We’re attempting to use a PCB as a sort of wire harness to connect a set of 15 LED strips in parallel over 23". Circuit is 12V and 3.6 amps, so 1 oz copper trace width needs to be over .070" according to calculators.
I assume as large of zones as possible should be used, see attached top copper layout where half of the front and back is used for each of the positive and negative circuits with thermal relief around the pads to be able to solder. Or should each copper layer be devoted to one of the circuits even though all soldering will be done on the top?
Is there a best practice for how to layout this PCB? And do the standard thermal relief calculator values apply for higher currents?
Thank you very much for any advice to someone new to all of this!
Is that 3.6A total, or 3.6A for each of the 15 LED strips (54A total)?
PCB trace width calculators are often based around a guesstimate for a 10 degree Celcius (F?) temperature increase and not about voltage drop.
With such a long PCB the voltage loss may become an concern.
How critical are the locations of your mounting holes? You are creating thin spots on half the mounting holes with the Zig-Zag line. It would be better if you can put all the mountingholes in the “wide” parts of the Zig-Zag.
It does not matter much wether you put both conductors on both sides, or reserve each side for a single trace. The main difference is the clearance between the traces. If there is only one trace on each side, the Zig-Zag line is gone and you have more copper left.
A simple trick is to make the power connection in the middle instead of one of the ends. This halves the maximum current through the PCB.
I do see the voltage drop could be about .5V more than say, 18awg copper wire.
The 30 smaller pads’ locations that connect to the strips need to be as above. The Power actually does connect to the larger thermal relieved pads in the middle, see image below. Does that mean I could use half the current and distance in the calculators for trace width or voltage drop?
It does sound like it would be best to split the board in half by top and bottom, if the thinner areas caused by the zigzag gap could be problematic.
Yep. You’ve got 1.8A going to the left, and 1.8A going to the right, and both have at most 30cm before they reach ther LED strip.
Both currents diminish quickly because each LED strip takes it’s own current.
But you loose a bit of voltage over both the “+12V” track and the -12V track (Euhm, that’s a 24V difference ???)
I grew up in a metric country.
Your board seems about 60cm long and 3 cm wide. Your 0.07" trace translates to 1.8mm, which feels about right for 2A. Your actual trace is however closer to 10mm, more than 5x what you need. There does not seem any reason for concern for the current handling of the traces themself.
Brightness differences in the LED’s is harder to predict. LED strips usually have 3 or 4 LED’s in series, which need 8V to 10V to even start lighting up. Which means, if you start with 12V and you loose 200mV in the +12V trace and 200mV in the “-12V” (=GND?) trace, the last LED strip only sees 11.6V, and you will loose (0.2+0.2)/2 = 20% of your LED brightness. (Brightness is aprox. lineair with current).
You will have least loss in the strip if the copper widht is uniform, but the differences between a straight line or Zig-Zag are probably minimal.
I thought you previously suggested using the entire TOP layer for 12V and the Bottom layer for the return path.
That will give you approx 10% more copper (and thus less resistance) because the empty zone of the Zig-Zag separation is now filled with Copper.
Another issue to watch out for are short circuits. I do not know how you want to mount your strip, but solder mask is not guaranteed to be an isolating layer.
Some more notes:
Make sure the holes are big enough to fit your wires.
Soldering wires directly to a PCB is not the best, expecially for flexible wires, and if they can bend.
The wires tend to bend and break right behind the spot were the solder ends. Tying the wires with for example a ty-rap to the PCB prevents the wire from bending (and breaking) right after the solder joint.
Wow, thank you for all the info, Paul! And also for bearing with my ignorance!
Yes, board is approx 60cm x 3cm. I have read that you want as full of a plane as possible for power and ground, and here that is all there is - so yes it should be more than enough for the 3.6A total. I’ll post a screenshot when I get that top vs bottom setup.
We’ll have to see what happens with a first printed test batch. We didn’t notice a difference in brightness among or within LED strips when working with milled PCB prototypes.
The strips are mounted to a separate part, that the PCB is also mounted to with nylon locking teardrop supports that raise it up. So both are static that the wires shouldn’t want to move once in place. I’ll look into the Ty-Rap though, especially for the power wires.
So these (long?) LED strips have wire leads, and come ‘out of the page’ vertical to the board ?
How much movement/stress is on those leads
Interesting question. Going top/bottom saves the gap, so gives slightly more copper, but does then rely on the plating to get current from the rear plane.
Yes, and with wires-into-board, also be wary of stress on the hole plating.
Higher current loose wires, not soldered on the current-side, rely on the plating, and we’ve seen that fail due to flexing and stress.
If you can tie-down the wires that helps, or solder both sides (but people forget can to do that…) or add redundant vias (which you may have in this case on the zig-zag design)
You could move those thermals, so they do not neck-down the copper as much - ie you have placed then almost where the copper is most narrow, a nudge to where copper is widest gives least-milliohms.
Correct, short jumper wires approach vertically through the board that should have little stress and no movement as the LED strips and mounted PCB are both locked in place.
Exactly, couldn’t find which is preferable. If the hole is sized right, shouldn’t some solder be pulled through to other side - I guess we might need to leave some pad masked off on the back then.
What layout of vias might you suggest for this case with the split zigzag layout?
Got it, I can move those anywhere I like - that makes sense.
The zig-zag design already has redundant thru holes, as I presume you clone F.Cu and B.Cu the same.
Sounds a good idea.
Where high currents and large pads are involved, we often use oval pads, rather than round, as the soldering iron can better fit and transfer more heat, and you get more copper adhesion.
Offset ovals can give the best solder results.
In some cases we have made the copper pad larger than the solder mask, for even better adhesion on parts that might be whacked during their life.
That said, 3.6A total current is not especially high.
jumper wires can be a pain to handle, if the LEDs do not already have bare wire leads, but do have solder pads, you could consider some sort of staked connector pin ?
(unless you need a bend in the wire link for stress relief)
If you can do without the thermal via’s then the solid copper connection of the hole to the copper planes makes them a lot stronger.
Another trick sometimes used is to use 2 holes to connect a wire.
You stick the wire complete with insulation through the first hole, and then solder it in the second hole. This ensures the wire does not get bend at the transistion area of the copper, which is the weakest.
Yet another option is to not use holes to solder the wire in, but a rectangular pad. The pad should have some (4 or 6) via’s for mechanical strength. I do not know if this is applicable in your situation.
I see, yes - they are the same. That redundancy in the split-halves option might be worth giving up a little bit of copper.
Thank you so much @PCB_Wiz and @paulvdh for your time and advice - it is very appreciated! We’ll do some testing of those pad and wiring options and find out what works best for our situation.
