Reader questions: Thickness of PCB traces for a 120A solenoid

Posted on Friday, August 17th, 2012 in help by DP

George asks how thick PCB traces should be for driving a 120A solenoid:

 I’m doing a board design to power a solenoid that apparently draws 120 amps, but it’s only on for about 300 ms (I imagine it has some kind of “inverse curve” looking decrease with time, I haven’t thought about it too much). The weird thing is that the solenoid comes with 18-gauge wires, or something close. I’m wondering how thick I should make the traces. I’m assuming I don’t have to go the recommended 5-inch wide (using PCB calculators). In fact, I started thinking I would only have to go as wide as the max diameter of the pad, which is TBD.

Now, the only reason one has to design wider traces for higher current is to avoid heat, correct? Because this is for an electric shifter (In Ohio State University’s new formula car), so it’ll have a very low duty cycle. What do you think? Should I just go ham on the traces and make them huge? Or should I limit them to the width of the pads (or the circumference of the wire for that matter)?

Via the contact form.

This entry was posted on Friday, August 17th, 2012 at 11:00 am and is filed under help. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

21 Responses to “Reader questions: Thickness of PCB traces for a 120A solenoid”

  1. Tony says:

    If the traces are on the order of 8mm or less, they will vaporize rather quickly with 120 amps flowing through them.

    Whatever you have driving the solenoid, say a MOSFET, place one terminal for the solenoid as close as possible to MOSFET drain and the other terminal as close as possible to the Vsupply terminal, leave the soldermask off in high current regions and cover the traces with lots of extra solder. If you need to make longer traces, again leave the soldermask off and solder on a heavy gauge bare wire directly onto the trace.

    • Jack says:

      Yep. With 1oz copper it’ll take under 2 seconds. With 2oz copper, less than 5 seconds. But with 4oz copper he’d get a full 12 seconds before fusing!

  2. Steve Bennett says:

    Are you not better off with 18-gauge wire and crimp terminals, like what came with the solenoid?
    Why does this need (or want) to be on a circuit board?

    • Filip says:

      my thought exacly, Wheneve I came in contact wiht high amp solenoids, wire’s were used directy,, or a circuit locker, with large copper stips/bars….have the mosfet, and the solenoid adn all the high current traces, as wires, only use the PCB for small controll currents like the mosfet gate etc,…

      P.S. that solenoid must have one motha of a protection Diode

  3. Matseng says:

    What you need is a PCB BUSBAR. They are strips of copper sheet with mounting pins that is soldered as a component on the component side of the pcb.

    This model is common in medium to high current power supplies.

    Then there are this version that is used for even higher currents.

  4. Cabe says:

    Would something like an SCR not be a better choice for the intermediate stage? Remove the need for having any high current on the board.

  5. Chuck Saunders says:

    On a 1oz board, the copper thickness is 34um, or 0.034cm.

    If you have a trace width of 8mm, (8)*(0.0034) = 0.272 mm2. Now, admittedly from Wikipedia, the maximum current density for PCB traces on the outer (top and bottom) layers is 35A/mm2.

    This means an 8mm trace on a 1oz copper board would give you 9.5A before it blows.

    If you’re looking to design to spec for 120A using that 35A/mm2 rating, then you’d need 3.429 mm2 cross sectional area of the trace, which at 0.034cm means your traces need to be just over 100mm wide.

    That converts to 10cm or 3.96 inches wide. This is the minimum you need to not vaporize the copper. Any margin of error or factor of safety will increase the width.

    You can reduce the width by increasing the copper weight on the board. A 2oz board should halve the width requirements, but you’re still left with 2″ traces. As others have mentioned/asked, this shouldn’t be on a PCB unless you positively need it to be.

    I work with unmanned vehicles for Virginia Tech, so I absolutely understand that there is always space and weight requirements to be met, but if this must be on a PCB I would look at running wiring to the solenoid and using vias strictly as a physical mount, no tie-ins to a circuit.

    Keep in mind, too, that anything that supplies power to the solenoid needs to use the same width of traces. This means the solenoid, whatever does the switching, the power supply, and anything else that sees/feeds 120A.

    If I were you I would use a relay to switch the solenoid. Small automotive relays draw something on the order of 0.5A or so, and you can easily run one on a PCB. It reduces trace width everywhere and the relay does the high current handling. Just be sure you get the right specs for it, as this is an issue we had with our vehicle. The voltage and current ratings for the coil and the contacts are not necessarily the same, and if they’re different they don’t mark both ratings on the relay. You have to look up the part number. We have a small relay drive a large 500A contactor for one of our electric cars.

  6. CoytHV says:

    Very interesting. I’m not sure, but the 120A rating may be the initial inrush current of the solenoid, and not the actual working current rating. I mean the solenoid may draw 120A for an extremely short amount of time when it is first turned on in order to magnetize it’s coil. Once the coil is fully magnetized, the current draw should go down. This would explain the small gauge wire connected to the solenoid.

    If this is what’s happening, your circuit doesn’t need to be able to supply a continuous 120A. It only needs to supply a short pulse at 120A, and this can be done by adding a capacitor on your solenoid power supply board. The capacitor will charge and then be able to supply the solenoid with the 120A inrush current that it wants during start up instead of burning up your board.

    Try to look at the data sheet for the solenoid and see if the current draw curve goes down over time after the initial “ON”. You could also measure the current draw with the solenoid on for a longer amount of time. Design your board for that amount.

    If the solenoid truly draws 120A continuous, then follow the directions of the other posters and use something different such as a small relay, mosfet, SCR, etc to drive your solenoid.

  7. Rich says:

    The following site has a calculator for temperature rise/fusing current that you may find helpful. Unfortunately it does not cover a pulse situation

  8. SQKYbeaver says:

    make the distance between your mosfet and the solenoid as short as possible don’t forget the protection diode.

    the high inrush and low duty cycle, sort of allow you to use an average current for your calculation. the solenoid will disapate most of the energy

    if your duty cycle is going to be less than 1% there is a fair amount of time for the trace to cool before the next cycle, i would bet you could get away with a 10mm wide trace with minimal extra precautions(solder over traces) as long as you keep the trace short enough(about 5cm)

    what is the average current of the solenoid? how long does the inrush current stay at 120A?

  9. sam says:

    Here is a link to an article discussing this topic.
    based on the equations in the article and your specs of 120 amps for 300ms you would your traces to be 6.55mm wide with 1oz copper, this is without any safety margin. doubling the with or copper would probably be a good idea to avoid de-laminating the traces, also the suggestions of coating the traces with solder is a good idea.

  10. If I were you, I’d use the PCB bus bars mentioned by Matseng (for reference, that’s Corsair’s solution on their 1200w single 12V rail power supplies – those are rated for 100A continuous so should be good enough for not-too-big). But beyond that, go with your gut: use the bars, make pads nice and big, as mentioned, leave any power traces unmasked and gob solder on them.

    But then once you’ve made something you FEEL should work, test it! Build one up and just fire the solenoid over and over and over until you blow the board out, then you’ll know if you’ve got enough headroom for practical use. If you can manage it, try to borrow a FLIR camera from some other department or see if your local FD will come by and let you experiment with theirs for a couple hours.

  11. Chuck Saunders says:

    A coil like a solenoid is like a giant inductor. Inductors act initially as an open, then the steady state impedance is that of the wire resistance.

    The maximum current draw for the inductor isn’t going to be an initial “in rush” of current, it’s going to be the long-term steady state draw.

    • That’s KIND of true. It has much more than anything to do with the real energy required for actuation, though. If it takes a lot of power to throw the solenoid, but then once it’s there it requires no power to remain in place, the inductor simile is largely moot. Especially if, as many of these kinds of actuators do, it has a limit switch at either end. Take a regular brushed DC motor, for example – the inductor simile is usually more or less true. EXCEPT that a stalled motor draws less current than a motor operating at maximum power (usually somewhere just above the torque peak in the torque-rpm curve), where it draws more current than in an unloaded spin.

      The real point is, george states 120A for 300ms, and I’m inclined to believe that’s accurate enough.

  12. piotr1600 says:

    Been involved in controls for engine driven systems for quite awhile.
    This is very similar to old school (pre-electronic engine) diesel engine fuel rack solenoid pull in currents.
    If the solenoid is reliably operated by its attached 18 ga wires, then you can almost certainly simply replicate that level of trace for similar reliably. 18ga wire definitely will not handle 120A continuous!

    Alternatively, if you decide that you really have to have that level of output, there are a number of inexpensive relays that will serve as an intermediate step.
    Something like this:

    Good luck!

  13. If you do a little bit of calculation, using values for
    the resistivity of copper and heat capacity of FR4 board material, and
    some pretty rough assumptions about how heat transfers from the copper
    track to the FR4, you end up with something like a 20 deg. C temperature
    rise for an 8mm wide track of 35um copper, after passing 120A through it
    for 300ms. That’s the bare copper – piling extra solder on top will
    easily lower the resistance hugely. So I think something like a 10mm
    track, with added solder, would easily be enough.

    Something to keep in mind is duty cycle. Even if a single “pulse” resulted
    in a large-ish temperature rise (say 50 degrees), that wouldn’t matter if
    the pulses only occurred every 5 min or so, giving the board time to cool
    between shots.

    Probably easiest just to try a length of PCB trace with the specified current
    and see what happens – does it heat up too much. What about connecting the
    trace to a welding power source (e.g. a basic buzz-box welder) and pulsing
    the welder with a relay in the mains?

    I modded my TIG welder ages ago to do pulsed welding. For a 100A pulse lasting
    100ms, I can use pretty thin wire (0.6mm solid core) and don’t notice any
    appreciable heating.

  14. Ross_Down_Under says:


    Your ” piling extra solder on top will easily lower the resistance hugely ” might need a rethink in light of Dave Jones’s practical evaluation in



    • Matseng says:

      If I remember that episode correctly the resistance was halved – a rather good improvement that you basically can get for free. It would also increase the thermal mass which would enhance the pulsed load carrying capacity.

    • Yep, should’ve remembered the resistivity! The main advantage in the case of coating the trace with solder would be to increase the thermal mass – a 0.5mm layer of solder on top only drops the resistance to about 40% of the bare copper trace, but increases the thermal mass nearly 10-fold, reducing the heating effect.

  15. hardcorefs says:

    You need a complete re-think, if you are actually contemplating such a design.
    Design defensively… for example what happens to the design “IF” the power is on for longer due to a fault……

    Just remember that electric car in China that burst into flames………….

  16. Vonnie says:

    hello!,I really like your writing so so much! percentage we keep in touch extra about
    your article on AOL? I need an expert in this area to solve my problem.

    Maybe that’s you! Taking a look forward to peer you.

Leave a Reply

Notify me of followup comments via e-mail. You can also subscribe without commenting.

Recent Comments

  • Kevin: I to think it is wonderful but really only a Darwin contestant if he has it next to his rubber ducky in the bath.
  • KH: *applause* I think this is good enough for a mention in the Darwin Awards. Ballsy and courageous. *thumbs up* Awesome, mate.
  • KH: It may not be very powerful, but regulations-wise, this is probably a dicey thing to do in every country that has modern wireless infrastructure. "For...
  • Blaine: You guys are awesome!
  • Samuel: have a nice week everyone