When do you allow for redundant traces?

[LoveLaika]

I did this practice recently for a board, but I was curious about how common it is these days, if at all.

So, usually, CAD software (in my case, KiCAD) doesn't allow for redundant traces. In a recent board of mine, I overrode this option so redundant traces are not deleted and stay in place. In my case, it allowed me to have traces on both sides of the PCB for redundancy from the same two points. That way, if one side is bad due to circumstances, the whole board can still be salvaged (probably).

I was able to do this because my board was quite simple. I was wondering if this was a common practice and when professionals use it. What justifies its usage, and what are its risks?

[tszaboo]

ATEX/IECEx calls for parallel redundant 1mm+ thick traces for some cases. So yeah it happens. It's far from common

[CaptDon]

Hi current is often shared by several traces, each on a different layer on a multi-layered board. They generally look exactly alike so the current divides equally and predictably.

[LoveLaika]

Hi current is often shared by several traces, each on a different layer on a multi-layered board. They generally look exactly alike so the current divides equally and predictably.

Sure, I can see this for high-current traces. But, can you predict such a thing as you said? I never really thought about it all that much. This certainly wouldn't be feasible for signaling.

[LoveLaika]

Interesting. Thanks for letting me know. I suppose in such an environment, redundancy is mandatory.

[tszaboo]

Interesting. Thanks for letting me know. I suppose in such an environment, redundancy is mandatory.

For safety components, it either double or triple redundancy.
It's so that we don't set people on fire by accident, serious stuff.

[SteveThackery]

I don't know what the MTBF of a PCB trace is, but I would have thought it would be longer than most components. Do you have to provide redundant resistors and capacitors, and semiconductors as well?

I ask because, although I was involved in electronics reliability, resilience and longevity, I've never come across redundancy in PCB traces, so I would love to learn more.

For the systems I worked on, resilience was usually provided at the module level, for example duplicated plug-in cards, etc.

[LoveLaika]

I don't know what the MTBF of a PCB trace is, but I would have thought it would be longer than most components. Do you have to provide redundant resistors and capacitors, and semiconductors as well?

I ask because, although I was involved in electronics reliability, resilience and longevity, I've never come across redundancy in PCB traces, so I would love to learn more.

For the systems I worked on, resilience was usually provided at the module level, for example duplicated plug-in cards, etc.

Given the end purpose of my circuit (working as a heater, being partially exposed to water), my simple circuit would fail over time due to corrosion and such. It's mainly the resistor that breaks, but sometimes, PCB traces would break/corrode/wear away as well rendering the whole board unusable. I just wanted a way to be able to reuse boards. If a trace was corroded on only one side, the board could be salvaged if the trace on the other side was still intact.

[tszaboo]

I don't know what the MTBF of a PCB trace is, but I would have thought it would be longer than most components. Do you have to provide redundant resistors and capacitors, and semiconductors as well?

I ask because, although I was involved in electronics reliability, resilience and longevity, I've never come across redundancy in PCB traces, so I would love to learn more.

For the systems I worked on, resilience was usually provided at the module level, for example duplicated plug-in cards, etc.

For protective components the simple answer is yes.
For a PCB trace, imagine that it's going to a connector, that connects current into a device. Connector gets abused, and one of the traces is broken, then it will go from short circuit to open circuit, which will make spark, which can set the explosive atmosphere on fire.
Basically imagine the worst case scenarios. Like a fan controller (inside the fan) in a coalmine. The standards were written to provide guidance, to be able to make safe electronics devices for these industries.

[SteveThackery]

I  just wanted a way to be able to reuse boards. If a trace was corroded on only one side, the board could be salvaged if the trace on the other side was still intact.

That's interesting. But, wouldn't it be easier just to repair the trace? I've done plenty of repairs which involved soldering a copper wire along the path of the original trace, fixing it with adhesive, then covered in a conformal coat.

In fact, wouldn't a conformal coat avoid the corrosion in the first place?

[SteveThackery]

For protective components the simple answer is yes.
For a PCB trace, imagine that it's going to a connector, that connects current into a device. Connector gets abused, and one of the traces is broken, then it will go from short circuit to open circuit, which will make spark, which can set the explosive atmosphere on fire.
Basically imagine the worst case scenarios. Like a fan controller (inside the fan) in a coalmine. The standards were written to provide guidance, to be able to make safe electronics devices for these industries.

That's most interesting - thank you.

One of the things I had to consider was the issue of fault coverage, that is, the ability of the diagnostics to detect faults, and what proportion go undiscovered. One problem I can see with what you have described is that the first break will never be discovered, because the fan keeps working just fine on the other trace. Then when the second trace fails, bang.

For this to be useful in a safety critical role, surely you want a way of detecting the failure of the first trace so that you can safely take the fan out of service and repair the broken trace. Am I wrong about this?

This issue of redundancy is more complicated than it first appears. There is always a risk that the standby unit has already failed, and the only time you discover that is when the primary unit fails and the system tries to change over to the faulty unit.... and down it goes. In general, fault detection on a standby system is harder, because it isn't exercising all of its functionality.

A neat way around this is to have load sharing (where "load" can be network traffic, power feeds, whatever), such that when one of the units fails, the other will continue but with a reduced level of service from the system. This reduction is detected and triggers the repair process. It is only useful when a reduced service level is tolerable, but the benefit is that you almost completely avoid the issues around undetected faults.

Sorry! I'm off-topic now, but I would like your thoughts on the risks associated with an undetected broken trace.

[tszaboo]

For protective components the simple answer is yes.
For a PCB trace, imagine that it's going to a connector, that connects current into a device. Connector gets abused, and one of the traces is broken, then it will go from short circuit to open circuit, which will make spark, which can set the explosive atmosphere on fire.
Basically imagine the worst case scenarios. Like a fan controller (inside the fan) in a coalmine. The standards were written to provide guidance, to be able to make safe electronics devices for these industries.

That's most interesting - thank you.

One of the things I had to consider was the issue of fault coverage, that is, the ability of the diagnostics to detect faults, and what proportion go undiscovered. One problem I can see with what you have described is that the first break will never be discovered, because the fan keeps working just fine on the other trace. Then when the second trace fails, bang.

There are periodic inspections described. Including ones that might require to take apart the device, by a technician, and inspect every safety aspect.
I'm not that familiar with that part of the standard, I'm a designer. Someone from the oil, mining or chemical industry will have much better answers for you about how that goes.

[LoveLaika]

I  just wanted a way to be able to reuse boards. If a trace was corroded on only one side, the board could be salvaged if the trace on the other side was still intact.

That's interesting. But, wouldn't it be easier just to repair the trace? I've done plenty of repairs which involved soldering a copper wire along the path of the original trace, fixing it with adhesive, then covered in a conformal coat.

In fact, wouldn't a conformal coat avoid the corrosion in the first place?

We don't have anything like a conformal coating available. It's just easier and cheaper to replace the resistor and/or board in our case. Usually, the resistor legs snap off at the joint. For some boards, the trace gets so bad, they "peel" off (if the board was old and quality was cheap). I think we talked about it, and I can't really remember how it went, but ultimately, this was the path that we chose.

[CaptDon]

How on God's earth do you produce a harsh environment product without conformal coating??? If electrolysis is eating your boards then there is obvious electrical leakage!! How does this pass any kind of testing? We built 1400VDC bus bars in the locomotives that could withstand damp iron powder dust from the mines in Austrailia for Rio Tinto and Roy Hill. Seems like you could get some coating on your product. It pains me to think how under-engineered the product must be?

[SteveThackery]

I was thinking the same thing. The boards are obviously under environmental stress and need protecting. I don't understand LoveLaika's comment "we don't have conformal coating available". It comes in an aerosol for a few pounds/dollars.

https://uk.rs-online.com/web/p/conformal-coatings/0557991

You can get it in silicone as well, if you need a wider temperature range.

LoveLaika's original post talks about duplicate tracks for redundancy, and I was misled, thinking it was something to do with reliability and resilience, which is the usual context. But in fact it's more about extending the longevity of the PCB: if one side is damaged beyond repair, the other side is still available to use, viz:

"... sometimes, PCB traces would break/corrode/wear away as well rendering the whole board unusable. I just wanted a way to be able to reuse boards. If a trace was corroded on only one side, the board could be salvaged if the trace on the other side was still intact."

This seems a remarkable approach, to me. The obvious first thing to do is to protect the boards from damage. In fact, done properly, probably the entire rest of this topic is moot. The idea of traces breaking, corroding or wearing away... well, I'm gobsmacked that the chosen solution is to duplicate the traces on the other side!  A proper engineering approach would be to implement two layers of protection from what is obviously a very hostile environment. The first would be either moving the boards away from their current position to somewhere safer, or enclosing the boards in a waterproof case. The second layer of protection would be a conformal coating. Those two combined would count as a proper job.

Anyway, that's just my advice.

[LoveLaika]

I was thinking the same thing. The boards are obviously under environmental stress and need protecting. I don't understand LoveLaika's comment "we don't have conformal coating available". It comes in an aerosol for a few pounds/dollars.

https://uk.rs-online.com/web/p/conformal-coatings/0557991

You can get it in silicone as well, if you need a wider temperature range.

LoveLaika's original post talks about duplicate tracks for redundancy, and I was misled, thinking it was something to do with reliability and resilience, which is the usual context. But in fact it's more about extending the longevity of the PCB: if one side is damaged beyond repair, the other side is still available to use, viz:

"... sometimes, PCB traces would break/corrode/wear away as well rendering the whole board unusable. I just wanted a way to be able to reuse boards. If a trace was corroded on only one side, the board could be salvaged if the trace on the other side was still intact."

This seems a remarkable approach, to me. The obvious first thing to do is to protect the boards from damage. In fact, done properly, probably the entire rest of this topic is moot. The idea of traces breaking, corroding or wearing away... well, I'm gobsmacked that the chosen solution is to duplicate the traces on the other side!  A proper engineering approach would be to implement two layers of protection from what is obviously a very hostile environment. The first would be either moving the boards away from their current position to somewhere safer, or enclosing the boards in a waterproof case. The second layer of protection would be a conformal coating. Those two combined would count as a proper job.

Anyway, that's just my advice.

I don't know what to tell you. The boards are exposed to water to heat the water up. That's the end usage. I believe I mentioned that before. Only part of the board is exposed to the water to allow a resistor to heat it up. If I recall correctly (again, it's been years), the usage of a coating would impact the heating ability of the resistor, so that idea was tossed aside. It's not like this was in a harsh environment, nor was it for commercial use.


EDIT: Yeah, coating was a no-go given the end usage.

[Terry Bites]

Use a better CAD package!

[LoveLaika]

It was simple once you figured out how to do it.