It was truly a delight to talk with Alun Morgan at productronica this year. He may be the most enthusiastic person in the field of electronics that I have ever met, as you will certainly understand as you read this interview.
Patty Goldman: Alun, I know you’re with EIPC, but what is your position?
Alun Morgan: I'm the chairman of the EIPC.
Goldman: That means you're in charge, right?
Morgan: Well, not really, no. I’m the functional head but don't actually do any of the work. Kirsten Smit-Westenberg does all the work for the organization, but yes, I've been chairman now for six years, so three terms of two years. So probably it’s time to move forward. But it's been a great honor and privilege to be the chairman of EIPC. Our role basically is to promote electronics in Europe, the whole supply chain, but focusing on the PCB Community, our roots. We have members who design PCBs, who make PCBs, who assemble PCBs, who use PCBs and who supply companies who make PCBs, so the whole supply chain, and our job is to promote their businesses and to keep Europe's profile as high as we can in the world.
Goldman: I know EIPC has a couple of well-attended conferences every year, summer and winter.
Morgan: We do, indeed. We have two every year and we have workshops on top of that. But next year is a very special year for us; it's our 50th anniversary, and the first chairman, Dr. M.G. Fassini, will be with us.
Goldman: You're doing something special?
Morgan: We are. We're going to have a meeting in Bonn May 31–June 1, Thursday and Friday of that week, and we're going on a dinner cruise down the Rhine River. We want to make this a celebration of our existence for 50 years, and a special event for our members, they should enjoy themselves. We have a very strong social program as well as a technical program. We do this for one primary reason, which is networking. We find if we have a good social program, people can network together and make links that they would not otherwise make. Many people tell us they found business partners through these kinds of events. We have members from North America, members from Europe, members from Asia, members all around Europe of course, but we have 28 member-states of the EU. It's surprising the degree of cooperation you can get from these events. From my point of view, probably half of our activity is technical, and half of it would be networking, social, or business-oriented, let's say. So that's the way we see it. We have 120 member-companies now, which is a recent times high and good attendance at events.
Goldman: And when will your winter conference take place?
Morgan: Our next conference is going to be in Lyon, at the end of January. We're going to have a tour at Alstom, which is the large train maker in France. Huge trains. Actually, I was there a while ago, and you know trains traditionally are large mechanical devices so they've got big motors, big levers, and bearings and gears and stuff. But you'd be surprised, a lot of electronics are in trains now. Many of the actuators are controlled by electronics rather than big mechanical mechanisms, and many pneumatic actuators again are controlled by electronics. As we're moving forward of course we see more and more train control systems run by electronics as well. So we have a big place in electronics, not just in automotive, but in terms of mass transit through trains as well.
Goldman: Mass transit probably beyond the trains and the buses and so forth. Everybody's getting more electronics in the transportation industry.
Morgan: That's also a fascinating area. I spend a lot of time now talking about automotive electronics, and the way things are going. Of course, autonomous driving is the end goal, and then the different levels people talk about as defined by SAE International. So level zero, whereby the driver is in sole control going through to level five whereby the vehicle is entirely autonomous and may not even have a steering wheel.
Goldman: Where you sit in the backseat, tell it where you want to go, and you get there.
Morgan: Literally, off you go. But what's interesting is that when you get to that level of course the vehicle becomes very critical. You can't have any failures at that point because you have no backup option. There is no fall-back. If something goes wrong with the vehicle or its guidance system, there is no wheel to hold, there are no pedals to press, there's nothing you can do. When you get to that level you have be very sure that the vehicle is going to take care of you through thick and thin. Beneath that, levels that have slightly more or less automation may have, for example, a 10-second window, whereby the vehicle takes care of everything, but it may have to give you 10 seconds warning that something's going wrong and you need to take control, within ten seconds, of the vehicle.
It means put your newspaper down, stop doing whatever you're doing, and then take control of the vehicle. The vehicle needs to know the state of the driver. In the car now, there are cameras and sensors that monitor the driver and see whether the driver is awake, whether he's asleep, whether he's reading the newspaper, whatever he's doing, just to make sure within 10 seconds they know they can alert him or her to get back to controlling the vehicle.
This leads you to other things. If you imagine a car has collision avoidance, as I'm looking ahead, then it can look ahead a certain number of meters. Typical cameras that can look ahead perhaps 200 meters, let's say, might be the limit of technology. You have to say that 200 meters is as far as you can go safely, and you set a safety window of 10 seconds. So that limits your speed to 45 miles per hour (20 metres/sec) for example. Because the vehicle knows it's clear for that period of time, but if things go wrong it has 10 seconds to alert you, and they must have a 10 second safety window.
Think of this, though. If the vehicle is electronically linked and aware of other vehicles so that you know that there is a vehicle a quarter of a mile ahead driving at 100 miles per hour, for example, your car can read their sensors as well, then you know pretty well the road is safe ahead for at least another two cars. You can then go faster because you have your 10-second safety window by using somebody else's sensor.
Goldman: I can think of some problems with that in the U.S.
Morgan: There are a lot of problems; however, this is known as V2X awareness, standing for Vehicle to X where ‘X’ could be another vehicle or it could be two infrastructures, so it could be two traffic lights or it could be some other, let's say, roadside furniture. The problem then of course arises in how you communicate between the vehicles, because your car needs to know for sure the car ahead is really a car and it's transmitting a valid signal. So you get to the point of "How do you do secure communications between vehicles? " The thing I've been hearing, and hearing quite strongly of late, is that to move on to this level of driving that allows you to go completely autonomous, you need to have this vehicle-to-vehicle communication. It's a must. In fact, there was a projection I heard just a few days ago to say that all vehicles will have this by 2025. Now I think that's a bit ambitious. I'll be very surprised if we get anywhere near to it. However, there's a strong drive towards that, and people in the silicon industry of course are looking at what devices they need to produce to get there. In fact, they're often driving the technology with the vehicles.
The other thing to think about is not just communication between vehicles, but also the liability. What happens if a sensor fails? What happens if the communication fails, and you're doing 100 miles per hour on the road? What do you do? You need to be safe. Typically, in automotive, we talk about a defect-free system—zero defects. But imagine you’ve got an ASIC running this system; it may have 5 billion transistors in there. Defect-free doesn't exist for a 5-billion transistor die. It doesn't exist. So you can't have zero defects on this device. It's not possible.
Secondly, how would you test it? You can't test 5 billion transistors on a device. You can't do long-term aging on the device because nobody gives you the time to do that. So how do you warrant the electronic system in a vehicle that must last for 15 years? How do you that? And the answer is, you probably can't. Let’s look at thermal cycling. A vehicle cycles up and down six times a day. The chip runs at 125°C, for example. Therefore, we just up the temperature, and say “Okay, now we can simulate 15 years of life.” That doesn't really work because you're making different mechanisms of failure over this 15 years of life.
A great example that I will mention came from a chap from ASE Group, the semiconductor manufacturing services company , who said "Take an egg, and look at its thermal endurance. If you take an egg, and you have a hen sit on it at 38°C for 21 days, a chick hatches. But if you want to accelerate the aging and find out how you can make this shorter, take the egg, put it in a pot of boiling water at 100°C for five minutes, and then you get breakfast. You no longer get a chicken."
You've got a different mechanism, and that's where the issue comes along. And I think the concept we will see now is an upgrade path for vehicle electronics. For example, we'll say, "Here's a new car, and you have our vision sensor that will give you collision avoidance ahead. We warrant this for five years because we could test it to five years, and we're certain that a five-year life is reasonable. You want 15 years of the car, but come back in five years’ time and we'll swap out the sensor for you. We'll swap it out to a better sensor, because in five years’ time we'll also have a much better system, and we’ll warrant that for five years. Then come back in five years further and we'll give you another one.”
So then you'll see the vehicle having upgrades, and that's the key thing. As you do upgrades to your sensor technology you'll have the guarantee of the reliability over that period of time.
Goldman: You certainly need the upgrade capability because there's an awful lot of vehicles out there that people aren't going to part with by 2025.
Morgan: This is true. I think that's too ambitious. But the lifetime is 15 years for a car, on average, and there's obviously vehicles around that are 50 years old for sure. But if you're trying to get this into all the new cars, you should get a critical mass coming along. The concept is this, really. You change the tires on the car, you change the windshield wipers on the car, and you change the battery on the car, why don't you change the electronics as well? Why don't you do this throughout the lifetime of the car?
Goldman: As long as they make it modular enough, right?
Morgan: That's a big difference, because they don't make it modular right now. But that is what the discussion is around. In order to ensure reliability and safety of electronics we're going to have to have a different mindset of how we move forward, and that's going to be a big revolution for us. The great thing there is rather than buying a car and leaving it as-is for 15 years or 20 years of life, you will then have an electronic supply chain still supplying that vehicle and a kind of service or upgrade plan over its entire lifetime. And that means more electronics, which is a good thing for us.
Goldman: And the vehicle may have a longer lifetime. There's also talk about how you won't own the vehicle but will simply schedule the vehicle for your use.
Morgan: That's another big discussion, because if you imagine vehicle-to-vehicle communication, you can then get vehicles traveling like trains, traveling all together on the freeway. In the end, what's the difference between that and having a normal mass transit system on a train? What's the difference? The difference is you've got ownership of your part of it, or control of your part of it.
Goldman: Because you want to get off here, and the rest want to get off somewhere else.
Morgan: So you can imagine in the end it's a bit like supply chain solutions. The first mile and last mile really is the tricky part. Everything else is quite straight forward.
Goldman: There's an acronym for that. TLM: the last mile.
Morgan: We'll have the same thing actually, because you know the mass transit will be organized easily, but you want to have the vehicle go to your address, not drop you off at some central station somewhere in the middle of the city. Your imagination could probably run wild. In the end, vehicles typically aren't used all that much. A vehicle spends most of its life doing nothing.
Goldman: You drive it to work and it sits there for eight hours, and then you drive it back and it sits there for another eight hours.
Morgan: My car lives at Edinburgh airport because I travel all the time. My car sits there most of its life. It does nothing. The only thing that's working on the car is the alarm system, and that works 24/7. So that's a different set of reliability concerns than the rest of it. But the car is only working, I guess, 1% or 2% of its entire lifetime. But if you had a car that would go autonomously to other users, you could then increase that usage from let's say 2% to maybe 25 or 30%, or even 50%. Then, of course, you bring other concepts in. Are you still wanting 15 years of life from that car? Probably not, because the duty cycle is much longer now.
Goldman: It depends on how well the engine is built, and you still have to do the regular maintenance, and change the tires.
Morgan: You may find these much shorter and you may scrap the car off more frequently as a result. So therefore the 15-year lifetime which we're used to having in a car for personal ownership, perhaps for a fleet car that was used for lots of other people, would be a lot less. Rental cars don't last 15 years because they rent them out all the time. Rental cars last up to two years before they get replaced. And that two-year window starts to change your concept of reliability and testing for reliability. Fifteen years is a long time to test for. Two years is possible to imagine. Think also of where we have to get to with reliability of cars. People talk about applying consumer technology to vehicles, and everyone I talk to says that's a big jump. Consumer reliability and automotive reliability are far apart. Consumer products have a short design life and if they go wrong, your phone fails for example, but no one dies.
Goldman: You just throw it out and get another one.
Morgan: It's inconvenient for a couple of days, but it's not really a big issue. If your autonomous vehicle fails it could be a very big issue. It’s interesting where we might start from. We wouldn't start from consumer, I don't think. We may start for military because they also have a different concept of reliability. But again, their systems are often built and then lay dormant for years, and then the day you want to use them they have to work. For example, the missile systems. If you're defending your country against incoming missiles, you don't fire one missile to defend, you fire five missiles to make sure. One may fail. One may miss. So you send five. You aren't going to send five cars to pick you up, are you? That doesn’t make any sense. The one sent has to work.
So many people are thinking that the analogy is better looking at, say, the server market, where you've got servers running 24/7 at data centers or attached network storage devices. These are running all the time, admittedly a fixed thermal load, but they are a 24/7 requirement. So that reliability requirement is probably far nearer to what we need for automotive than the consumer electronics would be. That might be another link I might bring into the conversation. As you know, I'm working also with the High-Density Packaging User Group as a facilitator and as a European representative. We do a lot of work on reliability on systems for companies such as Oracle, Nokia, IBM, Cisco, Juniper, a lot of the network companies who are providing this kind of system. I think the jump between that and automotive is probably quite a bit smaller than for a consumer.
We are very keen on getting more engagement from automotive, and I think we're going to have to have this discussion seriously about what reliability means, how we can measure it, and how we can deliver that performance reliability into the automotive sector. I've talked an awful lot there, Patty. I'm sorry, but I love the topic.
Goldman: It's a great topic.
Morgan: It gets me excited.
Goldman: It's all good stuff. Anything else you want to add?
Morgan: Let's just say one thing. Automotive electronics, of course, is very strong in Europe. Here we have some of the very big players. I was at the Taiwan Printed Circuit Association show in Taipei two weeks ago, and Hayao Nakahara gave a keynote speech, and you know he's very well-known in the industry and an acknowledged expert. He talked about automotive electronics. He talked about how this is now having a huge impact in China and China's manufacturing, and the Chinese mindset for manufacturing. So actually, I think automotive is a huge topic now and this opens up massive markets for everybody, and we'll also help our lives as well. If we can rely on transit systems that are autonomous it'll free us up to do other things and make our life easier.
Goldman: The thing is, five years ago we didn't give it a thought.
Morgan: Imagine five years from now where we might be.
Goldman: What else have we not given a thought to that will be big in five years? Electronics is everywhere now.
Morgan: I won't go into detail, but medical electronics is becoming very interesting as well. In terms of diagnosis, because I heard a phrase a while ago, “Diagnosis is the new pharma.” Instead of prescribing medication, diagnosis is going to be the key. And of course, electronics fits that very well. There are now microsensors that can give you a very fast diagnosis on a range of conditions. There are AI systems that actually analyze the data you give them from biosensors, and will then even prescribe custom medications for you for a condition it can diagnose. I think we're going to see control of that taken back into ourselves rather than relying on an external lab or external doctor. We may find a lot of things that we can have managed by electronics as well, which again should make our lives better.
Goldman: There was an interesting thing that Dan Feinberg saw at the Consumer Electronics Show last year where you start to get the merging of industries in an automobile. When he put his hands on the steering wheel (of this concept vehicle) it measured his heart rate level of stress, five or six different things, just from holding the steering wheel.
Morgan: Or maybe if you’re not fit to drive or something. We haven't mentioned IoT at all yet, and the last thing I'll say is that Internet of Things comes into all this. IoT basically is a whole web of sensors around us, like you just mentioned on the steering wheel. They could be medical sensors or other kinds. They could be sensors of vehicles ahead, vehicles around, so this sensor technology is really what we're talking about. This is really wonderful. The degree of autonomy of the sensors is also a big discussion. These would have to be autonomous to some degree because you can't manage them. There'll be billions and billions of them. I'm looking forward very much to this. This IoT revolution, whether it be autonomous vehicles or medical, whatever the application may be, will end up making our lives as humans better. And it's about time that we had some peace and look forward to enjoying our lives rather than worrying about all the tech.
Goldman: There you go. Yes, time for you to enjoy all the fruits of your labor. Thanks so much, Alun. We appreciate it.
Morgan: You're welcome, Patty. It was a pleasure talking with you.