point idea, you can use copper rivets. There are hobbiest size rivets that work nicely.

• posted

Last time I did that the military disallowed it... "solder joint not inspectable".

...Jim Thompson

```--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |```
• posted

You're right.

Those 'W's are also supposed to be ohm signs. Corrected elsewhere but missed here.

• posted

The hole needn't reduce the width. Normally, a trace bulges out where the via is placed.

I wouldn't go that way. This would be a lot bigger than the test point but may not provide as good of a connection. You cant see into the gap between the pad and the rivet.

• posted

If you're running a wide trace, you aren't going to see any bulge unless you purposely make an oversize pad to suit the situation. Staking along the run of a trace doesn't do anything electrically.

No worse than using a swage terminal post. Lots of power supplies use them.

• posted

Yes so you make the pad.

Yes it does in the AC case. That was where I suggested it.

I don't use those either. I only use things that can be inspected. I also think you are wrong in suggesting it is no worse than the swage terminals. Remember that those terminals were designed for the purpose by people who hopefully knew what they were doing. The rivet is not intended for the job.

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.