high current traces and thermal relief

[djacobow]

Designing a board that will be hand assembled.

It is a 12V power switching board and will see currents as high as 30A. Frequency is very low (quasi static, so no dynamic issues).

I'm using big polygon floods for the high current traces. I'm not certain they'll be enough, but the traces are short, so <shrug> and hope for the best. Calculators suggest it's iffy.

I'm debating whether I should enable thermal relief on the pins on the high current traces. On the one hand, IR heating in the spoke, on the other hand, a huge pain to solder.

What's the received wisdom?

If it were wave soldered, this would be an easy decision. I have another version of this board that I have not tested at 30A, but it did have the direct connections, and it was indeed very difficult to solder the high current pins.

Picture attached. Q201 and Q202 are low Rdson Pfets. F2 is an automotive type fuse holder. J1 is vertically stacked Anderson Power Pole connectors. The drawing is a bit weird, but the long slots are the connections for J1.

Thanks in advance!
-- dave j

[nctnico]

I'd leave out the thermal reliefs and use hot air to aid soldering.

[T3sl4co1l]

Use relief.

More detail:

What, you haven't calculated it yourself?  (Do you know how? -- Do you know what a trace calculator is doing, how it works? -- Baby steps, just in case!)

The resistivity of copper is a bit under 20 nano ohm meters.

What does that mean?

When you have a material property, it's usually got a power of 1 or 3 (or -1 or -3) length.  This is times meters, so, +1 power.  When it's 1, you multiply by the ratio of length and cross-sectional area; when it's -3, you multiply by volume (which is length times cross-sectional area).  The latter does not depend on direction, a volume is a volume.  The former does: it matters which way the length is, and the cross section must be perpendicular to it.

What's a thermal relief?  Four (usually) spokes of copper.  What's the length?  The length is in the direction of current flow.  So, radial (outward/inward from the pin).  That means the cross section is (spoke width) * (plating thickness).

For a typical TO-220 pad, I would use 10 mil clearance and 20 mil spoke width.  The spoke length is 0.25mm, and the total spoke width is 2mm (more or less).  If this is 2oz copper, that's 0.07mm thickness.

So we have:
20n ohm.m * (0.25 mm) / (2mm * 0.07mm)

Once you have the circuit resistance (i.e., ohms and no meters -- don't forget to factor out the milli's and nano, BTW), you can calculate the power very easily too.

You can also calculate the temp rise, using a similar calculation, starting with the thermal conductivity (400 W / (m*K)).  You will find that the maximum current handling is...rather stupendous!

This does assume that the resistance of the pad ring, and of the polygon surrounding the spokes, is much smaller.  In practice, the spokes will look longer, because it takes some distance for the current to diffuse out into the polygon.  The polygon also isn't infinite, nor is its heatsinking value infinite.

What helps the most for reliably handling high currents, is making sure all the traces are well heatsinked.  Copper carries heat very well.  Indeed, it can be valuable to use a 4-layer board just because the inner layers spread out heat from the outer traces (you also get more copper period).

Tim

[nctnico]

You'll also want to take the heat out of the contacts which is what thermal reliefs prevent.

[ahbushnell]

What weight copper are you using.  4 OZ?  Normal boards are 1 OZ. 

Andy

[Monkeh]

You'll also want to take the heat out of the contacts which is what thermal reliefs prevent.

That's what the wire's for.

[digsys]

I do heaps of 20A-30A boards. I use 4oz, top and bottom, with ~80% of the tracks center non-masked. If the track is longish, or GND, I leave provision for heavy links
down the center - along the exposed section. Naturally, I make them as wide as feasible. I also make control test strips to verify the PCB batch is ok. = worthwhile.
We have JBC HD irons with a range of tips, some quite huge. I can solder a transformer clamp to a copper rail in seconds :-)
All depends on your tools ! According to the FLIR, nothing breaks a sweat :-)

[djacobow]

Thanks for the answers, guys.

I have a test board right now that is 2 oz copper and it handles 20A without a noticeable temperature rise using the highly-accurate touch-it test. This board does not have thermal relief.

I will also do as Teslacoil suggests and take the calculation approach more seriously than I have done so far.

I don't want to go to 4 oz if I can avoid it, but I don't think I will have to.

I have thought about leaving the traces unmasked and added solder and wire, and I seem to remember Dave did a video on it, and there is some benefit. But it seems like I won't need that. I guess leaving the mask off is cheap insurance.

[SeanB]

No relief, and have a second dedicated high power soldering station, with a big hoof of a tip, to do the relay contact soldering. You will need it in any case for the power connections, as these need a big tip and plenty of heat capacity to solder them fast.

please also do yourself a favour and use a modified TO220 pin layout, with the centre pin offset to the others, so there is the ability to have a larger ring on the pad as well so you do not just blow the trace off there. As well use a heatsink that has solderable tails to hold to the board, so you can use them as an extra current carrier to the bare drain connection. Yes they will be at drain voltage, and thus will need slightly more spacing, but that will also keep them cooler.