So this is how our pipe looks like, so it's a dragon power principle, which means we have more and more generation, similar to mankind. And the question that I will be

talking about, or one of the questions I will be talking about is, what actually must happen if we have a short circuit with a cathode? Why the cathode must So this is not one, so first of all I will go through the requirements that I see that the power electronic system has to fulfill, then we will do a review of the

topologies that are published or have been used and proposed and investigated. Then I will show the topology that we are aiming for and the energy The requirements, some hard requirements about the power electronics topology.

So first of all for this system it has to be bi-directional, that means at the launch we have to transmit power from the ground to the kite, through the power electronics to the rogers, which then act as propellers when we are launching and flying the figure 8 on a circle, that's vice versa, and actually it's quite dynamic when you fly circles of the mechanical system and no wind when you fly up,

you have to put power into the system, that's a millisecond range, so it has to be completely bi-directional. Medium voltage due to having a low tetadiameter or tetamers, and we need n-1 redundancy. So no matter what, a certain

number of rogers must be available to still hopefully kite and land. So these are the hard requirements that I see, there are some soft requirements, the first one I would actually argue is also a hard requirement to have a low complexity or have it simple. Light level components, high efficiency,

modernity, useful, or low cost. And I want to point out, it is not possible to have a battery on a large scale kite. Once you scale up to megawatts, your battery on the kite would not allow to lift it up, and of course it would have certain issues on the

efficiency of the kite, if you have an even 100 kilowatt kite on a small scale. So if you have topologies, so this was the first one that was published by Koller in 2011 from BKH to me. He was collaborating actually back then with GOB Energy, and here you see the

system is, the diagrams always look a bit similar, so you have the rovers, you have the generators, and then you have the tender, and you have some ground instinct injectors. So what you see here, we have a DC transmission, and then you have the high voltage DC bus, we have some DC-A converters,

which we have by directional, and to reach a high voltage level, he proposed to use DC converters to convert between a low voltage here and a high voltage here for the transmission, low voltage here to have highly efficient generators. The generators actually want to have a really low voltage

to be very efficient. The alternative would be to have high voltage generators and high voltage DC-A converters, which then also pose challenges by the power electronics themselves. So if you want to have a few kilowatts, it's quite a challenge, especially back then. Today we have better power electronics, like silicon carbide, but still it's a challenge, and also for the efficiency.

So let's check against it. So it's bi-directional, we have a medium voltage, but we don't have an electronic density. For example, short circuit, we can look up a little bit more of that. Then we have a relatively high complexity, lots of light power components, lots of high efficiency. Of course,

you can always argue on this, and there has been a need for it, or you won't, but this one can be, of course, a lot more acceptable. The next one, published by Makani, I'm going to publish, of course the company eventually published, it's a patent, what they have, it's a high frequency bi-directional AC power transmission.

So that AC here, they have a transformer on the ground, on the kite, but which is high frequency, so the transformer itself is relatively small. And the interesting thing about this, you can have low voltage here, and low voltage there, just of the winding ratios.

So we have all the lowering points on the low voltage side, which is quite an interesting scheme, but still, of course, it's not a very good signal, not a very good signal, not a very good signal. So that's the result of this system. Also, actually, it's not low complexity, because you have to design electronics and all the hardware for it.

But let's go to the next one. So this is one, interestingly, that I proposed, and at the same time also Makani published a patent on it, without any knowing from each other.

Here we have the generator, we have low voltage power electronics, and on the different side, we connect the machine's voltage drives in series, and a few of them on the same voltage level, but essentially we have a stack of converters connected in series. And the monitors that we have, low voltage generators, low voltage power electronics,

are very efficient, very simple in itself, for important things. But we pose this to us, which is the voltage stabilization. So we have to control actively the voltage in all levels, which means you have to have the same common misvoltage level

as the top voltage level. If you look at the system on a few points, of course you can have somewhat variable voltages, but this poses even more complexity. So it's bi-directional, it's medium voltage,

but again, short circuit here. You can even have a fault over here, which might lead to a runaway of the voltage, so it's even harder to have fault. And this makes it also highly complex. At least this was my conclusion.

I assume it's anybody from Makani. I have some, but there are some indications that Makani is actually using this, but they don't. So there's another pattern, also from Makani, where they propose to not connect the converters of several machines,

but use one machine. Each machine has a certain set of windings, but then you don't have just three windings, you have a number of end windings, and each windings you make to a converter, and then on the DC side, again, you make the misuses. So this decreases the complexity by having,

again, you have to have a set on this converter and this converter, but the side constraint is that the overall torque of the machine has to match a certain desired torque. The other topology, if you change the power here,

you also change the moment of the cut, and this is not the case here, because this generator is not going to compare. But still, it's in a fault-effect. But it's getting greener and greener, so we're moving somewhat in the right direction. So this is then a topology that I proposed in 2018,

and some research on it, specifically targeting the issue of getting redundancy. And let's start on the pipe side. So we have a rotor, or we have two rotors,

which are on the point symmetric parts of the kit. Those can fail as soon as the other ones are still there. So just isolate them from everyone else, from all the other rotors. And then we have the tether here, and here we have a relatively large number of electric cables, and we'll come back to this later. And the idea is just to not connect all the pluses,

all the minuses of the tether cables together to one high voltage bus, but leave them separated. Then on the ground we have some power current converters, I call them A, B, and C, whereby they could be just electric cables. So you can think of it as a cable. Then we have an un-interrupted power supply,

Then again we have cables, so there are different possibilities what you do here. You can have these converters, for example. And then at the end, so that's two UPSs, you connect the first together, if you might want to do it. And then you have the grid injection. Again, you can also have here electric cables,

and this will be a disagreement, or you can have here electric cables and cables can have a DC-A converter here, so you just connect them, of course there's more, together on the DC side, or on the AC side. The important thing is that you connect them together

left to the UPS. And then you have a certain fault, a certain fuse scheme, so you can protect even if that plus is connected to that minus and you have only one of them moving. And I show this in this paper.

So what we get at the end, we have N minus one, we'll add them in here, but also I think we have very little complexity, but the chipology itself doesn't say anything about the rest. Let's dig a little bit deeper in how the tether looks like, how we can get so many potentials in the tether. So this is how we view the tether,

and how we actually move it. So we have in the center a cord, a mechanical load carrier, and around it we have electric cables relatively high now, especially for the mid-level systems. And each cable also has a, so of course I drew this wire, then I drew this flange on it, then we have a shield that is rolled,

and, next slide, why we have a shield, or maybe for any reason I skipped this, so yet, again, same paper, I show how you model it, and it's not easy to describe it, automatically, and then we see how many electric cables we have in the tether,

and how the mass is, and how the efficiency is of the tether. And what we end up with is, for example, this picture. So this is the result of optimization for a 4-megawatt kite. So we have the core, which is the case of the KEPTa port, and then with the same color scheme as the slide before, we have the electric cables. And here I did not impose any

kind of strain on the number of electric cables. So this is the optimal result, to get the lowest mass, and the smallest diameter. And you already see we have a really high number of electric cables. And the idea is that you don't have a class with this class and this class with this class, it's all together, but even it's not several. For example, over there,

this one is supposed to go down again. This is how the tether looks like for a kite rock kite. This is actually how we build it. And to see it for the small-scale systems, the dimensions just do not look so good. So we have actually just five electric cables here,

and not more than eight or so. But it's better when we scale up. So right now we are implementing the system on the power supply. So proposed technology, let's simply combine the best of those approaches, which is then the following. So we have this

isolation scheme. And for each electric machine, we have a machine which is multi-phase, which doesn't have three phases, but which has, for example, six phases or nine phases. So the number of three-phase sets, each three-phase set is then supplied

by one DCA converter. And then we connect them in series. Then we just have to supply and control the voltage over here. And only if the cycle strength that we need to have a certain torque or certain power or certain speed of each electric machine. And then for each electric motor plant we just have one or two

points-to-metric of the voltage. So then, everything turns green. So again, here's how we implement this. So again, we just have like, five electric cables that are one and the other, and then we have two times plus,

two times minus, and we go and connect the plus and the minus together. So we have the left side and the right side, which are isolated from each other. And we have already applied this. So this was three weeks ago. We just don't operate. We just see it over here. It's working really great.

So conclusions. About half of us in topology have been proposed and investigated on phot electronics. Most important problem that hasn't been solved so far before the population speculation scheme is the N minus one we don't see. And we found this solution

by simply not paralleling the plus cables and the minus cables and by adding a certain fuse scheme to it so that you can have plus and minus connected where you can cut the wires. You can short circuit any wire you want and the worst case that happens is on one. Most

or all other requirements matter both in the multi-phase machines and zero-connection on a DC link. And the topology is entirely implemented and used in high-craft food implementation as planned and for the bigger things. Thank you very much.