Maker Pro
Maker Pro

PCB track width and high current with Mosfets

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Having designed the PCB for my LiPo over-discharge protection circuit, it struck me that 40/80A is going to need quite a significant PCB trace to carry that much current. I checked on various trace width calculation sites and all came up with traces several centimetres wide, which does not seem unreasonable. HOWEVER, the Mosfet has a tiny leg that gets soldered into the track, that nowhere near matches the required trace thickness. It seems I am missing something here; why does the PCB trace need to be so much 'thicker' than the Mosfet leg?
I have now designed a PCB and kept the track to an absolute minimum length (5mm), but still envisage it going up in smoke.
What am I missing?
 

chopnhack

Apr 28, 2014
1,576
Joined
Apr 28, 2014
Messages
1,576
I ran it through the circuitcalculator and you could get down to a 20mm trace if you use a 4 oz. board. For the price of the board though...eep! Good question re: mosfet leg area. Another method that you could consider it over tinning the trace, I have read in a few places where this significantly increases the load handling ability. HTH
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
There ate a number of things to consider.

Firstly, the track width will be for a given temperature rise. It may be perfectly fine to have localised areas where that temperature is a little higher.

Secondly, copper is a very good thermal conductor, even if heat is being added at a point (where the current density is the greatest) then that heat will rapidly flow to areas of the copper that may not be carrying much current a all.

Thirdly, the package itself may become a heatsink for the copper in extreme cases.

Copper is a better thermal and electrical conductor than solder, do the closer tout can get copper to copper the better. In this respect, surface mounting may have a slight theoretical edge. However it is also true that filling vias with solder reduces their resistance and would thus increase their current carrying capability. You also see solder on traces for the same reason. One of the easiest ways of doubling the current varying capacity is to use both sides of the board though.
 

OLIVE2222

Oct 2, 2011
690
Joined
Oct 2, 2011
Messages
690
If the boards are assembled professionally you can remove the solder mask and add solder paste on the track. This is the industrialized equivalent of the over tinning method suggested by chopnhack.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
This is a homebrew board and as the trace length is only 5mm I can quite easily just solder a wire over it.

What is not clear is how the LEGS of the Mosfet copes with the rated continuous current of 40/80A as there is not much I can do about that (except try and conduct the heat away through the traces, but even that is limited by the connection space around the legs.
 

OLIVE2222

Oct 2, 2011
690
Joined
Oct 2, 2011
Messages
690
Assuming a D2PAK the leg cross section is around 0.5x1.2mm not that bad knowing the very short length and compared to a few mm wide PCB track who is very thick. Can you post a snapshot of your PCB layout?
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Attached pictures of how I did it in the end:

1. The top picture is the etched PCB and it is clear the traces will not cope (BAT & LOAD).
2. The populated PCB showing the connections on the copper layer(it is single sided).
3. The PCB from the top, showing the substantial wires for the battery and the load leading to an XT60 connector each.
4. A closer look at what I did with the FET connections. I pushed a longish piece of the wire through the hole and folded it over the track towards the FET and then bent the leg of the FET onto the wire and soldered the whole lot together. I would have liked to have embedded the FET leg in the centre of the wire, but that proved too difficult. The positive side(unswitched) just had the wires pushed through and soldered together.

I don't think my black wire will cope with 80A but it should handle 30A, which is normally the most I should need and approaches the limit on the single FET (I did not bother to solder in the second FET, I will do that if needed). Unfortunately I ran out of heavy gauge wire.
 

Attachments

  • 20140617_111151.jpg
    20140617_111151.jpg
    189.7 KB · Views: 291
  • 20140617_161408.jpg
    20140617_161408.jpg
    155 KB · Views: 263
  • 20140617_161421.jpg
    20140617_161421.jpg
    153.9 KB · Views: 233
  • 20140617_161534.jpg
    20140617_161534.jpg
    92 KB · Views: 238

OLIVE2222

Oct 2, 2011
690
Joined
Oct 2, 2011
Messages
690
That's look OK, the soldering is a bit generous however you can probably go for a maiden test.
To improve the layout a bit more room around the FET should allow you a cleaner wiring.
Maybe the XT60 connectors can be soldered directly on the PCB, no more wires so.
Olivier
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
@OLIVE2222: Thanks for your comments.
I have not been able to find PCB mount XT60 connectors, they all have solder cups for wires and I tried mounting one on a PCB and that is a REAL mess. The tightness of the circuit around the FEt is to try and keep the wires as short as possible on the copper side and also to keep the board as small as possible.

I have tested the board with 2A and it works great. I will gradually ramp up the power drain.
 

Rleo6965

Jan 22, 2012
585
Joined
Jan 22, 2012
Messages
585
Did you also consider placing a heatsink on FET? PCB trace might survive the high current but the FET might not.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Did you also consider placing a heatsink on FET? PCB trace might survive the high current but the FET might not.

That is why the FET and 7805 is 'upside-down'.
The heavy connection to the thick copper wire will also serve as a heatsink.
 
Top