First of all sorry for the lengthy post - I'm trying to give as much info as I can.

I’m currently designing a PCB for a wall-mounted control unit which has control logic and mains switching/detection. I’ve done low voltage boards before, but not with mains on board.

I'm keeping the mains side well away from the logic side, and setting track widths based on a max temperature rise of 4C on 2oz board, but I'm not so sure about the small taps taken for mains detection.

To describe the circuit (see diagram) - there are 4x opto-driven mains relays (feeding 4x IEC sockets), 4x opto mains detectors connected to the SWITCHED LIVE feeds to the IEC sockets and two 5v PSUs (one for logic side, one for the mains side of optos) - all optos have slots for isolation.

The mains fuse is 5A, and all four feeds out of the IEC sockets (1.1A, 1A, 1A and 200mA loads) are each on 6A cable, with the mains input "bus" track to the relays also rated at 6A.

Each relay output is fused at 2A on the PCB, with tracks rated at 3A and the mains inputs to the medical grade SMPSUs are collectively fused at 250mA (track is rated at 1A).

I've not forgotten NEUTRALS: IEC NEUTRALS don't go via the board and are wired directly on the IEC sockets with 6A cable. For the PSUs they're rated at 1A, for NEUTRAL to the detectors they are (at the moment) rated at 1A, with the NEUTRAL "bus" rated at 6A.

But I'm trying to figure out the best way to deal with the taps that feed the mains detection ccts.

To me, logic says that their (and the associated neutral) track widths also ought to be >2A as the tracks are effectively fused at 2A. However, the actual current is under 1mA (there are 2x 220k mains-rated resistors and a 1N4007 in series).

I had originally planned to have these tap tracks rated at 1A - bearing in mind that these taps are already protected by the 2A fuses do I: (a) set the track width to 2.5A (1mm and a bit tight) (b) protect each with a resettable fuse rated at 50mA or (c) is the risk so very low as to not be an issue? I don't think this is an option, but I include it anyway

And a related question - throughout the board LIVE and switched-LIVE will be at least 5mm away from neutral and other tracks/connections other than EARTH where it's 3mm. But should LIVE and SWITCHED-LIVE be similarly separated? I'm not sure that's possible, as even the mains relay contacts are only 2mm apart.

In writing this, I actually think I've figured it out, but any further input would be appreciated.


enter image description here

  • \$\begingroup\$ When routing on PCB, the fibreglass material is a dielectric, so instead of air gap (clearance) you have to consider dielectric path (creepage) , for any kind of safety factor refer to creepage tables for separation, they are a function of the signal voltage, pollution factor, and elevation \$\endgroup\$
    – crasic
    Commented Aug 29, 2020 at 15:58
  • \$\begingroup\$ Thanks for that - for spacing I'd checked various references, including tables on the PCB service's site and also advice from www.smps.us - but I also looked at the PSUs in a couple of big-name laser and inkjet printers. 5mm comes out about right, so I've gone just a bit bigger. But these clearances are only within PSU part of the board (eg L-N) - clearances between L and signal/logic tracks are MUCH bigger (and with slots) - clearances between L-E are less, but that's because the components don't permit anything greater (on the terminal block for instance, where E is between L and N) \$\endgroup\$
    – Gotty
    Commented Aug 29, 2020 at 16:57
  • \$\begingroup\$ Just adding that this is nearly all "functional insulation" as distances between mains voltages and control logic are well in excess of "reinforced clearances" \$\endgroup\$
    – Gotty
    Commented Aug 29, 2020 at 17:06
  • \$\begingroup\$ By the "IEC" block it says "Max Load 1.2 A", but the sum of the max loads is 2.2 A, according to the annotations. Which one is wrong? \$\endgroup\$ Commented Aug 29, 2020 at 18:43
  • \$\begingroup\$ The max load of the first block is 1.2A - the sum of the four block loads is 1.2+1+1+0.2 = 3.4A - so the "bus" feed to all relay inputs is design to handle 6A, and the mains fuse is 5A \$\endgroup\$
    – Gotty
    Commented Aug 30, 2020 at 12:09

1 Answer 1


On the face of it the fuse ratings and PCB track current ratings seem OK .Careful there is a trap that I have seen several times Now imagine that your load is a short circuit .Sure your fuse will blow in maybe 20ms .Check your fuse curves .What you do not want is the PCB tracks to blow before the fuse under such fault conditions .So I squared T is also a PCB track rating criterion .When you ballpark calc the melting temp of the PCB copper ,The resistance of the trace ,The Blow time of your fuse and the heat capacity of your trace you may want bigger tracks than you think .If you cant be bothered doing the calcs or are worried about the accuracy of the calcs then test under fault conditions to avoid feild failure under such conditions .

  • 1
    \$\begingroup\$ Good points, many thanks. All track widths have been calculated with very big margins. They're calculated based on 1oz copper, but the final PCB will be 2oz. Also, all calculations have been based on 2x expected max current and a temperature rise of no more than 4C. If I do a reverse calculation (using online trace-width calculators and 2oz copper) at 30A the big tracks should only get to 70C ... and to reach copper melt point it'll take about 120A. \$\endgroup\$
    – Gotty
    Commented Aug 30, 2020 at 13:25
  • 1
    \$\begingroup\$ I've just been doing some reading about using PCB tracks as fuses - may be an in-extremis option. In fact, in the process, I've also just found a really useful downloadable calculator (Saturn PCB Design - PCB Toolkit) which actually calculates track fusing current, and a whole lot more. I'll also do some experiments on home-made 1oz tracks. Thanks again for the thoughts - I would like to avoid failure in the field, but would also to avoid burning down the hall! :-) \$\endgroup\$
    – Gotty
    Commented Aug 30, 2020 at 13:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.