Trace width / spacing for Mains voltage

[yanir]

Hi, I'm working a project that switches mains voltage using a relay. I also power the micro controller and relay with mains (US 115v 60hz).

I've been using a variety of trace width calculators and in general they use the same equations but I occassionally get different results. Is there a calculator or method that you would recommend?

This is a product that will undergo certification and needs to be within spec.

Also, I have seen boards that are certified and have much lower track widths than the calculators recommend (10A @ ~74mil track and ~130mil space between next mains track.)
The calculators recommend way over 100mil for 10A, at 2oz/ft^2. why such a difference?

My design is space constrained, I am currently using 98mil tracks and around 75mil spacing. I can make changes but I a limited by a connector (that I would prefer not to change). So I can't go much wider.

Any one have any ideas or direct experience with this? Are there any good resources on this topic?

Thanks

[IanB]

Any one have any ideas or direct experience with this? Are there any good resources on this topic?

I'm sorry, but don't put currents like 10 A on a PCB. My dishwasher control board put low voltage control electronics on the same PCB as a relay switching mains power to the water heater. The PCB tracks could take it but the solder joints at the relay and connectors could not (solder is a high resistance connection once significant currents are involved). Every dishwasher with this design eventually failed with a burning smell, smoke, and a fried PCB.

I imagine you must of course be dealing appropriately with safety and isolation concerns. How do you isolate the microcontroller from mains spikes and surges that might get through the voltage regulator?

[yanir]

Adequate heat sinking should control the solder from reflowing. This too is an appliance control application. I have examined several pcbs for similar products and also much higher currents ones(cooktops). They all hold up, the cooktop has massive heatsinks on two mosfets that share the load.

As for protection, right now I just have a 1.5 VA transformer and a rectifier but a fuse may be used.
 I have seen several similar design that don't use any fuse. This is a cost sensitive application, as most of these are and this may the reason for lack of additional protection.

The certification process will ensure that the device is safe. I want to use the largest track width I can, but I have to work with a specific board shape.

 

[IanB]

Adequate heat sinking should control the solder from reflowing. This too is an appliance control application. I have examined several pcbs for similar products and also much higher currents ones(cooktops). They all hold up, the cooktop has massive heatsinks on two mosfets that share the load.

The issue with my dishwasher is that the relay controlling the heater current is through-hole soldered onto the PCB. No heat is generated that needs heat sinking, but the solder joints are a weak point. Any weakness in process control during board manufacturer can produce a marginally high resistance joint. Since the voltage is 120 V the high resistance at the joint poses no impediment to the current; the current simply keeps flowing and the joint gets hot. The hot joint starts to oxidize, and gets hotter. Eventually there is a cascade reaction and the joint burns out. When I examined my dishwasher after it failed I found the relay pin was islanded in the middle of the solder cone on the pad and surrounded by a black air gap.

The same thing does not tend to happen with low voltage applications like PC power supplies in spite of the high currents involved. At low voltages a high resistance joint simply reduces the current. At mains voltages the current keeps flowing until failure.

I recognize that many appliances are using this kind of design, but it does not make it a good design. It is merely a symptom of low cost engineering and short service life expectations by the manufacturer.

Design codes for high current mains wiring specify crimped or screwed mechanical connections for a good reason. These are more reliable and durable than soldered connections.

You will do what the job demands of course, but I'm afraid you touched a sore point with me about the quality of design in today's world. No longer are things built to last.

[jahonen]

When I examined my dishwasher after it failed I found the relay pin was islanded in the middle of the solder cone on the pad and surrounded by a black air gap.

A single-sided PCB? They are notorious for such failures, plated-through holes are much more robust against that.

Regards,
Janne

[yanir]

You will do what the job demands of course, but I'm afraid you touched a sore point with me about the quality of design in today's world. No longer are things built to last.

I agree with you about things not designed as well as they could be. I think that good engineering is about meeting the requirements of the job for: function,expected lifetime and most importantly safety.

This is a fr4 2 layer design that uses plated vias for all through hole connections. All of my calculations show that my current design will go to 7.5A. What I want to know is why does another board have much narrower traces and claim 10A is the limit ( it carries RU certification, that's ul for a component).

[ejeffrey]

The same thing does not tend to happen with low voltage applications like PC power supplies in spite of the high currents involved. At low voltages a high resistance joint simply reduces the current. At mains voltages the current keeps flowing until failure.

That isn't really true.  PCs provide relatively constant current or even constant power draw (especially on the 12V rail that powers a lot of SMPSs). Until the voltage drop is so severe the system cuts out, rising contact resistance will continue to increase power dissipation.  A PC power supply that can is sourcing 40 amps at 12 volts is perfectly capable of melting any solder joints that have out of spec contact resistance before the voltage drop gets so severe that the load shuts off.  The supply voltage does make a difference in fault conditions, 120 volts into a short of a few ohms can put a lot more power in than a 12 volt supply.  10 amps can be and often is done safely on printed circuit boards.

Domestic appliances tend to have poorer records in this fashion because of a couple of reasons.  The total power is usually higher.  600 watts is a big PC power supply (and most PCs will never draw that much) but a small appliance.  Appliances are often subject to much worse heat, humidity, and vibration than a PC power supply, and often don't have fan cooling.  Home appliances seem to be under even more cost pressure than PC power supplies, which is amazing to me...

All that said, I don't trust appliance safety certifications at all.  They appear to be completely worthless.  I know personally 3 people with difference models of electronically controlled electric ovens who have had them fail full power on.  In two cases, the owners noticed in short order and were able to shut them down with the breaker.  In another case it happened while the family was on vacation and destroyed all of their cabinets and floor in the vicinity of the oven through heat damage.  They were damn lucky that it didn't burn their house down.  None of the three ovens appeared to be equipped with the most basic piece of safety element every heating element should have: a thermal fuse, or it wasn't implemented properly.  If you can get safety approval for an oven without a thermal fuse, nothing would surprise me.

[IanB]

I know personally 3 people with difference models of electronically controlled electric ovens who have had them fail full power on.

That's an interesting point, and now you've scared me. My oven has electronic touch sensitive controls and there is no mechanical isolation switch anywhere, either on the oven or on the wall nearby. In the UK ovens have always had a dedicated supply with a two pole isolation switch in the kitchen where the oven is. I've only just realized that's missing here.

[IanB]

What I want to know is why does another board have much narrower traces and claim 10A is the limit ( it carries RU certification, that's ul for a component).

I don't think this kind of thing is clear cut. Conductor sizes and current ratings are set by design codes based on various assumptions about the surroundings, available cooling, service conditions and so forth. Derating factors may have to be applied under adverse conditions, or credit could be taken for certain positive circumstances. Fundamentally you have to look at the maximum current, the resistive heat load, the available cooling, and the allowable temperature rise, and then you have to ensure your design stays within the permissible limits.

[yanir]

In household applications this design will typically only carry ~2.5A. The higher currents are for more industrial applications.

So does anyone know why there are different ways to measure current capacity?

Shouldn't the transformers be sufficient isolation? They are two separate coils of wire and the higher voltage side will burn out before severe damage could occur. The fuse really just protects the trans and allows the device to be fixed easily by replacing the fuse.

The cooktop I examined uses capacity touch to control the burners. The burners get the mains voltage convert to filtered dc and then PWM controls the amount of current to the heating elements for a digitally controlled device. Your oven probably works the same way.

[yanir]

What I want to know is why does another board have much narrower traces and claim 10A is the limit ( it carries RU certification, that's ul for a component).

I don't think this kind of thing is clear cut. Conductor sizes and current ratings are set by design codes based on various assumptions about the surroundings, available cooling, service conditions and so forth. Derating factors may have to be applied under adverse conditions, or credit could be taken for certain positive circumstances. Fundamentally you have to look at the maximum current, the resistive heat load, the available cooling, and the allowable temperature rise, and then you have to ensure your design stays within the permissible limits.

 Are there any documents or resources that elaborate. I want to be more knowledgeable on this topic and use this to make better decisions earlier in the design process.

[IanB]

Your oven probably works the same way.

By oven I literally mean oven, rather than a stove top or hob (which is gas in my case). The controls are one of those plastic laminated touchpad things. You turn the oven on and set the temperature via a digital readout and then it is thermostatically controlled. I have not taken it apart to be sure, but from the clicking sounds it seems as if the heating elements are on a mechanical relay (but it is difficult to tell if the clicking is from the thermostat or the main power, or indeed if it has electronic temperature control and the same relay does both duties). Whatever the case, power to the relay is electronically controlled, and if those electronics should happen to fail closed circuit...?

[IanB]

Are there any documents or resources that elaborate. I want to be more knowledgeable on this topic and use this to make better decisions earlier in the design process.

Sorry, this is not my area, I am only speaking in general terms. There may be some experts here who can point to specifics. (This is an example where you would typically consult a more experienced engineer at your workplace.)

[Neilm]

One of the reasons you may have seen different trace widths is that there are different requirements for traces on an outer layer as opposed to inner layers. I have used this in the past. http://miscel.dk/MiscEl/miscel.html It is based on the relevent IEC standard so the numbers it produces will probably be similar to what you have seen in the past.

If you want to increase the current carrying capacity of a trace what you can do is remove the solder resist from the track and allow it to be soldered. This will significantly boost the current carrying capacity of that track. 

Yours

Neil

[yanir]

One of the reasons you may have seen different trace widths is that there are different requirements for traces on an outer layer as opposed to inner layers. I have used this in the past. http://miscel.dk/MiscEl/miscel.html It is based on the relevent IEC standard so the numbers it produces will probably be similar to what you have seen in the past.

I have noticed the calculators that allow you to choose between inner and outer layer modes. I have seen differences between calculations on the external layers. I also have a part here that has a really narrow track compared to mine that claims to support 10A. I just wonder how they got there.

If you want to increase the current carrying capacity of a trace what you can do is remove the solder resist from the track and allow it to be soldered. This will significantly boost the current carrying capacity of that track. 

I don't know if that would pass certification.

[IanB]

Remember there is no "limit" to how much current a conductor can carry. The only questions are how hot it gets in the process, and what voltage drop is allowable at the other end. For short conductors the temperature rise is usually more important than the voltage drop.

Also, even though a conductor may be able to carry a certain current, that current still has to get into the conductor and out of the conductor at either end. The joints or connections may be the weaker links in the chain and they also must be able to handle the required current. In my dishwasher example above the track on the PCB could handle the heater current but the soldered connection to the switching relay was not adequately designed for the same duty.

[yanir]

Remember there is no "limit" to how much current a conductor can carry. The only questions are how hot it gets in the process, and what voltage drop is allowable at the other end. For short conductors the temperature rise is usually more important than the voltage drop.

Also, even though a conductor may be able to carry a certain current, that current still has to get into the conductor and out of the conductor at either end. The joints or connections may be the weaker links in the chain and they also must be able to handle the required current. In my dishwasher example above the track on the PCB could handle the heater current but the soldered connection to the switching relay was not adequately designed for the same duty.

Right. but this is going to be manufactured and is undergoing DFM. I am being told to increase this track width. I'm just looking for more information on how to assess track current capacity. My manufacturer have not mentioned any potential issues with the joints.

[IanB]

Right. but this is going to be manufactured and is undergoing DFM. I am being told to increase this track width. I'm just looking for more information on how to assess track current capacity.

So ask them on what basis they think the track width is too narrow, and what track width would they consider to be acceptable...?

[yanir]

Right. but this is going to be manufactured and is undergoing DFM. I am being told to increase this track width. I'm just looking for more information on how to assess track current capacity.

So ask them on what basis they think the track width is too narrow, and what track width would they consider to be acceptable...?

I am, I just want another source of info. Geez.

[IanB]

I am, I just want another source of info. Geez.

Well sorry about that. I don't know, and it appears nobody else here can give you an answer. I'll leave the thread alone now.

[yanir]

I am, I just want another source of info. Geez.

Well sorry about that. I don't know, and it appears nobody else here can give you an answer. I'll leave the thread alone now.

It's cool. I appreciate you trying to help. It does appear to be a tough topic to get info on. I've done quite a bit of searching, and looked for guidance in books but not much is in there. Perhaps this could be a good topic for Dave to cover? Maybe general high current routing or dealing with Mains voltage in projects. I know this was scary stuff when I first started dabbling with it.

[ivan747]

Adequate heat sinking should control the solder from reflowing. This too is an appliance control application. I have examined several pcbs for similar products and also much higher currents ones(cooktops). They all hold up, the cooktop has massive heatsinks on two mosfets that share the load.

The issue with my dishwasher is that the relay controlling the heater current is through-hole soldered onto the PCB. No heat is generated that needs heat sinking, but the solder joints are a weak point. Any weakness in process control during board manufacturer can produce a marginally high resistance joint. Since the voltage is 120 V the high resistance at the joint poses no impediment to the current; the current simply keeps flowing and the joint gets hot. The hot joint starts to oxidize, and gets hotter. Eventually there is a cascade reaction and the joint burns out. When I examined my dishwasher after it failed I found the relay pin was islanded in the middle of the solder cone on the pad and surrounded by a black air gap.

The same thing does not tend to happen with low voltage applications like PC power supplies in spite of the high currents involved. At low voltages a high resistance joint simply reduces the current. At mains voltages the current keeps flowing until failure.

I recognize that many appliances are using this kind of design, but it does not make it a good design. It is merely a symptom of low cost engineering and short service life expectations by the manufacturer.

Design codes for high current mains wiring specify crimped or screwed mechanical connections for a good reason. These are more reliable and durable than soldered connections.

You will do what the job demands of course, but I'm afraid you touched a sore point with me about the quality of design in today's world. No longer are things built to last.

So above 10A I should start considering more direct wire to terminal to relay connections, right?

[mkissin]

IanB is quite right. The current handling of a PCB track is often largely determined by how much of a temperature rise you are prepared to tolerate. There is no "right" answer, as it will also depend on your rated ambient temperatures.

The document you are looking for is IPC-2152, and the calculator I usually use is the Saturn PCB Toolkit (from http://saturnpcb.com/)

This won't help with your mains isolation question though.

[yanir]

Thanks for the pointer to the document. It looks great, but not free  , I will have to get it though. There are a lot of other great documents listed there.

[yanir]

So above 10A I should start considering more direct wire to terminal to relay connections, right?

Not really. I have designed a board to work with an H-bridge IC (VNH2SP30) that goes up the 30Amps (the IC does, not my board, made that to support the 10A @12VDC I am drawing). But this is a surface mount device designed for automotive applications. You just need to make the tracks wide enough (given your copper thickness) and provide adequate heat sinking. The IC has a huge thermal pad on the bottom to allow you bring all that heat to the bottom of the board through vias to ground plane.

I have also found some mosfets that can handle up to 50 amps (FGA25N120ANTD) This comes in surface mount and through hole.

The cooktop I examined used two of those to carry the load simultaneously. The relay still has to have traces running to it. So they need to be capable of carrying those currents. I would just keep these traces as short as possible to the connectors.

I have used the sunstone calculator to determine these widths given the copper thickness. I just want to figure out why this other product has such narrow traces for 10A and it's certified by UL.

[IanB]

...has such narrow traces for 10A and it's certified by UL

I believe UL are only testing for gross safety violations that could pose a threat to life or property? I don't think they are concerned with good design or bad design. If your product self destructs but causes no harm, that is not their problem.

(Edit: Underwriters Laboratories essentially are acting on behalf of the insurance industry. They are testing whether use of a product may give rise to any insurance liabilities.)

[yanir]

I believe UL are only testing for gross safety violations that could pose a threat to life or property? I don't think they are concerned with good design or bad design. If your product self destructs but causes no harm, that is not their problem.

I guess I had put to much faith in them. I do want this to last, but I'll take causing no threat to life or property.

[ciccio]

I believe UL are only testing for gross safety violations that could pose a threat to life or property? I don't think they are concerned with good design or bad design. If your product self destructs but causes no harm, that is not their problem.
(Edit: Underwriters Laboratories essentially are acting on behalf of the insurance industry. They are testing whether use of a product may give rise to any insurance liabilities.)

I've never had a design certified by UL, but currently I have two prototypes under certification for CE marking by a Notified Laboratory.
They are big multichannel audio power amplifiers with a maximum ac drain of 15 A at 230V.
The mains 15 A fuse is on a PCB, that acts as a power distribution to  multiple SMPS. The track width is about 4 mm. The lab made a number of thermal tests, including one with a blocked cooling fan and one with blocked airflow, reading a lot of thermocouples inside the amplifier, and one thermocouple was taped over one of the tracks carrying mains current.
They said that the temperature rise should not exceed the maximum temperature allowed by the PCB laminate manufacturer, in the worst condition. There are no minimum sizes.
Obviously, with fiberglass  double layer laminate, this temperature is higher than with "cardboard" single layer, so your track will be allowed to carry an higher current value.

This is not a case with wires: there are standards that set minimum copper sizes in function of current (e.g 1.5 mm2 from 10 1o 15 amps,  0.7 mm2 for 5 to 7.5 A, etc...

[IanB]

They said that the temperature rise should not exceed the maximum temperature allowed by the PCB laminate manufacturer, in the worst condition. There are no minimum sizes.
Obviously, with fiberglass  double layer laminate, this temperature is higher than with "cardboard" single layer, so your track will be allowed to carry an higher current value.

This is not a case with wires: there are standards that set minimum copper sizes in function of current (e.g 1.5 mm2 from 10 1o 15 amps,  0.7 mm2 for 5 to 7.5 A, etc...

We can believe of course that exceeding specified thermal limits might lead to the possibility of failure with fire or shock hazards resulting. So this kind of testing seems quite reasonable.