I am building a board to power solenoids. The solenoids require 40amps to open. I am using a P-Type MOSFET to supply each solenoid. Attached is a drawing of the board. My question is supplying the voltage to the MOSFETS. I am assuming that a power plane across the board is going to be the best I can do. I also know a ground plane is good to have in a design as well. All of the components are on the top layer and it is a two layer board. Can I flood the entire top layer with the 12v input to help out with the supply to the MOSFETS and use the bottom layer as the ground plane and not have any problems? Better yet, can I put a power plane on about a third of the bottom layer and the other 2/3rd of the bottom layer a ground plane? Meaning have the ground plane on the bottom layer underneath the PIC chip and other components, then have the power plane on the bottom layer, along with the full top layer, just underneath the MOSFETS and connect to the top plane with multiple VIAs?

 

Neat as it would be to have everybody on the PCB, I would think that pushing 40 amps through any reasonable area of foil would be iffy. Especially where it make connection with the source pins.

I would locate the FET/s on a separate heat sink with heavy wire connections to the solenoids. The gate drive would come from the PCB, but that is all.

 

For MOSFETS that are rated say 50 amps, how do they connect to the board besides the source pins and the foil? For connecting to the solenoids, I am using two 14awg wires. There are actually eight MOSFETS, four are for the high current.

Note: the solenoids are only activated for five seconds at the most and I understand the 40amp draw is just initially to open against 1000psi.

Neat as it would be to have everybody on the PCB, I would think that pushing 40 amps through any reasonable area of foil would be iffy. Especially where it make connection with the source pins.

I would locate the FET/s on a separate heat sink with heavy wire connections to the solenoids. The gate drive would come from the PCB, but that is all.

 

One thing which works nicely is to use vias to help you distribute the current over multiple layers around the point of contact. So at the pad where it will solder into the FET, use vias in/around the pad to break it to and from the solder joint with nice thick tracks on all layers connecting into the pad. This keeps all the current from having to flow down one annular ring of the pad.

The biggest concern is having one pin have to get all of that current from one contact point on one plane.

 

Consider enforcing solder on the top and bottom layers also. The middle layers will have weak contact at best with the pin itself, and therefore all current from the inner layers has to flow down the annular ring to get to the solder joint.

If it can flow to top or bottom, doubles the throughput.

 

Thank you for the reply. This mosfet, Infineon SPD50P03L G has two pins for the Source. It is a surface mount deal. I am pretty new on working with something that requires high amp loads. I can see the concern for the small foot print of the two surface mount pins that source the voltage. I just have to think that Infineon wouldn't build a part unless it can handle the load. Hopefully thats a correct assumption lol.

I have one board built that I plan to test next week. It has a power plane on it, just one side though. I had started drawing up a board that has the power plane on both the top and bottom layer and was wondering about using a bunch of vias like you mentioned. I found a couple busbars that I might use also or instead of the power planes.

I have some potting compound that is mixed to help with heat dissipation. What do you think is better for a heat sink? Compounds or metal formed?

Consider enforcing solder on the top and bottom layers also. The middle layers will have weak contact at best with the pin itself, and therefore all current from the inner layers has to flow down the annular ring to get to the solder joint.

If it can flow to top or bottom, doubles the throughput.

 

Would a possibility be to put the MOSFET off the board, attach it to the casing or a big heat sink and run wires to it? No need to worry about the PCB traces burning up, as long as the wires go directly to your power source of choice.

I'm not sure though. I've never worked with really high currents. I do know however it only takes about half a second for a trace to burn up with too much current, because I've tried it before, with about 8 amps D.C. at 12 volts through something which looks it could have carried maybe 1 or 2 amps at best.

 

Here is a trace width calculator based on IPC-2221 (formerly IPC-D-275). It can give you an idea of where to start with your track widths are for minimal distances on broken up planes. I think I remember one of the comments saying it extrapolates the curves above 35 Amps. Can you go to a 2 oz/ft^2 copper board or higher?

http://circuitcalculator.com/wordpress/2006/01/31/pcb-trace-width-calculator/

 

Alot of times you'll need to run two tracks of equal width on two independant layers and bring them together near the pad with 5-10 vias to help the current spread itself between the two. If it is just off a power plane, just make sure you consider how much current the vias can handle. I use a standard of 14 mil vias with 20um minimum annular ring and generally let 1A loose on em without problems.

 

I am considering a copper busbar. I found copper bar online pretty cheap and I have a friend with a CNC machine that can make the bar fit for me. I will need to change my PCB around a bit. 0.125"x0.500"x4" copper bar will handle it pretty easily. Will the copper work as a pretty good heat sink?

 

I think copper can be a pretty good heat sink as long as you have it thermally well connected and there is enough surface area to dissipate the heat. You will have to test in your application if it is enough for your conditions. As far as a heat sink goes, your bar is not very big.