We want to connect two relay boards to a front-panel board. Each will need 64 connectons at, ideally, 5 amps each. If (heaven forbid) all run at 5 amps simultaneously, that's 320 amps. On each side of the front panel board!
TE claims a standard 100 mil header pair, with 25 mil square posts, is good for 5 amps.
We figure that a lot of copper, like 2 oz pours instead of traces, will help keep pins cool. And we can put a pair of ground pins every
3rd pair, so no pin has hot neighbors on both sides.
Manufacturing would like thermal spokes on all the pads, and won't get any.
That is if all the circuits/relays are on. Your power supply is unlikely to support that much current anyway.
On a similar issue, but smaller scale, I am working on redoing a design with doubling current cap by external mosfet switch, or OOB (Out Of Band) current/power. The current implementation derates the design by 50%, due to small PCB and SMT devices. I am replacing some critical path components with THT (Through Hole Tech.) to handle 3x to 4x more current. I already increased the PCB area by 4x, but there are still critical current paths. There is only limited trace width i can bring in/out of the chips, without going to expensive multi-layers or heavy copper dosing.
I am thinking about unmasking some traces, and let solder to fill up the center, and thus increasing the current flow. Namely, a 20 mils mask opening for a 60 mils trace. So, a "half-naked" trace is defined as a trace with two masked trace on either side, but unmasked in the middle. It needs to handle 50% pulsed 5A to 10A.
BTW, part cost alone is more than the existing retail price. All the PCB mods and assembly costs are extra. Don't know how they can sell for so cheap, even if they sell millions.
I wonder if some special silver based solder would help.
But for some SMT parts, i am restricted to 20 or 30 mils, or even smaller.
Price is not really a problem, but wondering how they can sell below cost. I figure that even if i order 10,000 life time supply, parts alone would still be more expensive than buying the completed module. Although it's power corner cutting module.
but for how long? the limit is mostly a limit on temperature rise and a short thin trace connected to long fat trace isn't going to be as hot as a long thin trace
Sot25 hooking up to couple of vias 50 to 100 mils away. The top is all full. The bottom has more room for fat traces. I can probably feed some copper wires and solder directly to the pins. I changed the heavy duty inductors and capacitors from SMT to THT. A 1M resistor across 200V should probably be 1/4W THT, not an SMT device.
Can you bring in power by different means? I've done high current work like this before with a power coming in on a high current contact mechanically/electrically fastened to a bus bar which in turn was bolted to a pad on the PCB.
My notes say things like:
"9.2 Amps/via for a 20 mil via with a 25C temperature rise"
"11 vias/bus bar connector * 9.2 Amps/via = 101.2 Amps / bus bar connector pad."
Looks like I had plenty of room for more vias, so I doubled the amount I needed to meet the spec so I would have plenty of margin.
But damage is a function of total heat capability. As John said, solder might not be a good conductor, but still a good heat absorber. Large amount of solder on "naked trace" can still protect the trace. The solder will vaporize before the copper.
Probably not relevant, though; the heat isn't shed faster because of a solder layer, so temperature will rise until it IS shed fast enough, or until the copper delaminates and the trace, solder and all, melts at the least-heat-sunk portion.
If your relays handle lots of current, you should expect lots of power and prepare for it. Old RETMA racks had steel for heatsinking, and more modern (telecom?) relay boards have been made in porcelain-on-steel (thick film wiring) for the same reason. One hopes these are lighter relays, and the '5A' is more of a limit than an expected operating condition, but... if there's dozens of 5A paths, you might want to connect relays with fusible links (fusible resistors, and/or wire-wrap?) or other non-printed wiring, rather than trying to bake relays onto surface-mount pads and rely on two-ounce printed wire.
Damage is a function of temperature. I assume that this will operate steady-state, so power matters, not energy.
I don't think a sliver of solder would help. What I need is wide traces, 2 oz copper, adjacent copper planes, and a bit of air flow. 5 amps isn't a lot of current per conductor, but 60 of them close together could get warm.
Two sources now suggest that a standard dual-row 100 mil box connector pair, with 25 mil square posts, is OK at 5 or 6 amps, but a lot of them adjacent will warm things up. Gold plating is best.
No, the current circuits are isolated. The ground pin function is purely thermal; the pins would be soldered to some fat internal ground pour layers to put some cool spots between hot pins.
Single connector pins can carry a lot more current than pins with current-carrying neighbors. They heat one another.
The box here has two 61-pin connectors, cable in and out. It's a fault generator; it can break any through-path (simulate a cable open wire) and can short any pin in the cable to any other pin.
For my circuit, the issue is for the protection of 20A inrush. Steady state up to 5A is probably not a problem. I might end up with some empty header pins as heat sink.
And BTW, on the topic of PCB. I know it's bad for over-lapping drill holes. How about touching or almost touching holes? What is the minimum gap between two holes in general.
We use plated slots for things like fastons and the blade pins on barrel connectors. And pave them over with layers of copper for high current... no thermal spokes.
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.