KB1GTT's design practices
Posted: Sat Jan 11, 2014 12:45 pm
I figured I'd share some of the constraints of my design practices, as it seems that on occasion it's not clear to other folks.
I always engineer to work under what ever conditions I can think of as important. This may or many not follow code and compliance regulatory standards. I'm a big fan of regulatory standards like medical standard EN60601, as it specifies the hazards that have to be protected, then it's up to the design to ensure it's protected. While other standards like NFPA70 and UL establish a set of rules an claim that if you stay in those bounds you'll be safe. I know several engineering failure that have resulted in significant loss of life that followed the code, but were not safe. So as a design spec, I put compliance and regulatory constraints at the bottom of the barrel. They are typically in the barrel, but are generally only used to work as a minimal guide of safety specs. If I have a commercial product that has some kind of potential for inspection or a customer demand, then I'll follow regulator constraints more stringently.
As an example, when it comes to wire selection, I generally don't select based on NFPA70's recommendation for branch circuit wiring, as I'm generally not designing a buildings electrical distribution. I generally design machines. I some times use NFPA70 as a generic guide but that's all. I've read several of the standards they use to establish when a wire is safe vs. unsafe and I understand that those standards don't always apply to every machine out there. Take an EKG for example, where the bare metal is pressed against a human, and the human is electrified. There are places for these standards and places where they don't apply. Or take the example of coil wire, by NFPA70 you can't use a traditional coil wire, it's voltage and current rating are not appropriate per NFPA70. So I try to design to the physical worlds needs, not to a spec. I'll use the wiring standards established by NFPA70 as a guide, but that's about all. I've actually used safely 30A on a 18awg wire in a heating platform design. This is really off base with NFPA70, but was OK with UL.
So for automotive environments and DIY projects what does this mean. For something that's going to be build by a DIY'er, it means the IO should be really rugged as to prevent common mistakes in wiring from causing damage, as the damage can be costly. It means that ESD controls are minimal, so it should be well protected against ESD as that's often not understood by many mechanics. It also means that brain farts happen, and people flip the + rail with the - rail when making a split second decision. To me those are very clear probably situations, and we can protect against them. As well it's easy enough to verify that it's designed against such issues. So those should be design considerations. When it comes to temperature, that's a bit more wishy washy. There aren't any regulator laws that I know of, so it's a design practice that's base on reputation more than anything else. The environment that these will be operated is unknown by the designer. Are these going to be north of the arctic circle, or in the middle of a desert. We don't know. So we should try to keep a wide temperature range, and publish it what ever it is. However we really have to rely on installer and user to ensure it's used in an appropriate environment. I see 130F as 55C, so an ambient spec of 85C seems fine to me. This often means a device will need to be rated for 125C as it will likely make heat and there will be a temperature rise before it can dump the heat to the heat sink at 85C. However where I live I rarely see temperatures above 100F (40C) so I could suffice with a much lower ambient temperature. As well with automotive stuff, you generally have lots of air flow, however with industrial devices they are often in a box which can have an increased ambient temperature, relative to the environmental ambient. So in my take a device that's rated for 85C is good enough, but we really need to measure what it can handle for the max environmental ambient. Then we should publish that spec and let the install / user decide if it's OK for their needs.
Any how, those are just some misc notes about my general design practices.
I always engineer to work under what ever conditions I can think of as important. This may or many not follow code and compliance regulatory standards. I'm a big fan of regulatory standards like medical standard EN60601, as it specifies the hazards that have to be protected, then it's up to the design to ensure it's protected. While other standards like NFPA70 and UL establish a set of rules an claim that if you stay in those bounds you'll be safe. I know several engineering failure that have resulted in significant loss of life that followed the code, but were not safe. So as a design spec, I put compliance and regulatory constraints at the bottom of the barrel. They are typically in the barrel, but are generally only used to work as a minimal guide of safety specs. If I have a commercial product that has some kind of potential for inspection or a customer demand, then I'll follow regulator constraints more stringently.
As an example, when it comes to wire selection, I generally don't select based on NFPA70's recommendation for branch circuit wiring, as I'm generally not designing a buildings electrical distribution. I generally design machines. I some times use NFPA70 as a generic guide but that's all. I've read several of the standards they use to establish when a wire is safe vs. unsafe and I understand that those standards don't always apply to every machine out there. Take an EKG for example, where the bare metal is pressed against a human, and the human is electrified. There are places for these standards and places where they don't apply. Or take the example of coil wire, by NFPA70 you can't use a traditional coil wire, it's voltage and current rating are not appropriate per NFPA70. So I try to design to the physical worlds needs, not to a spec. I'll use the wiring standards established by NFPA70 as a guide, but that's about all. I've actually used safely 30A on a 18awg wire in a heating platform design. This is really off base with NFPA70, but was OK with UL.
So for automotive environments and DIY projects what does this mean. For something that's going to be build by a DIY'er, it means the IO should be really rugged as to prevent common mistakes in wiring from causing damage, as the damage can be costly. It means that ESD controls are minimal, so it should be well protected against ESD as that's often not understood by many mechanics. It also means that brain farts happen, and people flip the + rail with the - rail when making a split second decision. To me those are very clear probably situations, and we can protect against them. As well it's easy enough to verify that it's designed against such issues. So those should be design considerations. When it comes to temperature, that's a bit more wishy washy. There aren't any regulator laws that I know of, so it's a design practice that's base on reputation more than anything else. The environment that these will be operated is unknown by the designer. Are these going to be north of the arctic circle, or in the middle of a desert. We don't know. So we should try to keep a wide temperature range, and publish it what ever it is. However we really have to rely on installer and user to ensure it's used in an appropriate environment. I see 130F as 55C, so an ambient spec of 85C seems fine to me. This often means a device will need to be rated for 125C as it will likely make heat and there will be a temperature rise before it can dump the heat to the heat sink at 85C. However where I live I rarely see temperatures above 100F (40C) so I could suffice with a much lower ambient temperature. As well with automotive stuff, you generally have lots of air flow, however with industrial devices they are often in a box which can have an increased ambient temperature, relative to the environmental ambient. So in my take a device that's rated for 85C is good enough, but we really need to measure what it can handle for the max environmental ambient. Then we should publish that spec and let the install / user decide if it's OK for their needs.
Any how, those are just some misc notes about my general design practices.