Talking About Microchip's Latest Motor Control Solutions in an Expanded Design Ecosystem

EE: When we talk about motor control, no pun intended, there are a lot of moving parts. There are a lot of different solutions for all of those moving parts that make the world go around. So then from that point of view, we have both incredibly powerful tools now, but we also have very, very daunting challenges now with motors and motor control, we're better at it now, but they're asking us to do more with it, right?Patrick Heath: That's the whole point of the multitude of different support items that we have. Different customers approach motor control from different aspects, and so we try to cover all of those with tools that make their job easier to make spinning the motor easier.

EE: Before we get too deep, let's talk at a higher level for a second. There are people in the industry that aren't even aware that there's a migration in some application spaces back to reluctance motors, right?

Brett Novak: Interestingly enough, I caught myself turning YouTube on, for background noise yesterday, and I think it was Yaskawa, they had a fantastic quick video on exactly that point. And you're correct. We're seeing a phase shift from the old ACIM machine...

EE: No pun intended.
Brett Novak: No pun intended, but the traditional ACIM machines which we've all been very familiar with for years. Of course, ACIM's, until you get into industrial variable-frequency drives, ACIM's were pretty simple. We didn't really have a lot of control. It was on, off, maybe a TRIAC or something like that. And we saw the increase in the variable-frequency drives throughout the 90's and early 2000's. And then we saw this huge shift, I mean, almost a polar shift from the synchronous type machines and the AC style machine, the three phase, the brushless-motors type applications.

EE: Well, that's empowered a lot by software. Isn't it? The fact that we have the software now to drive the reluctance motor, I guess you could say reluctant, haha, properly, isn't it that empowering aspect of it that it's behind that explosion you're talking about?
Brett Novak: That's correct. You get the benefits of the variable speed, the variable torque, or the constant speed and the constant torque, but you're absolutely correct. The age of microelectronics has really allowed us to develop the market. And Patrick, he's, he's been in this for, Oh, what, what were we talking about Patrick? About 20 years

Patrick Heath: 30 years.

Brett Novak: 30 years. That's right. 30 years. Patrick's been doing it for 30 years and I've been more on the power electronics side for, we'll say the last 10 years, 10 to 15 years. I actually came from a background in switching power which is comparable to motor controlling. We do switching power very fast, we're in the hundreds of kilohertz and megahertz range. We're in the motor control domain because we do have a big chunk of iron going around in circles.

We're a little more limited on the frequencies that we deal with in motor control, but that's the migration. The migration path has been very visible. I mean, the traditional... The microchip eight-bit machines, we've had huge success in driving the simpler motors and either synchronous machines, or the brushless, and brush DC machines.

But as the microcontrollers got better and got less expensive, that enabled the design space to really take advantage in cost-sensitive applications. I mean, today's microcontroller, a low cost microcontroller has the performance of a $15 or $20 unit from 10 or 15 years ago. So that really enables cost competitiveness for the manufacturers. And that of course gets passed down to the consumers. We talk about motor control and, and at the heart of story is, I think the last number I saw was I think 46% of global electricity generation goes to motor control. I might be a little off on that number, so we'd have to double the numbers.

Patrick Heath: I would agree with a number similar to that. Think about, just all the motors in... I mean, just refrigerators and HPAC.

EE: Right. Right. And if you look at that as global consumption, of course, one of the things now we want cleaner energy. We want greener energy. You look at at some of the states; New York, California, they want to go completely green energy by 2030. And this means an increased demand on the grids, which we can't go out and just build new grid infrastructure. So what do we do? We increase the efficiency at the end point. So we go from a machine that might've been 60% efficient in converting electrical to rotational energy.

Every percentage point that we gain is a percentage point less in the global dominance of electricity for motor control. And let's face it. I mean, there're more things with motors, electric motors coming out every day than there are things with internal combustion. If you look at the automotive market, for example, Tesla's migrating to reluctance motors. They've already migrated to reluctance motors.

Brett Novak: They've already migrated to it. Yeah (affirmative). The new Teslas use the synchronous machine for the drive train.

EE: Let's get a little bit more granular now and let's give Patrick a chance to give us a little bit of a viewpoint because from your history, Patrick what I'd like to let you give us an opportunity to let us experience through your eyes is, think about the task of qualification and specification as it is migrated. Today we're almost in a time of eternal test, eternal system awareness, because if you're designing in a software simulation package and you're building Six Sigma, and you've got a test station after every stage of manufacture, and then you're in the field and you're using the internet of things and getting telemetry back from your stuff. In essence, you're constantly testing, or you could almost say you're not testing at all. You're just aware. How would you address that? Because I mean, motor control is very specifically an area that involves a lot of tweaking or used to, and maybe not now, but maybe just in a different way.

Patrick Heath: I see two areas or two trends that touch on what you're talking about. The first is something that is, in a big thrust for us at Microchip, and that's what we call functional safety. Or FUSAF for short. So functional safety was born out of perhaps out of the automotive industry, but it certainly has expanded, very recently, to many other industries. Industrial medical, and even consumer and functional safety has a bunch of requirements on the motor control device itself, and the MC or the digital signal controller, as well as the software that you're running to ensure that things are going as planned. 

So there's code that's running in the background to check, to make sure that hardware elements of the processor are still functioning correctly and that the signals from the motor are still within the range barrier to be expected. So Microchip now provides a bunch of collateral to support that. Things like FMEDA users, diagnostics, software, and users manuals about how to tie all that together to support certain levels of certification like ASIL B or ASIL C types of applications. And so that's a big thrust for us and Microchip.

EE: when we talk about this advance that came out with the release, would you say that it's an evolutionary aspect? I mean, was this something that was expected?

Patrick Heath: Partially, yes. Motor control has been a real, for us, a real focus for us for more than 15 years here at Microchip. I started in 2001, started directly doing motor control type of development here. And so over those years we've created lots of different families of devices that focus on motor control, as Brett already mentioned, our dsPIC is currently 16 bits, the digital signal controller as been focused on motor control, as well our 32-bit motor-control devices. We have different families at different price points, different performance levels to address the whole spectrum of motor control applications out there.

What we have in the press release, we introduced a very low cost dsPIC family to try to push down the cost of field-oriented control. This supports the migration of people that have gone from, maybe originally they were AC induction motors and moved to brushed or brushless DC and have asked for, or needed more efficiency. Their customers are yelling for higher efficiency. And so they moved up the chain to PMS in motors, or even IPM motors, which provide even more efficiency. This new family, a distinct family of digital cleanup controllers, provide the ability to run those PMS and IPM motors, running field or into control, which is the king of efficiency as far as motor drivers go.

EE: The aspect of the evolutionary aspect of the migration aspect expected the aspect of more power, more functionality at lower and lower levels. Eventually there'll be a baseline expectation of functionality regardless of the application's pace. You know what I'm saying? 
Patrick Heath: At some point our customers realize that it's basically the costs of doing field oriented control is so low now that there's no point in using simpler algorithms to control the motors. The trade-off for it, the lower efficiency is, it's just not worth it.

EE: You're not saving enough money.
Patrick Heath: Yeah. You're not saving enough money to warrant it. We introduced an algorithm called ZFMT, zero feedback from torque, as part of our press release. This algorithm works with field-oriented control to basically remove something that impeded people from switching from six-step block commutation to field control. It provides the maximum torque out of the motor at startup and at low speeds, which was a weakness of field oriented control.

By combining this new algorithm with our field-oriented control, we can get the best of both worlds. So we've really removed all the hindrances for customers to jump from six-step commutation. You know, they can hitch those call centers, and wires, and cables and save all that money, just to senseless deals or to control super high efficiency, as well as maximum torque, even at startup at the motor. And so we're providing customers with this capability that was, as Brett said earlier, what a cost, at $15 or $20 a DFP to do 10 years ago.

Brett Novak: I'm actually going to add on the second part of that question that you'd started with. One of the other things that we see becoming... I want to use the word more model-based, in addition to what Patrick was saying about performance, and of course every time we increase the performance of MCU, that gets passed along to the designer as, "I can do more." So as Patrick mentioned, the functional safety aspect of the dsPIC with the dual-core machine. You can segregate your functional safety to one core, with your motor control on another core.

So you have a nice clean split system in a single core device. The easiest way to do that is you drive the frequency up so you can do more in the same amount of time. But one of the other things, and you had sort of touched on it with your question, one of the other things that we're seeing acceleration in, is the move to... I'll use the traditional term burn and learn.

You burn it, you spin it, you test it, you go back, you write some code, you spin it, you test it. Every time you're doing that, you're rewriting your code on the device. What we're seeing now is real-time hardware in the loop. We have third-party partners such as MATLAB support for these devices. And you can do real time simulate. You can take your simulated model and then directly compile code to your microcontroller, and at the same time have hardware in the loop.

So you're doing real-time tweaking of your motor application, instead of going through the regimented steps of, "Okay, I'm going to go tweak this knob over here, and that knob over there, and see what happens. I got to burn it and, and start over." We are seeing... again, I mentioned MATLAB, they support both the 16- and 32-bit devices in that software environment. We also have a partner which is LCM and LCM has what's called Psylab. For both Patrick and I, for the 16- and 32-bit, the demonstration software, we use a real-time plugin as a back communication channel.

It plugs right into MPLAB X and allows you to talk directly to the hardware. You can have a motor spinning and see all of your parameters come back in real time. So, if you see a quiescent current on one of your phases that seems out of place, you can go in and tweak your settings to bring that under control without having to go through every step of rewriting the code, programming it, and retesting it. It's all seamless in line.

EE: This speaks exactly to that point that came up earlier. There's actually an anecdote I want to bounce off both you and Patrick on this, and get your reactions. We were talking about Tesla. Let's talk about another one of his creations and that's SpaceX. They were looking at the old Saturn  Rocketdyne F-1 Engines. They were told that it's impossible to remake them. And they're like, "That's insane. We have the diagrams here in front of us. It's all documented. Every single thing."

The one thing NASA did not document was, as you said, the turn and burn... These engines were designed on paper, but they were built in the ongoing learning flow, as the engineers tweaked and dialed in the subsystems in each of the systems, like an old Mercedes-Benz. In the old days they would make the car on the assembly line, and then a team of white-suited engineers would dial the car in perfectly.

You couldn't make that finished Mercedes on the assembly line. It was impossible. You needed to know what tweaks they made to that vehicle. And the same thing with the F-1 engines, they have the plans, but they don't know what the engineers did in the process of tweaking and development to literally almost hand build each of the engines that went into that spacecraft. Whereas today, a SpaceX engine is one of the most efficient, beautiful pieces of hardware ever made. And it's all designed with a team of very gifted people, but they're using advanced software tools. And so they don't even have to... That's a foreign concept to them: Test to destruction.

Brett Novak: Right. It has gone from physical tests-to-destruction or physical tests to limit, to simulation, to pass the limit, and then back it off a little bit and try it. So you've shortened the design cycle, while at the... I can't really say shorten the design cycles. You've increased the efficiency of the design cycle. Not having to go through all of the iterative steps of the downstream testing. And of course with the algorithm development, Microchip, we develop algorithms and some customers take our algorithms and plop it on a part and off they go, they're off and running. Most customers will take an algorithm and they will add their own secret sauce to it.

They'll go in and tweak things and manipulate the code to their liking, which we allow it's primarily open source, open source code as long as it's on our devices. But really the simulation aspect, especially in today's consumer-esque market, a lot of the motor control, a fair majority of the motor control we do still see is in the industrial domain. Pumps, fans, all of the manufacturing equipment, heavy construction equipment, things like that. But as Patrick mentioned, with the cost of the microcontrollers coming down, the cost of the overall system comes down as well. And it puts higher performance in the hands of the consumer.

Now, I won't use a direct name, but we'll say a large vacuum cleaner company, they went from a traditional AC motor plugged into the wall, to a battery-operated six-step, and the price point has gotten to the point where they can now put a brushless DC field oriented control, no sensors, we're doing everything via back EMF. They've put that in directly in the hand, literally, in the hand of a consumer where, and they have accelerated design cycles. What was a two to three to four year design cycle previous, through the simulation, through the advancements in software, both the development on the supplier side and the development on the customer side or the designer side, they've really been able to shorten their time to market, from that three to four years to 18 months.

Motors are a tricky things and they're physical things. They're real things. I'd like to mention one other part of the press release that we did, and that's for a tool called motorBench. Now, motorBench is a GUI that runs on your PC or your Mac, but it also connects by a serial cable to one of our development boards, which is connected to a motor. What we found is that every motor is subtly different. Even if you take a motor that's manufactured on a high-volume manufacturing line, and you pull two of them out of the line at the end, and you compare those two motors, you'll find that there're small differences here and there in the manufacturing on the diameter of the wire, the number of windings round, a stator or a pull, or, in the thickness of the aluminum somewhere.

 And these all end up making some difference in that motor and how it's controlled. And so we have this tool, this great tool called motorBench, that will, in five or 10 minutes, if you hook up a motor to one of our development boards and run this GUI, it will go through a process, a three-step process that starts with what we call self commissioning. And in self commissioning, it will extract the motors, electrical, and mechanical parameters from the motor and board in the system and capture those parameters. Things like the state of resistance, the inductance of the DNQ axes and the back EMF constant and things like this. And so it extracts all those parameters in step one. And in step two, it will tune the motor using, those parameters, it will tune out the motor for sealed out controls.

EE: Would it predict damage if, if I had foreign object damage in it, would it be able to detect it from the drag on the motor?
Patrick Heath: Yes. Thethe state of resistance would have increased, and you would see that. Let me touch on that in just a little bit later, because I think that brings up another aspect that we're focusing on, which is kind of machine learning, which is looking at temperature, vibration, or noise flooring,

Brett Novak: Everything.

Patrick Heath: Resistance to predict when motors don't have a problem and send out, and call the maintenance man to go take a look at it. Right. But motorBench is the initial production line for the motor, to whoever them making the manufacturer is using that motor. And, they can run this tool and extract the parameters at two to three controllers for FLC, and then provide the code that will run the motor.

This whole process takes five or 10 minutes. And it used to take a couple of engineers sitting around doing, as Brett said, the burn and learn routine, could have taken them weeks to do before this, right. To try to get the first run of that motor using field-oriented control because of the tough algorithm. It's very difficult. But with the tools that we have now, any engineer, even a novice can sit down and look up the board, look up the motor, press go, right. And in five to 10 minutes later, they have a running system using field oriented control for that particular motor, with all of its fine intricacies and differences from every other motor out there in the world that this software is tuned to run that motor for field oriented control. Ready for them to start playing with.

I say start playing with it because he can't just take the first path. You have to do all sorts of testing and say, "What if the voltage, tide, the current and the speed as well, and it basically compresses the amount of time that it took a couple engineers. So weeks to do down to just five or 10 minutes now.

EE: Increase productivity.
Patrick Heath: This is the power that Brett talks about. The productivity and efficiencies of time have been increased now because we've been able to push... With the march of technology, we've been able to lower the cost of entry into this, the highest level of performance now.

EE: Before we close, turn off the recorder, why don't you each, give me a little bit of a closing thought to leave with the audience? Brett, you go first.
Brett Novak: So I think the best closing thought from my aspect is the beauty of the Microchip model is we do have support for, and Patrick and I use, we use the pyramid, the pyramid of motor control when we discuss this. The beauty with Microchip is we have the experience. We've, as Patrick said, we've been doing motor control for 15 plus years. And with that experience also comes the ability for the customer to pick and choose the solution that fits their application. 

Patrick Heath: Having lived out through that for the last three decades, it's been an exciting ride. Well, I think my closing statement would be; Microchip has been committed to motor control for over 15 years. As Brett says, we provide a wide range of control options, from eight to 16 to 32 bits. But we've also provided a wide spectrum of development support for our customers.