Inverters with constant full load capability for electric drives | Hacker News Reader

Inverters with constant full load capability for electric drives

izm.fraunhofer.de

74 pointscl3misch46 comments

If you are wondering about the project name "Dauerpower": "Dauer-"/"andauernd" means permanent/permanently/continous and and - equally important - it rhymes with power ("d-ower").

Here [0] is a longer article by Fraunhofer on silicon carbide power electronics. Depending on how much you want to know, there are Wikipedia articles on a number of terms used in it (SiC, MOSFET, wire-bonding, micro-via, parasitic inductance, IGBT, ...; there is also an explanation of "PCB embedding" on the Fraunhofer website [1]).

[0] https://blog.izm.fraunhofer.de/silicon-carbide-for-power-ele...

[1] https://www.izm.fraunhofer.de/en/abteilungen/system_integrat...

What is the cheapest/simplest configuration for an agricultural PV-to-always-on-e-motor (1MW)?

We can serialize the PVs to get 1000v? And then feed that directly to a (suitable) drive without an inverter? Possibly even a DC motor?

This cuts out half the components compared to an EV drive train, since we have much simpler cooling/packaging/response demand?

What is even the the point of that? An always-on (presumably only on during daytime?) 1 MW motor with enough solar panels to power them and something that actually requires a full megawatt would be pricy enough that the cost of an inverter would not be all that large percentage wise. Especially since you almost certainly would want to have some sort of controller for the motor anyway, which would need the same type of electronics as an inverter would need.

I would assume a huge pump of some kind.

Probably an irrigation pump that needs to move water uphill. Water is heavy.

What sort of agricultural PV-to-always-on-e-motor object would require 1MW continuously? That's a lot of power for agriculture purposes.

Most modern panels seem to be rated for a system voltage of up to 1500V; i.e., you're allows to connect enough in series to get up to but not beyond, as long as your MPPT can cope.

And yeah, 1kV target is practical, you could run a triple half bridge inverter from that into a motor with enough stray inductance to smooth the PWM into pure sine, yeah. It can do the MPPT task at the same time, btw.

The issue is available power is almost never going to be constant enough to directly couple to a load - a cloud passes by and your motor will stall. 1PM on a sunny day, and you’ll be using 1/3 of your power, etc.

If your feeding DC from PV to a motor without an inverter, it will be a DC.

If you hook a brushless motor straight to PV panels, the speed the motor runs at vary throughout the day as the volatage output of the panels waxes and wanes. You'll need to make sure that it the motor has sufficient cooling to not damage itself when running at full power on the hottest sunniest day.

Generally, almost every type of PV or DC electric motor setup has a one or more systems that manage volatage, either in the form of a charge controller the outputs a constant(ish) voltage given the varying input voltages provided by the PV, or an ESC that outputs varying voltages to the motor to change it's rate of speed.

To get good efficiency from a solar panel you must continuously track the “Maximum power point” (MPPT). You would never want to run a motor directly off the solar, you want a power converter to maximize efficiency.

Closest English cognate is probably "enduring".

in both cases, German and English, through Latin borrowing

https://en.wiktionary.org/wiki/dauern

Inverters are becoming a larger fraction of the cost of PV systems, so improvements in the technology are welcome there as well.

Cool project, I like the incorporation of 3d-printed copper heat syncs with integrated coopant channels to the specific components so that they can control coolant distribution more accurately. That has applicability in many other areas.

As battery tech gets better and energy densities increase, these improvements in inverter tech are critical to keep up. This could also mean improved AC output in battery energy storage systems as wel.

What's the real application here? It surely can't be automotive, the amount of time spent at full load is minuscule at least in cars. My old Model S has a full load output of 250 kW but the typical load is less than 30 kW. Lower power cars spend a bit more time at a higher fraction of full load but still typically far less than 50%.

Of course efficiency and cooling are important but EV drives are already quite efficient and rarely operate at full load so the improvement in practice will be small.

Or is the mention of automotive use relevant only because Porsche is involved in testing?

100% duty cycle capable means more durable and capable of industrial use, racing, aerospace, military.

Think mining haul trucks, industrial process control, towing, semi trucks, race cars, electric helicopters, water pumps, etc.

also, likely capable of 150% for short bursts (military power)

> also, likely capable of 150% for short bursts

Existing inverters are already capable of 150% for short bursts, if you define 100% to be the constant full load capacity.

At 1MW?

Why not. 1.5 MW is only six times the maximum power of my ancient 2015 Model S 70D. The inverters are tiny.

‘Only’ And not rated at 100% duty cycle probably?

The 'full load' designation may be a distraction as the research appears to be focused on improving the cooling in general which obviously enables the use of less efficient and cheaper electronics.

Related: An article comparing Si IGBTs and SiC MOSFETs. https://www.arrow.com/en/research-and-events/articles/advant...

Which components in a modern house could run as, or more, efficiently if fed DC power?

I'm guessing this includes:

- Most electronic devices that require AC->DC power adapters. Including CPUs, GPUs, and everything powered by USB.

- Electric stoves, ovens, and other simple electric heaters.

> Electric stoves, ovens, and other simple electric heaters.

Not really. The current would be too high on low voltage DC. And high voltage DC is dangerous.

> And high voltage DC is dangerous.

Is this also at 120V or 220V DC? Is it due to how the alternating current allows muscles to release? (Or was that just a myth?)

It matters for switches and things releasing in a physical sense, so muscles may not come into it. Also, there are issues with high voltage DC contactors welding themselves closed in high demand EV situations because they were sized incorrectly or had poor control.

Thanks, dangerous in the sense of damaging equipment/starting a fire? (As opposed to say shocking someone)

> Also, there are issues with high voltage DC contactors welding themselves closed in high demand EV situations because they were sized incorrectly or had poor control.

Would this have have made a difference if it were AC? I think AC welding is also a thing.

AC definitionally has zero voltage 60 times a second, so when you try to "disconnect" by breaking the switch, the flowing electricity doesn't hold the switch closed. It's why when you look at relays they're rated for 12VDC or 120VAC (that and the commonality of house voltage and automotive voltage). I think the true values are probably a little higher in each, but you'd find that relays cannot break contact at 120VDC where they can at 120VAC.

Thank you! So it's the "stickiness" of DC causing these problems, eh? I wonder if there are applications where the DC could temporarily be converted to AC or turned into some kind of oscillating DC temporarily to use more hardware.

>So it's the "stickiness" of DC causing these problems, eh?

Yes, but it's more than just sticky in the sense of welded-contacts. A DC Arc is a continuous plasma that is conductive. That means the arc continues even with a significant air gap. The arc stretches as contacts are separated and yet the arc continues. That means that fuses can burn out completely but still conduct. Breaker-switches can trip and then catch fire while they continue to conduct rather than safely interrupting the arc. So fuses, breakers and relays all need to be designed specifically for DC or significantly de-rated compared to their AC voltage and amperage ratings.

> applications where the DC could temporarily be converted to AC

Yes and that involves an inverter.

DC arcs don’t self extinguish like AC ones do, because there is no zero-voltage crossing phase point. For a given voltage, it makes DC much harder on relays, and DC relays are more expensive and harder to produce.

This is true even though AC peak voltage is quite a bit higher than the RMS AC voltage. 170V for ‘120V AC’ for instance.

Thanks!

I'd imagine DC-DC conversion to be a bigger pain. Inc comparison is AC-AC is very easy, while you can use switched supplies and what not it can be noisy in an EMF/RF context. And low voltage DC (even something as "high" as 24V) can have massive sag/voltage drop off over 10-20m of wiring, similar to what the other commenter mentioned.

(Technically I'm sure using for eg motors, DC-DC could be done with minimal EMF noise, but you might end up with audible noise and efficiency losses.)

Unless you’re using a transformer, AC adds a complication: energy storage. A device that takes AC in, wants to have a high power factor draw’s power that’s proportional to V^2, so the power is a sine wave at twice the input frequency. Most loads want power that doesn’t have 100% ripple at 120 Hz, so the power supply somehow needs to store about a half-cycle worth of power to out the ripples. As a practical matter, you end up with two-stage power conversion, where the first stage is a “power factor corrected” conversion to a high intermediate voltage and the second stage converts to the final voltage.

Similarly, for AC output, you want that 100% ripple on the output but not on the input.

Three-phase AC avoids this particular problem — power factor 1.0 with >= 3 passes has constant total power. But even a three-phase-AC motor drive producing variable frequency three-phase output has an internal DC bus.

As a practical matter, IMO all large residential loads except resistive heating either should be, or already are, either DC or variable frequency drives.

- Electric stoves, ovens, and other simple electric heaters.

Resistance heaters don't care at all about DC or AC. And with induction you actually have to make the current AC with frequency around 50khz so I don't think it will matter that much in the grand scheme of things if you start with AC or DC.

DC would even be cheaper, you could skip the PFC front-end.

You could surely build an induction heater that has high power factor without a PFC front end by modulating the output at 120Hz. The result might be a loudly buzzing pan, though.

Heaters don't care. Zero effect.

High power ~300W AC/DC conversion is 90% efficient.

Low power ~1W AC/DC conversion is typically 65% efficient, but the energy used is also very small.

Very small but arguably 10x-100x more numerous, so it's not negligible. Especially once you start accounting for the spread of LED lighting.

Not negligible, but small. Lighting is roughly 9 percent of home electricity usage, TV and Media Equipment: 4 percent.

LED lighting almost invariably wants constant current DC, so there’s a conversion stage regardless. The only major sort-of exception is LED tape, which uses a constant voltage supply, but internally, and lossily, regulates current.

A modern high-quality LED light bulb uses a little IC that controls a non-isolated switching converter. You can find excellent datasheets online.

If we assume induction motors get replaced by ECM motors or 3 phase induction motors with VFDs, then everything but aquarium pumps and hair clippers? /me ducks.

Basically, nothing.

The problem is voltage. USB needs 5V, CPUs/GPUs need 0.8V-1.4V (you feed them 12V, but that gets down-converted), plenty of other chips need 3.3V. You can't wire a home for 5V or even 12V DC because the losses would be unacceptably high.

This means a full-home DC grid would need to run more like 100V-200V DC, so you need DC-DC conversion at every point of use. And efficiency-wise AC->DC or DC->DC don't differ much. They're both around 95% in ideal scenarios, or more like 80% in real-world use. It really isn't worth the effort.

The big question would be at what voltage? Most simple electronics want low voltage, but things like stoves and water heaters would want high voltage. To keep cable costs and resistive losses down, you'd want to keep the voltage as high as possible until just before it gets to the device. I doesn't generally make much sense to have one +5v power supply for a whole house, because that power would have to traverse a lot of cables before it gets where it's going.

Historically, AC has been a lot easier to step up and down as needed, but maybe these days buck/boost converters are cheaper and just as good as a transformer.

One advantage of DC is it doesn't have a "skin effect" where the current tends to just flow on the surface of the conductor. So, you can move more power over a solid DC cable than you can with AC at equivalent voltage. That might mean you can save on cabling costs with DC, but I don't know if it actually matters for household wiring.

Just separate all electronics to low-current vs high-current.

Most digital home equipment are low-current.

Electric stoves, ovens, and other simple electric heaters, air conditioners are high-current. EV charger, electric cycle are also considered high-current.

Also in some houses, like Elon Musk house, could be servo-doors, like doors from Star Trek - also considered high-current.

You could use Solar panels with tracker - it also considered high-current.

So, all low current devices best to power from something about 48V, which is safe for people and easy to achieve for electronics, but is high enough to limit currents in power wires. All high-current devices should be powered by 220V or higher, because even on 220V at 1kW power will be significant losses.

> Following a simulation phase, the prototype is currently under construction and will ultimately undergo an extensive testing process at Porsche AG

Reply on news.ycombinator.com