30kW High Voltage DC Smart Inverter and Progress on Tesla Model 3 Batteries.

Our adventure in solar began about three years ago. I kind of had a sense of what I wanted to do and you all have misled me. We originally used a Tesla Model S full pack and a Sandy inverter that would do, I think at the time, 10 kilowatts of AC, 240 volts.

And it worked from 300 to 400 volts input. Most of our viewers did not want to deal with a full battery pack almost entirely because of the expense and they didn't want to deal with a high voltage inverter almost entirely because of the expense. And they wanted to put a tow in the water and take two Tesla battery modules and a cheap 48 volt inverter and try to make a solar system out of it.

And I kind of acquiesced to that, sympathetic with that desire. And we did do a Tesla Model S module controller that is our most successful product today. And we did locate an inverter by Sigineer somewhat better suited than what they were trying to use and enable this low end solar.

It's proven dissatisfying to almost everyone who's tried it because it's not enough power and it's not enough storage, which I knew in the beginning that it would be. One of the issues while I went along with that was the announcement of the Model 3 automobile, which was some distance off, but I knew would be produced in much larger quantities than the Model S or the Model X. And that's proven to be the case. I still think that the correct model is using the high voltage battery packs directly as they come from the car.

And that requires a high voltage inverter. And so that's what we're kind of working back toward. What are some of the reasons? The 48 volt systems require an immense amount of current to produce useful quantities of electrical power.

12 to 15 kilowatts being kind of a minimum, well, that's up in the 400 amp range. And your I squared R losses due to those high currents are high, but more importantly, they're somewhat dangerous. And when I say dangerous, I don't mean you're going to get electrocuted.

I mean, they're going to start a fire or they're going to burn something up simply because of the heat generated in the conductors. I don't like dealing with those kind of high currents, but I don't mind dealing with high voltages at all. I was taught the Marine Corps voltage check when I was in the Navy, we had marines, and they would check voltages like this and after a while you get used to it.

So we're kind of back. One of the central tenets of what we're doing involves the use of remaindered salvaged batteries from electric vehicles to provide storage for solar energy. Tesla has made it very easy for me in that I can safely ignore Chevrolet, Volkswagen, Nissan Leafs, and so forth.

Tesla makes a quite better battery in much higher capacities and they sell more electric cars than everybody else put together. And so all I need to do is focus on the Tesla Model 3 battery. We developed a controller for it and that went surprisingly well.

We can read all the cell voltages even on the later batteries now. And we've developed a controller much like our Model S module controller, but for the entire Model 3 pack. We got one and we're powering the shop with it right now.

But we got another one to use as a test thing and a guy bought it from me before we even had it completed. So we got two more and another guy bought those from me before we could get them working. So we've sold three of them literally before we could get them out of the crates and get them thoroughly tested.

I got two more in this week, and we already have them working. One of the things that all three of the first ones wanted was our 6.6 kilowatt Chinese TCCH charger. To charge it from the grid or generator or whatever.

And so we've been playing with this one this week and we have found a way to use the Model 3 PTC heater and AC compressor high voltage connector on the battery and it's associated cable right into our 6.6 kilowatt charger. Which the first three we had kind of routed into our controller. But we have contactors inside the Model 3 and we have a perfectly good connector there and so that's what we're playing with today.

I'm very optimistic. I am enormously excited by what I think. I'm sitting in my little shop working with what I think will be one of the huge inflection points in the history of humankind.

I struggle a bit. First, this is a picture of me being really excited. That's what a really excited Jack Rickard looks like.

I'm sorry. I'm not a very expressive person. Somewhat saturnine and somewhat skeptical of everything.

Particularly nocelliums, which you call no-see-ums and are now in our environment. Boy, these little buggers bite like a mother. Anyway, I'm very excited about solar energy.

One of the biggest struggles I have talking to you about it is that you already know all about it. You've got a full set of information about solar and how it works. You have a full set of of reasons to be interested in it and opinions about it.

The overwhelming thing is your concept of alternate energy and green and most particularly global warming. Perhaps it was because I was born on July 24th of 1955 in Missouri, and it was 105 degrees Fahrenheit that early afternoon when the occasion of my advent on planet Earth occurred. But I don't care about global warming.

I don't view it the way you do. I don't view it the way your opponent do. I am left out of the conversation somewhat disabled, mostly by you.

But suffice it to say, you can't make me care about it because none of it's real. It's not in my task list to persuade you of any of that. But we do share an interest in electric vehicles and solar power and driving on sunshine

That is an accident. I'm not one of you. I don't want to save the planet.

I think it's actually the loveliest planet in the universe and God's creation. And I am commanded to spend my time here on this earth in what little way I can, assisting to its perfection and saving it from global warming. It's not on the list.

However, solar is. For 25,000 years, mankind has sought to improve their lot and increase their wealth and their ability to feed themselves by manipulating energy, starting with firewood, deadfall from the forest. This is actually the result of a photosynthetic process where the sunlight converted by vegetation trees primarily into take our air and our water and primarily our water and our sunlight and to rearrange things a little bit to form a carbohydrate, which you can store as food for the plant.

Well, if the plant dies, it doesn't need it much longer, but we do. And we burn those carbohydrates to cook our food and to stay warm. And that was the beginning.

If you rearrange that just a little bit by putting it in the ground for a million years, now you get a hydrocarbon, coal, oil, natural gas, and we can burn those. But the highest order of our achievement has been the development of electricity. But it's a really ugly process of burning a hydrocarbon to heat water, to make steam, to turn a turbine.

And the turbine turns a alternator and the alternator produces electricity, which we then wire to each other across the entire continent on cables. This reduces our problem of transporting fuel because you can transport it instantly and it weighs nothing. You do lose 10% in the wires.

But so all in all, it's a magical epoch based on electricity and it's only 150 years old at most. But there's a bypass and we are on the cusp of it, at the very beginning of it. You all think this is a done deal.

It's not started yet. It's right at the very beginning. And the bypass is we take the sun, you keep the water, and we're going to convert it directly into electricity, eliminating the transport of the fuel, the burning of the fuel, the heating of the water, the turning of the turbine, which in turn turns the alternator, and then all the transportation.

We can put photoelectric cells directly on the roof of where we need to use the electricity and we can convert sunlight to electricity directly. This was first and best described by Albert Einstein. And you'll be shocked, perhaps some of you, to learn he did not win a Nobel Prize for the theory of general relativity.

And he didn't receive it for the theory of special relativity. He received it for a description of the photoelectric effect. Probably kind of a backhanded version of anti-Semitism, but that's what he got the Nobel Prize for.

Today we're right at the cusp of this. 22 years ago, by the way, I installed at an expense of about $275,000 a then enormous 12 kilowatt array at my home in Morrison, Colorado. So I'm not new to solar, but I recognize that with all of it you've seen, with the utility grid scale solar systems going up, with the huge amount of residential solar in the United States, that it hasn't actually started yet.

This past month, Lawrence Livermore Laboratories has introduced or published a paper on a multi-layer, multi-junction photocell, power cell that will operate slightly north of 47% efficiency. Our most efficient commercial cells now are about 22.5% efficiency. In their paper, they alluded to a ready and imminent path to exceed 50% efficiency using this technique.

And this is not the end of the road, guys. The quest for efficiency in photovoltaics is ongoing. But you know what? They published a thing, they publish every year actually, called the Estimated U.S. Energy Consumption in 2019, 100.2 quads.

Same Lawrence Livermore National Laboratory. I'm going to put it up on the screen here. And the first thing I would note is that out of all energy sources, solar is 1.04%. It's barely detectable with test instruments that it exists.

But the reason I refer you to this diagram is the curiously inelegant way we manipulate industry energy. We have solar, nuclear, hydro, wind, geothermal, natural gas, coal, biomass, and of course petroleum all accounted for here. On the right side of the page, you'll see energy surfaces at 32.7 and rejected energy at 67.5. We did not reject this energy because we don't love it.

When they say rejected energy, they're talking about wasted energy that is given off in all the processes to produce and use this energy as heat, waste heat. When you run up a coal-fired plant or a nuclear-fired plant or a gas-fired plant, you have to eat the water, but most of the heat just escapes and is given off into the atmosphere. Very little of it winds up being electricity.

When you put petroleum in your car, let's see, what are they saying here? Oh, well, that can't even be right, but I figure you probably get 15 percent forward motion out of 100 percent of the energy that you put in your car in the form of gasoline. With electric vehicles, you can convert much more efficiently up to 85 or 88 percent of the fuel in your batteries to forward motion. That is a huge gain in efficiency, a huge gain in efficiency, and a huge decrease in heat loss or waste.

In fact, I view an internal combustion engine vehicle as a power plant designed to blow heat out the front of the car, and forward motion is a byproduct of that process. An electric vehicle more directly, more elegantly, and more efficiently converts energy into forward motion, which is what I want, but also radio, internet access, the ability to honk the horn and keep me warm or cool. It's used energy services as opposed to rejected energy, and so you can see from this diagram that two-thirds of our energy is wasted as heat.

If you're worried about global warming, you don't need to invent a CO2 fantasy. How about just the heat you give off in all these activities from the tree to the turning wheel or the air conditioning that you use? It's a huge amount of wasted energy. In fact, you're wasting most of it.

No, you're not wasting most of it. You're wasting two-thirds of it giving off heat, and that is a wasteful and crude and inelegant use of energy. It would be a good mission for my life and I would suggest for yours to change that from two-thirds wasted and one-third used to one-third wasted and two-thirds used.

In fact, if we could get the world to be 90 percent energy efficient, we would be that much wealthier. You see, energy is our base wealth, and money and houses and cars and clothes all that are proxies, one-off proxies for our ability to manipulate energy. And right now as of 2019, we're doing a piss poor job of it.

We waste two-thirds of the energy we harness, and we benefit from one-third of it. There are efficiency losses in photovoltaics as well, but they're much less and they don't give off much heat. We're converting the sunlight to electricity directly and that's a cool thing and it would be cooler at 47 percent than 22 percent.

And so, this is why I'm saying we're at the cusp, the very beginning of a revolution in the generation and use of energy. And I believe that it begins at home and I believe that it starts with energy being generated where we use it and use there. But that of necessity, because of its photovoltaic nature, firewood burns day or night, but the sun is not overhead day or night.

In fact, some days it isn't overhead, and we haven't had much today. But that's where the batteries come in and the ready source of lower cost, less expensive batteries of the very latest architecture will are now and will be in the future left over from electric vehicles. The reason I'm not totally dedicated to building electric vehicles anymore is we already won.

Tesla is producing millions of them. Others will follow. They probably won't be Volkswagen and General Motors and Ford.

Boy, in my life, two years ago, I told you all those companies were on the very edge of collapse and would be bankrupt and everybody would get to carry their desk adornments out to the parking lot in a little cardboard box. This year, everybody knows this. There was no prophecy to it.

Anybody knew that. Two years ago, it was like, he's a madman. That can never happen.

Today, it's like, oh, everybody knows that. GM is about to go out of business right now, so is Ford. Well, yeah, they are.

So what are we about today? Back again with the Model 3 battery, but I'm unhappy with the availability of high voltage inverters, particularly because I've become enamored of the Enphase microinverter. Note that a little over a year ago, I bought some Enphase stock at $5.90. It hit $63 this week and is quickly becoming the 800-pound gorilla. In the grid-tied inverter world, 97% of the solar installations in the U.S. are simply grid-tied where they try to use the AC grid as a storage battery.

It never did make a good one, in fact, a piss poor one, and now we have an open war going on between the grid and the solar people, leading to what we call selfish solar. Cut them off. Don't give them anything.

Don't sell them anything. The 46 bucks a month you might get for your $80,000 or $100,000 investment is not worth it, and so we're talking about selfish solar. Grid independence.

We keep a grid, so if our system goes down, we've still got one, or if we have seven days of bad flooding, um, we can top off our batteries with it, but we don't send them anything. We don't sell them anything. Um, that's why it's called selfish solar.

Your dream of a kumbaya moment where you and the utility company walk off into the future hand-in-hand with a very delicate music playing in the background didn't pan out. That's not the way our corporate world works now. Worse, they have robbed the bank, um, taking all the money out of the grid and giving it to retirees in the form of dividends, which their salaries are based on, and have let the grid devolve into a teetering, doddering old wreck that's kind of falling down around our ears, slow motion, and it's not going to get better.

So, I see a future of point-of-use generation and use of electricity through the magic rocks of photovoltaics and lithium batteries, and I'm kind of committed to that, um, without taking on all the, um, fantasies and pseudo religions of global warming and peak oil and all of that. You don't need all that. It's right here.

Two-thirds of your energy is squandered like whores, sailors in port towns. One-third actually does you any good. Can we improve on that? And I believe we can.

Again, I've been frustrated about the inverters. Sandy makes pretty good inverter. They're heavy.

They take weeks to ship. Now there's a 25% tariff on them. Um, it's just kind of a shit show, and they don't do what I want them to do.

I cannot talk these people, the Sigineer people, the Sandy people, anyone into doing an inverter my way, and the central issue is the grid-tied inverters all now use UL 1741 Supplement A, which allows me to very, uh, in a very fine way modulate their power output by varying their frequency. But Sandy won't do it, and Sigineer refuses to do it. They, I don't know what I'm talking about, and they're not going to do that.

And by the time everybody's doing it that way, then they'll wish they had listened. But it's very difficult to persuade people about the future without the future arriving. And then their story is, well, everybody knew that.

And so that's just the world I live in. If you recall, Colin Kidder joined us here, as he does rarely every year or two. He's worked for me for six or seven years now.

A brilliant guy, and a lot of what we do around here, he's kind of at the heart of. And I was complaining about the inverters, and he happened to mention that, you know, he kind of liked inverters, and would kind of like to have one at his home in Sparta. And he'd like to do an inverter.

I said, really, you have an interest in it? Oh, yeah. I said, well, you know what? He says, the problem is you blow up a lot of expensive IGBTs. I said, you know what? I know a guy who just loves to buy expensive IGBTs.

This guy. I'll buy him just to look at him. So we have since corresponded, as we do daily anyway, on the topic of a massively overbuilt, over-engineered, overkill 30 kilowatt inverter that will work on the voltages of the Model 3 battery.

By the way, these voltages did not arrive on a pallet from heaven. It's no accident that electric vehicles overwhelmingly operate between 300 and 400 volts. If you rectify 240 volts AC, which is the overwhelmingly almost universal power that we have here in the United States, you get about 340 volts.

And if you will do a battery and charge a battery and discharge a battery, somewhat close to that, the expense and size of the components to boost that up to 400 volts or buck it down to 300 volts is very minimal. And so the tie between the 300 to 400 volts and the 240 volts AC is pretty natural. Similarly, if we take 360 volts and try to invert it, that's a lot easier, a lot more efficient, and in all ways better than trying to do that with 12 volts or 36 volts or 24 volts or 48 volts or a thousand volts.

And so the 300 to 400 volts is not an accident for the EV batteries. It comes from the 240 and it'll go right back to the 240. So we started playing with the idea of an inverter.

Hey. Hi. How are you? Good evening.

So what's the plan? You jump in there, the drought ends, and then the crops grow. Everyone's happy. Right.

Okay. What if instead we built some sort of system that would get the water from the river down to the fields, like with tubes or something? A gravity-driven water distribution network. Yes.

Huh? You got it here? Um, no. Oh. Well, everybody walked all the way up here, so... Go on, get on with it.

In you go. Yeah, no, you're right. I liked him.

You like everyone. Well, I'm a people person. Eventually, somebody realizes there's a better way to do things.

So we get to start an inverter with a clean sheet of paper and design it any way we want to, and it can have any features we can code. And that's kind of interesting, so I'm sure it will surprise you that our first two prototypes, one with Colin in Sparta and one with me here in Cape Girardeau, Missouri, are 48-volt inverters, 15 kilowatts. Why would that be? Well, the trick is in the transformer, and I happen to have a couple of Sig and Ear 15-kilowatt transformers here, and that's actually kind of the tough and central part of an inverter the way we would do one.

Ziqi has negotiated the construction of three other inverters that will operate at 180 volts input and 240 volts output at 30 kilovolt amps, kilowatts, and that will be our second prototype. But it takes a couple of months to get rid of these nociliums. But no, it takes a couple of months to get the transformers built and shipped here, and they're 800 pounds a piece, so they have to come by sea freight.

And so we're going to go ahead and do a prototype at the lower voltages, 48 volts, and do our experiments and code and so forth with that, and we can make a full-fledged 15-kilowatt inverter using the Sig and Ear transformer and put a lot less stress on our switching semiconductors. And that takes us to the next thing, and I'm embarrassed about this to the extreme, and Colin makes fun of me, because these are about 900 bucks a piece, 850, 900 bucks each. This is a Wolfspeed silicon carbide MOSFET half-bridge module.

A MOSFET is a metal oxide semiconductor field effect transistor, and this is the latest in switching technology. Circa November 2019, it was released. It's very expensive right now, but very new, and it's quite high power.

This will do up to 1200 volts and up to 450 amperes, and that's over 500 kilowatts, which is a lot more than we need, as Colin has pointed out quite correctly a number of times. But overkill is always appropriate, and I am very drawn to any new shiny object or bubble in the universe, and this is just pretty damn cool. It has a on resistance between the drain and the source of about 2.7, 2.4, 2.7 milliohms.

Now what that means is that if I put 450 amps through here at 1200 volts, that my total loss in forward conduction would be a little over one watt, 1.17 watts. With switching losses, it's somewhat higher, but this will run cool, and we have so much overhead. Even at 48 volts, we're never going to run it at 480 amps, 450 amps, and we're, of course, a long way from the 1200 volt limit, and you mostly blow these up with ringing and noise and back feed and so forth that gives you voltage spikes, and it's going to be tough for us to get to a 1200 volt voltage spike with that switching.

It also diminishes the amount of heat sink we have to provide for them. They're very small, and it's the latest and greatest silicon carbide switch device, and so I fell in love with it. It's stupidly expensive.

It requires its own gate driver board, and that's stupidly expensive, and then we had to develop a shield for our ESP32, which is also stupidly expensive to drive the gate driver board. So you have to have a board to drive this, and you have to have a board to drive with a differential pair for the high and the low side to drive the gate driver board, and, of course, a microprocessor to process all of that, but that's what I've selected on the input and the output, both sides actually. One of our viewers sold me his attempt to build our PowerSafe 100.

He got in the middle of it and just gave up and sold me the batteries and the enclosure and so forth, but he had two of these. This is a temporary service box by Siemens. They have kind of a subline called Talon, and this is a 137 US.

It has a NEMA 5-1550 amp, a 30 amp 240, and a 120 volt AC GFI outlet and a couple of circuit breakers here. The 240, I think, is 80 amps adequate for the 15 kilowatts, and we put a midnight solar metal oxide varistor surge suppressor on the output, and all that is repeated exactly on the input, and the idea is that, like the Sigineer, we can have a grid input and run either off the grid or off the inverter, but I've got long-term designs which may prove difficult, but we're going to try it. That is to link this output to the input of the next one, where you have two 30 kilowatts operating at the same phase and in conjunction with each other, and in fact, you could take the output of that one, connect it to the input of the next one, and I have 90 and so forth up to some point, and so that's our basic enclosure.

This is the same enclosure we use for the PowerSafe RV or PowerSafe 15, which has been quite popular, by the way, 24-volt system, little six kilowatt inverter with a crummy waveform, I think, but man, we've sold a dozen of those things. They like them in the RVs and boats, cabins, but we're using the same enclosure. Let's take a look inside.

So inside, Dan, can you pull that yellow thing off? We have a bus bar, and we can connect some things to that, but most notably, we'll have our battery input at 48 volts that goes to that bar. It's connected by copper bar to a 477 microfarad film capacitor. These are kind of, again, pricey overkill capacitors, film capacitors.

They cost about 200 bucks, and we may need a couple more of them, which I'll probably just connect to the bus bar. Here's our two switches on a heat sink. This is the gate driver board I was talking about, and there is our positive and negative battery input, and this goes to the transformer as an output.

And so that's our basic copper to connect our WolfSpeed XM3 series silicon carbide switches. Each module is a half bridge, and we've got them connected to the DC inputs, which would be high current, so we actually have three pretty big conductors going to the transformer primary. And so that's the mechanical layout of our switches, gate driver boards, and the output of the transformer goes over here to some bars.

We have a neutral and an L1 and L2, and I have some bars over here, and then we connect that to our output box. The input box feeds into two contactors, so we can completely isolate the system from the AC grid or, in some cases, the previous transformer, but we can connect it here for sensing of the voltage and phase, regardless of whether the contactors are on or off. If they're on, we really only bring L1 and L2 in.

We do not bring in the neutral, you will never bring the neutral in from shore power or from AC grid, and that's because eventually you're going to connect to a loads panel where its neutral will be tied to ground, and if you have your system tied to two different grounds at the same time, particularly in a boating situation, but an RV situation, but any situation, you don't want to connect to have two separate ground loops. And so this is the transformer. This one's 132 pounds.

I think the one that Ziqi has on order is like 287 pounds. Maybe it was 187, but it'll still fit in the box, I think. But this is our 48.

It steps up nominally a 31-volt AC sine wave from our switches to 240 volts AC output. The other transformer will do about the same with 180 volts RMS to 240 RMS. And when I say 180 volt RMS, so we're up at 250 volts a peak, and so a high voltage peak-to-peak value on the primary.

And so it'll be less of a step up and somewhat more efficient in the transformer. And so mechanically, that's what we've got laid out. We have to add our controller to drive the gate drive boards.

And right now, we're kind of leaning toward having that being a dedicated controller that communicates with the second EVTV ESP32 CANDue, which has a sensor shield on it. And the sensor shield will measure our current and voltage of our output, our input here. And it'll also take care of our Wi-Fi communications, Bluetooth if we want to communicate with the display, and talk to our battery controller.

And we'll be able to do CAN and so forth. So we'll have two microcontrollers in here, 240 megahertz controllers with Wi-Fi and Bluetooth capability. And it kind of has a built-in PWM function, Pulse Width Modulation function, that we're taking advantage of.

What is Pulse Width Modulation, and how do we drive these switches to make the sine wave output for 240 volts AC? Let's take a look at that. And I want to do a long, boring, and stupid description of that just for you, and just because I can. All right.

We need to produce a sine wave output out of our inverter. And we're going to use Pulse Width Modulation, of course, to do that. And let's talk a little bit about what Pulse Width Modulation is.

If we set up a square wave output at 20 kilohertz frequency, the duration of each waveform will be 50 microseconds. In our first prototype, we use a 48-volt source. And in the top waveform here, you can see that we have a 50% pulse width.

That means we're going to turn on our Cree 450 switch for 25 microseconds, and then we're going to turn it off for 25 microseconds. And then we're going to turn it on for 25 microseconds, and turn it off for 25 microseconds. That'll give us a 50% pulse width, and a 24-volt average output.

We're on at 48 volts half the time. And so our average voltage output would be half of our 48-volt source, or 24 volts. Similarly, the waveform below that, we show a 25% pulse width.

And that would give us a 12-volt average output. If we turn the switch on for 75% of our 50-microsecond duration, and turn it off for 25%, we would have a 75% pulse width, and a 36-volt average. Now, 20 kilohertz is a fairly high frequency for this kind of work.

The Cree XM3 series can operate at up to 100 kilohertz. If we fire that output of the 48-volt battery through the switch into a large inductor, inductors resist any change in current levels. And so it would have the effect of smoothing this rather thoroughly to where we would follow the average voltage without really seeing any of the pulses of the higher frequency 20 kilohertz.

And so this is your basis of pulse width modulation, is you can vary your output based on the percentage of on-time during your 50-microsecond duration of your pulse. And there's only two levels here. That's full-on to 48 volts, and full-off to zero volts.

But by changing the ratio of the amount of time on to the amount of time off, we can directly control the average voltage. And into a passive component such as a capacitor or an inductor, that will smooth very nicely to the average. So that's the basic concept of pulse width modulation, is simply switching an output on and off at a variable rate or pulse width of our carrier frequency, in this case 20 kilohertz.

Okay, let's apply this pulse width modulation concept to produce a 240-volt AC sine wave output at 60 hertz. And let's take a look at our little diagram here that I have up here. This is our transformer, and it has sufficient turns ratio to step up our 24-volt peak-to-peak, or 24-volt peak, 48-volt peak-to-peak output to 240 volts AC RMS in the secondary.

And we center tap the secondary, so that's our neutral L1 and L2. And for AC, we actually want the current flow to reverse. And so we use two of the Cree switch modules.

And the modules are half-bridge modules. They contain two silicon carbide switches. And we have two of them, and they're connected to the ends of the primary of the transformer.

So if this is our positive 48 volts here with the orange at the top and our return and negative terminal of the battery at the bottom, then we can see that following electron flow, that our current would go from the negative terminal through this switch, if it's on, up through the transformer in this direction, and down to the midpoint of the other module, and go up through the upper switch on this one to our positive. And so we'll have full conduction when these two switches are on. At the other end, if we reverse that and use the other two switches, we have current from the negative terminal up through the lower switch on the left module, through our primary, and then to the midpoint of the other module, and through the upper switch of the other module to our 48 volt positive terminal.

And so you notice that the current flow through the transformer actually reverses. Let's take a look at our waveform here. On the positive switches, what I'm calling the positive switches, here is our 20 kilohertz pulse width modulated waveform.

And we go from five percent here of this waveform to 50 percent of the midpoint, and then we go back to five percent. And we're going to do that in 333 pulses at 20 kilohertz equals 60 hertz. And so this will cause our sine wave, our average voltage, as we increase this pulse width modulation from five percent to 50 percent, our average voltage is going to increase to a maximum of 24 volts at the peak, or at 50 percent of our 48 volt pack.

And then it's going to decrease back to five percent, and we're going to let it go to zero for our crossover, at which time we switch these switches off. They're off and will remain off for the second half of our waveform. And we're going to use the other two switches, and we're going to do those at 20 kilohertz, and again 333 pulses across that duration of 50 microseconds per pulse would give us 60 hertz.

And we're going to switch these to reverse the current flow through the transformer, and we're going to go from five percent to 50 percent, or minus 24 volt peak, let's call it, and then back to five percent, and then to zero. And this will give us a sine wave on the primary at 60 hertz, varying from 24 volts peak at one direction to 24 volt peaks in the other direction. And that is going to cause that waveform to appear in the primary.

And if we step that up through the transformer, simply by number of windings, we can achieve about a 340 volt peak-to-peak waveform, an identical waveform. It will actually be flipped in phase, but it'll be 360, 340 volt peak-to-peak, and that works out to about 240 volts rms sine wave. And so this is the situation we want our gates on these switches driven at a rate that will vary from zero to 50 percent to zero to generate one half of the waveform, and the other two switches to generate the negative side of the waveform.

And in this way we receive a true reversal of current flow through the transformer to generate a positive and a negative voltage and current. And so that's the objective. So that's an explanation of how we do it.

This simply shows individual pulses from 5 to 50 and back to 5. But that winds up being what 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18. That winds up being 20 pulses. We're actually going to do 333 pulses.

And so we would take our 333 divided by 20, and we're actually going to do about 17 of each of these pulses at each of these, about 17 at 16.65, 17 at 5 percent, 10 percent, 15 percent. And in point of fact, we probably won't do that. Instead of 5, 10, 15, and 20, we might do 2, 4, 6, 8 all the way up to our 50 percent and back down to get a finer granularity.

I have Collin working on that code. I don't know what exactly he's going to do there, what the increment will be. But we're changing the pulse width from zero to 50 percent and back to zero.

And by the way, that's arbitrary. Depending on what we want to induce in the primary, we can go zero to 90 percent or 10 percent to 90 percent. But nominally, zero to 50 percent and back will give us what we need.

And so that's the sine wave we'll generate at 60 hertz. Now by varying this 20 kilohertz and doing the same number of pulses, we can vary the 60 hertz sine wave output. And one of the design goals in this particular inverter is we need very fine control of our output frequency.

And that is because we're going to use 60.1 hertz or 60.2 hertz, 60.3 hertz. By changing the values between 60 and I think about 62 hertz, we can modulate the output of grid-tied inverters such as the N-phase and completely control them. Instead of turning them off for 5 minutes and allowing them to come back on and later turning them off for another 5 minutes, we can actually vary the output of them and have them produce all the time to control the charging of our battery and the power production to our loads.

This was the critical feature. I could not get other inverter manufacturers to perform. And so there's some other things I'd like to do with inverters, but that was the central one, is that UL 1741 Supplement A of 2019 establishes this frequency control to a fine level of the output of the grid-tied inverters.

And I could not persuade any inverter manufacturer to take this at all seriously. It's actually the future. I think grid-tied inverters bringing 240 volts AC from the roof are the future.

And Enphase is the 800 pound gorilla there right now. By the way, their stock which I originally bought at $5.90 is trading at $62 today, a little over a year later. And so I consider it critically important to be able to control the output of those grid-tied inverters based on our battery state of charge and our loads.

We'll have them on fully until our batteries get full and then modulate them to where they basically provide what's necessary to the loads and no more. And so our loads during the day, while our battery is full, will be provided by the grid. Of course, we can supplement that from the battery, or if our battery is below 100% state of charge, we can take excess power from the roof.

And so that's why I want this fine control of our output frequency is to be able to control the output of the N-phase grid-tied inverters. And so this is how we're going to generate our sine wave, true sine wave waveform. I'm pretty confident that using these advanced switches at a 20 kHz rate, we can get a very smooth sine wave output.

And we can very, to a very fine degree, modulate this 60 Hz output to whatever meets our needs. And so that's the basis of pulse-width modulation to achieve a 60 Hz true sine wave at exactly the voltages we want using pulse-width modulation. Let's talk just briefly about voltage.

Okay, Jack, you're using a 48-volt battery. But in truth, the battery voltage can vary from 48 down to 36 or 37 before our battery is fully discharged. How do you maintain the 240 volts AC RMS on the output when your source voltage is not 48 volts? In fact, it could be 40 volts.

Well, that's kind of a big question. And it not only applies to the output of the battery varying, but as we cut in loads and cut out loads, our output voltage will vary based on the amount of load we put on it. And so what we're going to do is put some voltage sensors and current sensors on our output.

And we can monitor those for voltage and current. And recall that we went from 5% to 50% or whatever. We can adjust that voltage, of course, by simply widening our pulse-width or narrowing it slightly.

And so small changes in the pulse-width can result in significant changes in our average voltage. And so for batteries 40 volts instead of 48, we simply widen the pulse-width very slightly in each of these increments to maintain our 24-volt to 24-volt waveform. Similarly, if we detect a decrease in the output based on load, and that's why we do the measurement on the output, we can again vary this to maintain a specific voltage in the face of ever-changing loads and battery voltage.

And so this modulation figure is going to have to be altered. And I'm very pleased to announce that's Colin Kidder's problem. Because of the speed of the microprocessors, and ours is actually a pretty impressive 240 MHz, you would normally look up these values in a lookup table rather than actually calculate them.

I'm not too sure how he's going to do that, because we have the necessity of constantly changing the values based on our output voltage, depending on battery voltage and loads. But the floating-point math required to calculate them each time is significant. So I imagine he will have some sort of thing where he calculates the whole table and then runs it.

And if he needs to do a correction, recalculates the whole table and then runs it. Now that's kind of how I would do it. But he's a pretty bright guy, might come up with something entirely different.

But that's how you vary the voltage as you monitor your end result and adjust your actual PWM, may not be 50%, it may be 51.5% or 49.2%. But we are well able to control that. One of the issues is just simply the time slice of the microprocessor to maintain all this. And we have a lot of other things to do.

We need to measure these with sensors. We need to monitor sensors for current voltage. We're going to do that in the primary and the secondary.

And some other things, communications, both wireless for our display and by CAN. And so this is going to involve a lot of work that has nothing to do with PWM. What I have suggested, and it appears we're probably going to do, is actually use two ESP32 controllers.

One with a shield that directly provides the differential outputs to the gate drivers for these switches. And one that just handles communication duties and sensor measurement. And so that's probably the direction we're going to take.

It's just overwhelming with horsepower. Understanding that we're using a $4 microcontroller chip, but one that I like and we've developed some tribal knowledge of because it's wireless Bluetooth and Wi-Fi capabilities. And so that's probably the direction we'll go is two controllers.

One for the basic PWM and one for just ancillary duties, communications, and sensor measurements, and that sort of thing. And have them communicate between each other. So that kind of wraps up.

We can change frequency and we can change the voltage and hopefully get a very smooth sine wave output. I would mention and probably have elsewhere in this video, I'm kind of enamored with these XM3 series switches from Cree Wolfspeed. They're silicon carbide.

One of the issues is the amount of power they consume internally. MOSFETs have a very low resistance when turned on as compared to IGBTs. And these are extremely low, as low as 2.4 milliohms in the full-on state.

They also have a high tolerance for operating temperature. We can run them up to 175 degrees centigrade. Most inverters, you start to cut back at 80 degrees centigrade.

These can operate up to 175. But that low resistance internally when they're on means you dissipate a much lower level of power as heat within the device. And you need much less of a heat sink to operate them at quite high power levels.

In theory, these switches can handle over 500 kilowatts of power. We're going to be looking for 15 in our first prototype and of course 30 using our high voltage version. So that's a little peek in what we're doing with Model 3 batteries and with our clean sheet paper design on a 30 kilowatt high voltage inverter to mate with the Model 3 battery.

And most importantly, allow us to control in a very modular fashion the grid-tied inverters. You see that UL1741, if we can achieve a fine resolution control of our output frequency, we can get a fine level of resolution of the output power of the microcontrollers. And that's kind of important to me.

Right now when our battery gets full, we have a crude way to shut them off for five minutes. We shift the frequency to 62.5 hertz and they go off entirely for five minutes. The new firmware in the end phases and the SolarEdge microinverters, grid-tied inverters, allows you to actually modulate the power output by varying the frequency in fine steps from 60 to about 62 hertz.

And I couldn't get anybody to give me this function in an inverter. And that's kind of all I want for Christmas. I don't need two front teeth, I need that frequency control.

And I've talked about it to these people for two years. They are certain, I don't know what I'm talking about, don't know what I want, and it'll never be important. Ha ha ha.

Boy, are you guys going to look stupid here in a year or two when everybody's doing this. But it's entirely possible to control that output in a very fine way. And the utility company wants to be able to do that.

That's why it's in there. They want to be able to shut off your solar. I kind of do too.

We intentionally have too much solar here so we can find the edge case. But I want an elegant way when we get up to full battery to match your solar output to the loads and vary that output up and down to approximate the loads using the battery as kind of a buffer. But maintaining it up near full state of charge during the rest of the day and operating your air conditioner and so forth from the solar panels.

Without this, it's effective what we're doing. We cut it off for five minutes and then it comes back on. It just feels inelegant and stupid with a fine resolution of frequency.

I can modulate that output power to a much finer degree and actually cause this to operate in a much more elegant fashion. You're never going to have the right size photovoltaic array. Every day that you operate it, it's either going to be too small or too big.

For too small, we turn it full on. But for too big, which it will be in June if it works at all in December, we need to be able to control that and we need to be able to control it season by season, week by week, day by day, hour by hour, minute by minute. And really, we need to be able to control it second by second.

I know how to do it. I can't get anybody to help me. And so Colin and I are going to do it.

Or blow up a shit pot full of MOSFETs on the way to trying to, which could happen. So that's the mission is a sufficiently large and I want to kind of just say a word about that. You're all streams of running your house on a five kilowatt inverter with 10 kilowatt hours of storage.

I'm sorry. I went through this at age five. We got a little electric motor out of a little toy car and built the go-kart with this little motor and a diesel.

It just wasn't big enough. What a disappointment. In the future, you're going to need a hundred or so kilowatt hours of storage and 30 or 40 kilowatt of power.

We know this here today because we've got two Model 3s charging in the afternoon and we drive on sunshine, but it takes a lot of solar and it takes a lot of battery. And so we're kind of in touch with things on a larger scale and that's where you're going to wind up. Fortunately, as Elon makes millions of Model 3 and Model Y cars, the price and availability of the batteries will go down.

And fortunately, the solar panels themselves will become better and more efficient and less expensive over time. Now, one of the issues with the Lawrence Livermore development or with, you may have heard of the Perscovite cells, is their new technologies and it's very difficult for them to compete with the economies of scale that the old panels get. But that either means you get the 45 or 50 percent panels at a high price or as they come on and start to scale up, you get the 22 percents for 25 bucks a panel.

Either way, the price goes down. More electric cars, more batteries. And so in five years, you won't be paying any more for the 75 kilowatt battery and the 30 kilowatt inverter than you are today for the five kilowatt inverter and the 10 kilowatt hour battery.

Things happen with economies of scale. We're looking forward to Tesla's battery investor day and everyone has jumped the gun and announced what it's going to be, although none of them actually know when it's going to be. He announced it very clearly.

It'll be held in March of 2020 and we're eagerly awaiting that. Well, he did change it to April. Well, then he did change it to May, but they're sure now it'll be in May.

And so within the next couple of weeks, I think you're going to see some surprising things and be made to look very foolish on some of your prognostications into the future. But I do expect some cool stuff to come out of it. One of the issues I don't think people have grasped is that when Elon Musk announced the Gigafactory, what he was saying was, with the building of one building, I'm going to double the world's output of lithium batteries.

Well, several years have gone by and it's no longer half the output of the world's supply. Batteries have become kind of a big business, but it has become apparent to him in a way I don't think it has to you that he doesn't need a gigawatt of batteries. He doesn't need a terawatt of batteries.

He needs several terawatts of batteries, and he needs them kind of right now to do everything he's planning on with the cars, with the power walls, with the rooftops, with the semi-truck, and with the utility grid storage, utility grade, utility scale grid storage. There's not enough batteries in the world again, and he has to figure out a way to make batteries faster, cheaper, and more quickly than you can imagine. This is what's keeping him awake at night, is where do I get a terawatt of batteries by the 4th of July? That's the pressure he's feeling.

He doesn't really need it the 4th of July, but he thinks he does, and so it's going to be kind of interesting to see that coming battery day, and I'm going to leave you with my favorite commercial. At some point, somebody does have a better idea. Stay with us.