PCB design: High current tracks + low voltage measurement (Please review)

[superKris]

Hi all,

I have been working in a complicated design for some time now. Its finally time to make the new PCB which i hope is the final version. Before i continue working on the design i would love your feedback about the PCB trace layout. I tried to include all critical info in the pictures below.

This part of the PCB can be considered a 12V automotive fusebox. The power comes in from the main terminal to the righ. Runs trough a shunt, trough a relay, and to different outputs. There is also an unswitched output and 2 inputs both with a shunt. The other large resistors and TVS diodes are there to suppress voltages spikes like automotive load dumps.

I have a couple of concerns.
- Will the 40A traces be able to handle this current with 2oz copper?
- The shunt voltage will be VERY weak. I want to measure voltages between 50uV and 50mV. Will the high current traces induct noticeable voltage?
- I used vias to route current to the shunts on the bottom of the board. What is a sensible via size and setup for this?
- is 0,3mm distance enough between all tracks?

Please keep in mind this is a work in process. Finetuning is needed. There also is no ground yet, but bottom will be used for this.

I would really appreciate you feedback!

[daqq]

- Will the 40A traces be able to handle this current with 2oz copper?

Download the Saturn PCB toolkit software, there's a lot of useful calculators. Your proposed design will (10mm track width from both sides, 2 oz copper) will be able to handle the current.

- is 0,3mm distance enough between all tracks?

At 12V yes.

Other comments:
The way you are routing your shunt isn't exactly proper at these currents, since you are using two in parallel and are tapping only one. There might be some errors introduced with this. I don't suppose you could use a single shunt?

Beware of PCB bending - when tightening your high current screw terminals, there may be some bending/flexing of the PCB and this may lead to a cracking of the fairly long 2512 shunt since it's pretty close to the terminals. Try it first before going full production.

[superKris]

Thanks for your reply!

- Will the 40A traces be able to handle this current with 2oz copper?

Download the Saturn PCB toolkit software, there's a lot of useful calculators. Your proposed design will (10mm track width from both sides, 2 oz copper) will be able to handle the current.

Thank you! That looks like a very use full tool. I downloaded it, but it gives lower max current than other tools i used. Other tools predict a trace width of around 25mm (45 degC, rise 10) i'm currently at 10mm(x2), but the tracks are short. and a little higher temperature is acceptable i guess.

Other comments:
The way you are routing your shunt isn't exactly proper at these currents, since you are using two in parallel and are tapping only one. There might be some errors introduced with this. I don't suppose you could use a single shunt?

Beware of PCB bending - when tightening your high current screw terminals, there may be some bending/flexing of the PCB and this may lead to a cracking of the fairly long 2512 shunt since it's pretty close to the terminals. Try it first before going full production.

A single shunt will be hard to do if want to keep using "cheap" SMD versions. Assuming current will flow trough both of them equally it will be 20A each. I choose a very low value of 2mOhm. This will result in a dissipation of approx 0,8W per shunt resistor. I would prefer to not dissipate to much power. Additionally i think 40A is lot of current to pass trough such a small SMD connection on the 2512 shunt. It seems to make more sense so split this current.

I tried to keep both paths trough the shunts at the same length hoping that the current trough both of them will be the same. If you or anyone else has any info to confirm if this is true or not, i would love to know! I have done much research on this but to me the answer inst that clear.

I would be able to add a additional shunt measurement as i have 1 input on the ACD free, but this would add costs, and i could really use all PCB space possible at this location on the PCB.

This is the shunt resistor im planning to use https://lcsc.com/product-detail/Low-Resistors-Current-Sense-Resistors-Surface-Mount_RALEC-LR2512-22R002F4_C108929.html

I dont think PCB bending because of the terminals will be a issue. I'm using cage lift terminals.

[bob91343]

Having had to repair poorly designed high current paths on PCBs, I can say that it's a can of worms.  The biggest issue is entry of the current.  Some boards use eyelets and some solder to pads or use connectors.  Each of those solutions has traps.  The current distribution at point of entry is the problem.  Hot spots due to nonuniform current will cause lifting of traces, corrosion of contacts, etc.  Once they begin, they usually get worse due to the heat.  Generally there are dissimilar metals involved as well, with thermal interfaces.

You won't accept my advice but here it is anyway.  Do not use PCBs to carry large currents.

[superKris]

Having had to repair poorly designed high current paths on PCBs, I can say that it's a can of worms.  The biggest issue is entry of the current.  Some boards use eyelets and some solder to pads or use connectors.  Each of those solutions has traps.  The current distribution at point of entry is the problem.  Hot spots due to nonuniform current will cause lifting of traces, corrosion of contacts, etc.  Once they begin, they usually get worse due to the heat.  Generally there are dissimilar metals involved as well, with thermal interfaces.

You won't accept my advice but here it is anyway.  Do not use PCBs to carry large currents.

You are absolutely right, but without high current paths my whole project is pointless. I am however looking for tips to minimize the problem. The inlet terminal is up to the task. It has very thick double pins and is rated 80A or so. The relay socket im not that sure about but its rated 40A. Pins could be a little thicker for my taste, and double pins would be nice. Hopefully a shitload of solder will help.

Your tips on what can be improved are however very welcome. Things like the use of vias, track width, corner shapes, etc.

On a side note: It will probably run between 10 and max 20 amps most of the time.

[superKris]

So i decided to add an additional current sense amplifier. I have space on the ACD, the cost are not that high, and it adds accuracy.

I am however still looking for advice on the placement of the high current vs super low voltage sense tracks, and the usage of vias.