Guidence for large PC traces, for high current relay

Using Arduino Motors, Mechanics, Power and CNC
#1

In a current project, I needed a contractor that will reliably run a 1HP 120VAC motor, or 2 HP 240V, and continue to do so for a long time, (I calculated that I needed 20,000 on off cycles over the product life). After messing with several relays that theoretically "should" work, but probably wont have the longevity I wanted, I had a good long consult with one of the friendly Engineers at Potter and Brumfield, and decide to use their T92 series relay. Here's the data sheet.

This is a very robust relay with two 40 amp contacts, rated for switching the loads I need 100,000 cycles!. OK, it will cost a little more, but the product will be reliable.

Unfortunately, good things often DON'T come in small packages when it comes to power. My only hope of making this relay fit in my enclosure is to use their "Mounting Code #1" version, which is a PCB mount.

My question is about figuring out the most economical way to make PCB traces that will safely carry the current, without heating up much. I DID say this relay was overkill, as I'll never need to carry much more than 13 Amps continuous, based on the the name plate data of the motors I'll be controlling. So I thought I would just run the biggest traces I could (I did copper pours within drawn shapes actually), which came out to about 0.350 inch wide, and ran them to a hefty 30 amp PCB mount barrier strip. Then I duplicated those main traces on both sides of the PCB, and I figured I'd also go for 2, 3, or 4 oz copper.

Well apparently the China board house I usually use ( PCBWAY.COM ) is very good about doing 1 oz copper boards at super low prices (like 10 for $5 plus shipping). But as soon as you go to 2 oz copper, the same number of boards would be $36, and 3oz jumps to a whopping $140!!! (they don't even offer 4 oz).

I don't know why the higher density copper costs rise exponentially like this, but I'm still prototyping, and would rather avoid the expense until its proven to be necessary. So I'm trying to figure out whether I need that extra copper thickness. Based on what I explained already, can anyone offer some advice as to whether I could get away with plain 1oz copper? This image probably won't show up in a browser to scale, but my board looks like the below. Remember the wide traces are about 0.350 wide, are duplicated on side 2, and will never have to carry more than 15 amps continuous.

|500x321

If 1 oz copper is definitely NOT up to the task, a couple of other things crossed my mind...

  • I could wad up a lot of extra solder on the bottom side of the board to increase its overall metal thickness
  • Knowing the Relay pins are about 0.156 long, I could stack two boards (0.062) together to simulate a single higher density copper board
  • Someone could tell me my board house is being ridiculous, and tell me about another "budget board" company that doesn't treat copper like gold
#2

Depends how many these boards you actually need.

If it’s just a couple, order regular board and while your waiting buy appropriate thickness copper sheet for the power circuit and cut out with dremel / hobby hacksaw.

Layer on top board trace and epoxy in place.

If it’s a bunch, maybe find a local metal shop that can cut you out the pieces in bulk.

#3

Slumpert:
Depends how many these boards you actually need.

If it’s just a couple, order regular board and while your waiting buy appropriate thickness copper sheet for the power circuit and cut out with dremel / hobby hacksaw.

Layer on top board trace and epoxy in place.

If it’s a bunch, maybe find a local metal shop that can cut you out the pieces in bulk.

Not a bad idea. But I’m hoping to better learn how to predict, or at least find some guidelines. Just how wide a trace of what thickness copper for a given current, what temperature rise to expect, and what temperature rise is “OK” vs “Unacceptable”. Everything I’ve found on the subject is VERY confusing.

#4

Eureka! I think this is just the calculator I was looking for!

https://www.7pcb.com/trace-width-calculator.php

#5

If this is going to be a commercial product, you could save yourself time, money and misery by using an external definite purpose rated contactor. A board that small, even with proper layout, will have issues with testing agencies due to spacing.

At a minimum, if you’re married to the idea of that relay, allow for a 50% increase in board size so you don’t paint yourself into a corner.

BTW, you also need to investigate the importance of extra vias on top/bottom dual traces.

#6

WattsThat: If this is going to be a commercial product, you could save yourself time, money and misery by using an external definite purpose rated contactor. A board that small, even with proper layout, will have issues with testing agencies due to spacing.

At a minimum, if you’re married to the idea of that relay, allow for a 50% increase in board size so you don’t paint yourself into a corner.

BTW, you also need to investigate the importance of extra vias on top/bottom dual traces.

In this case, the relay is available with screw contacts. The ONLY reason for using the PCB version was because it takes much less space in one critical dimension. I'm basically just moving the screw terminals to a separate barrier strip on the same side of the board, which reduces height. Yes there are a few other items on the board i didn't mention (the rectangle on the right for a small SMPS s8upply), and there's a reverse blocking diode. But thats all thats on the board. The rest of the product is on a separate board, so having this separate "relay and power supply" board helps keep all 120/240VAC parts completely separate from the rest of the product, which is all low voltage.

But I don't think I'll need to add anything else to the board I've shown here, and there is a limited amount of space inside my chosen enclosure. I can expand it some in the uppermost direction, and I have a little room to expand it on the rightmost side. But you know, I've taken apart some professional load management devices use by electric companies, and I can't believe how much they cram into a small enclosure sometimes. I'l trying to follow the U/L guidelines I've been able to put my hands on though, and this particular relay is about as close to a definite purpose contactor as I believe I'm going to fit. If it doesn't work out, I'll look into external contactors. But I think it would add needless cost, and I have reason to believe that if the product has to come in two parts, it will make it a harder sell.

About the vias though... if I have the same wide traces duplicated top and bottom, won't vias cause me to lose conductor surface area?

#7

Clearance between the two high voltage tracks is inadequate.

You should perhaps use a fusible resistor in series with your voltage sense take-off line to protect the traces from being vaporized and flashing over (which would cause complete destruction of PCB), should the device on the right fail short-circuit.

Is the device on the right a small (voltage) transformer?

You should orient both current sensor and voltage transformer so mains is only at one end of the board, and there is a clear isolation gap between high and low voltage parts. You can then have a nice milled slot for better isolation.

#8

Hi,
Turn Relay 1 around, so that the coil pins are at the top, this will enable you have shorter low voltage tracks.
The power terminals will then be where the coil contacts used to be, and your mains wiring to the terminal block will be away from the low voltage tracks.

If this is a low run prototype, use copper wire soldered to the tracks from the relay to the terminal block to give you the current capacity.

Did you ask if they had similar relays with spade terminals on top, so you wouldn’t have to put mains on the PCB?
relayspade.jpg

Tom… :slight_smile:

#9

MarkT: Clearance between the two high voltage tracks is inadequate.

Well now that you mention in, it only costs another 50¢ to add a 500V bidirectional TVS diode, so I will do that. I guess I could add a snubber too, but still, I hope you're wrong! The screen shot is not to scale, but the gap between those traces is actually about 0.060". According to this chart, that should be good for better than 600V on external traces. I understand about load inductance causing arcing when the contacts open, but the minimum contact gap of the relay is significantly less.

MarkT: You should perhaps use a fusible resistor in series with your voltage sense take-off line to protect the traces from being vaporized and flashing over (which would cause complete destruction of PCB), should the device on the right fail short-circuit.

Is the device on the right a small (voltage) transformer?

You should orient both current sensor and voltage transformer so mains is only at one end of the board, and there is a clear isolation gap between high and low voltage parts. You can then have a nice milled slot for better isolation.

No, that space on the right is for a small SMPS supply on which the manufacturer kindly added standoff pins for me. It just supplies 12VDC power to a separate main board. I only put it there because the SMPS and the relay are the only components that are not on the main main product board, because they the only components that ever have to deal with the AC mains. That is a help in getting approvals, because if anything has to be changed it confines changes to that one board.

I'm not sure what you meant by a current sense, but this relay doesn't have anything like that.

Thanks for the input!!!

#10

TomGeorge:
Hi,
Turn Relay 1 around, so that the coil pins are at the top, this will enable you have shorter low voltage tracks.
The power terminals will then be where the coil contacts used to be, and your mains wiring to the terminal block will be away from the low voltage tracks.

If this is a low run prototype, use copper wire soldered to the tracks from the relay to the terminal block to give you the current capacity.

Did you ask if they had similar relays with spade terminals on top, so you wouldn’t have to put mains on the PCB?
relayspade.jpg

Tom… :slight_smile:

Hey Tom thanks.

P & B does have 3 versions of that relay (see the data sheet I linked in the OP). They have one with screw terminals, one with space connectors, and the PC mount. But the reason for choosing the PC mount, AND the reason the board has an “L” shape, and the reason for the relay orientation, all has to to with maintaining reasonable internal clearances to other boards and connectors inside my chosen enclosure. There’s no easy way to completely show this as I don’t have a 3D model drawing, but trust me… this is the ONLY way I can make this relay fit. And after pretty carefully examining specs and physical sizes on many relays over the past year or two, this one is the best I’ve found for the job. I’ve considered an SSR too. Really wish I could use one but they just generate too much heat.

As for the addition of copper wire, I know there are a few things I can do. But what I really want to know is how to make acceptable PC traces. Every pro device I’ve seen in the category I’m working on (load Management devices), where I was able to get a peek at the “innards”, are all using PC mount relays. So I really need to start becoming an expert on this. I think the calculator I found and linked in post #4 has given me some confidence. I’m still assuming two (front and back) 500 mil traces should be as good as one 1" trace, but I think that’s a reasonable guess.

As an aside, I see you pictured an OMRON relay. Although I’ve used many OMRON relays throughout the years, their current website IMO is the absolute worst I’ve seen for any product line. I can’t find anything there, and they are not 1/10th as responsive to customer inquiries as Potter & Brumfield (which is now part of TE connectivity).

#11

PeterPan321: The screen shot is not to scale, but the gap between those traces is actually about 0.060". According to this chart, that should be good for better than 600V on external traces.

What's suitable for 600V at low power levels isn't appropriate for mains where tens of kilowatts are microseconds away if any kind of discharge starts (insect, piece of metallic debris, condensation, nearby lightning, faulty fluorescent tube ballast inductor, induction motor in a fridge switching....).

Use much more separation than 1.5mm for mains. Consider how chunky mains plugs and sockets are and realize there's a reason for that.

Google terms are "creepage", "basic insulation" and "reinforced isolation"

#12

MarkT: What's suitable for 600V at low power levels isn't appropriate for mains where tens of kilowatts are microseconds away if any kind of discharge starts (insect, piece of metallic debris, condensation, nearby lightning, faulty fluorescent tube ballast inductor, induction motor in a fridge switching....).

Use much more separation than 1.5mm for mains. Consider how chunky mains plugs and sockets are and realize there's a reason for that.

Google terms are "creepage", "basic insulation" and "reinforced isolation"

Thanks. I did find this article which seems a good starting point. I likely will try to increase the spacing on subsequent versions of the board.

For now, one good thing is that in this case, they are not really "mains", as in Line to Neutral or Line to Line traces. The traces I'm routing here are to a relay that acts like a switch between one line and a motor. In other words a dead short would not roisk a direct L-L or L-N short. So there are no tens of killowatts available here. Also, the board is going in a watertight enclosure, IP-67 and NEMA rated for outdoor electrical use. So hopefully that screens out a lot of incidental problems (bugs, debris, etc). Now lightning? I don't know that there's any economically practical lightning protection, and I don't think ANY PCB trace spacing's could help much with that. Even my bi-directional TVS diode would probably explode like a firecracker with a lightning hit! :-)

#13

PeterPan321: Well now that you mention in, it only costs another 50¢ to add a 500V bidirectional TVS diode, so I will do that. I guess I could add a snubber too, but still, I hope you're wrong! The screen shot is not to scale, but the gap between those traces is actually about 0.060". According to this chart, that should be good for better than 600V on external traces. I understand about load inductance causing arcing when the contacts open, but the minimum contact gap of the relay is significantly less.

That spec for the gap is for dry air, the gap should be wider, 1.5mm is not enough, even with TVS. The spikes from an inductive load being disconnected at peak AC volts can be high and powerful. Don't forget 240Vrms = 240 * 1.414 = 340Vpeak I would be using at least 2.5 to 3mm and conformal coat it. Thanks.. Tom.. :)

#14

TomGeorge: That spec for the gap is for dry air, the gap should be wider, 1.5mm is not enough, even with TVS. The spikes from an inductive load being disconnected at peak AC volts can be high and powerful. Don't forget 240Vrms = 240 * 1.414 = 340Vpeak I would be using at least 2.5 to 3mm and conformal coat it. Thanks.. Tom.. :)

Yes thanks... and I will do that! Of course you know I'm losing trace width to add more gap, which naturally seems counter productive. I've also learned that NOT doing a conformal coating allows the copper to dissipate heat more effectively, thus allowing a lower trace width to handle a little more current. Ah the world of trade offs. :-)

I'm actually adding the TVS diode to help the relay contacts too! Despite being rated for 100K full load switching cycles to an inductive 1HP motor, it can't hurt. \

BUT my point was, those relay contacts are not even 1mm apart when open ( cut one open to see for myself, with some spark plug gap feeler gauges). I guess the modern contact alloys can take the momentary inductive arcing repeatedly, but the way I see it ANY arc that occurs for ANY reason will happen at those contacts without the TVS, because that IS the shortest gap. As with all my engineering, sooner or later I have to cross my fingers and see what happens.

I'll shave off a little before I submit the gerbers. If it works out thermally without heating up much under load, next time I'll go for 2oz copper (both sides) and will opt for an even wider gap!

#15

For now, one good thing is that in this case, they are not really "mains", as in Line to Neutral or Line to Line traces. The traces I'm routing here are to a relay that acts like a switch between one line and a motor.

You’re joking, right? Of course it’s a mains connection. The source of the power is a wall socket, right? It doesn’t matter that it’s only one side of the circuit.

Unless you’re providing the supply transformer for the motor and you can calculate what the available short circuit fault current is, you’re just kidding yourself about what those traces are carrying and what can happen when things get ugly.

#16

Hi,

For now, one good thing is that in this case, they are not really "mains", as in Line to Neutral or Line to Line traces

Unfortunately it is Line to Neutral when the relay is open. the motor even though it is not running is still a conductor.

Tom... :)

#17

WattsThat: You’re joking, right? Of course it’s a mains connection. The source of the power is a wall socket, right? It doesn’t matter that it’s only one side of the circuit.

Unless you’re providing the supply transformer for the motor and you can calculate what the available short circuit fault current is, you’re just kidding yourself about what those traces are carrying and what can happen when things get ugly.

Wait a minute. I DO know the exact short circuit current, because it is in fact the draw (inrush and/or sustained) for motor that I'm controlling. I also have a reasonable expectation that any such motor is being supplied through a breaker box, with a predictable trip current, likely around 20 amps, and certainly far less than electric company's transformer feeding the place.

If you are asking if I'm aware that the open circuit voltage is the same whether its through a motor or not, then YES... of course I understand that. But if you're trying to say the the short hazard for a one pole switch (what the relay is acting like) in series with the load is just as bad as if it were a two pole switch/relay, with mains on one side and the motor load on the other, then on that point we disagree.

I did say I intend to make the trace spacing wider in the next version. The current one is just 1 oz copper and I'll be anxious to see if the heating works out within the range I expect for the current trace widths. That in turn will tell me whether I'll need to go to more expensive copper thickness for a production board. I also expressed that despite the PC gap I asked about, the paralleled contact gaps of the relay are about 1/2 my copper gap. That also gives me some comfort for a first cut. Unless somehow the relay becomes entirely disconnected from the PCB, I can't imagine a situation where the arcing would not occur at the shortest gap available. Lighting, of course, is an all bets off situation right now.

#18

Wait a minute. I DO know the exact short circuit current, because it is in fact the draw (inrush and/or sustained) for motor that I'm controlling.

Okay, I’ll admit I posted a trick question. I thought you would say “sure, I know what the fault current is”. But I didn’t expect your response above as that is not the short circuit current. What happens if the motor shorts? What’s the current under those conditions? That’s your fault current and your relay and traces will be subjected to that current. Will the traces fuse and spew molten copper everywhere? Doubt it but you need to consider such things since, well, wait for it... it’s a mains application.

You said space is tight so I have to ask: you’re providing fusing, right? You cannot control the fault current levels without providing the fusing. The circuit breaker is only there to protect the wiring, not the device on the other end. Best to consider it now rather than finding out later when it fails testing.

With respect to trace temperature rise, you can test that on your first pass boards. Consider that it is most easily done with a low voltage, current limited supply. Your 13 amps continuous at 5 volts DC provides the same rise and is a lot safer and easier than testing with real world devices with ac mains.

#19

WattsThat: Okay, I’ll admit I posted a trick question. I thought you would say “sure, I know what the fault current is”.

You said space is tight so I have to ask: you’re providing fusing, right? You cannot control the fault current levels without providing the fusing. The circuit breaker is only there to protect the wiring, not the device on the other end. Best to consider it now rather than finding out later when it fails testing.

Well now that you mention it, I will do a live short test of the load a few times. But no, I'm not providing fusing. I will say that before I attempted this project, I examined at least two devices used by my local electric company for demand response, which simply meas devices designed to allow the company to remotely off-load customer devices during peak demand periods when necessary. Not only did none of them fuse their systems, but I don't think I saw any contractors as "hefty" and robust as mine.

But still, I'm dependent on the customer's breaker. If the breaker trips on over current, I WILL want to ensure my device can come out of it unscathed, even after a few customer breaker reset attempts.

But I don't think I;m alone. Very few appliance devices seem to have internal fuses anymore. Sometimes a push-to-reset- circuit breaker.

WattsThat: With respect to trace temperature rise, you can test that on your first pass boards. Consider that it is most easily done with a low voltage, current limited supply. Your 13 amps continuous at 5 volts DC provides the same rise and is a lot safer and easier than testing with real world devices with ac mains.

Good idea. Much safer. An isolation transformer and a hefty variac. Plus I can "touch" test the traces without risking a shock! :-)

#20