Google Offers a Million Bucks For a Better Inverter

There are several reasons for AC being chosen over DC for distribution. Transformers is a major one - for changing the voltage and for isolation. I suspect that AC generators were more of a driving factor than AC motors - since all electricity is generated, but only some of it ends up driving motors.

A key factor, that gets more important as voltages and currents rise, is switching (in particular, disconnecting). It is vastly easier to turn off AC with a contact switch or relay than corresponding voltage DC.

This is all true, and for the great majority of modern motors and their drives, the AC is first converted into DC. DC to DC conversion would work just as well (or could even be omitted, depending on the voltages in question).

HVDC lines are becoming practical over land too. There are many good reasons why HVDC is better for a grid backbone than AC. As you say, they can be buried - they can even be made with just one line, using the earth (or the sea) as the return line. Capacitance in the cables is a good thing rather than a bad thing. It is easier to connect up multiple sources to a DC backbone - you don't have to worry about phase and

voltage. And you don't have any skin effect - in copper, the skin depth is around 1 cm at 50 Hz, meaning you have to use Litz cables, hollow pipes, bus bars, etc. With DC, your conductors can be any shape you like.

There are, however, three big challenges in HVDC lines. First is getting your voltage up to the high voltage, including matching with existing generators or grids, then there is switching the current, then you must convert to something end users can use (lower voltage DC, or existing standard AC). Improvements in high power semiconductors and control systems have made the conversions a lot easier, though they are much more complex and error-prone than relatively simple transformers. And circuit-breakers for large DC lines are now available, though it is not an easy job to avoid arcing.

Where DC would make a big difference, I think, is for computers - and especially for server rooms. It is /insane/ that you have 220V AC coming in to your UPS, which converts it to something like 24V DC for battery storage and/or supply filtering, then converts it back to 220V AC. This is then passed on to each computer's power supply, which converts it back down to low-voltage DC. A single standardised DC supply voltage would make vastly more sense, and cut out two of the three wasteful conversions.

Reply to
David Brown
Loading thread data ...

The switching method used for HVDC is a combination of mechanical and semiconductor switches - it's not easy, as you can't (currently) make a single switch that handles high voltage, high current, and high speed.

---Ms-----Mf---Sf------- |-----Ss----|

Ms is a big, slow mechanical switch that can handle switching with some current, but not near the full HVDC power. Mf is a faster mechanical switch, but it can only switch with very low currents. Sf is a fast semiconductor switch which can handle switching the high current, but only with a low voltage drop over it. And Ss is a slow semiconductor switch which can handle the full voltage, but has a relatively high on resistance.

Start with all the switches on. As Ss has a high on resistance, the current flows through the Mf/Sf path.

To switch off, first turn off Sf - there is only a low voltage drop across it, so this is safe. Current switches to the Ss path, which starts to heat up. Then turn off Mf - this is safe as there is almost no current here (only leakage through Sf). Now turn off Ss. Finally, turn off Ms to stop the leakage through Ss.

Turning on is done in the reverse order.

Reply to
David Brown

Why not put a semiconductor and mechanical isolation in series ?

In normal operation, the semiconductor switch will cut the current and the mechanical isolation switch can be opened when no current is flowing.

Of course, it would be nice if the mechanical switch could interrupt the full current at least a few times, if the semiconductor fails.

Reply to
upsidedown

Seems a waste.

Barring a dramatic and significant increase in the efficiency of photovaltic panels, if you have the physical surface area to deploy 2kW worth of PV panels, you clearly have sufficient space to house a power inverter that is the size of a cooler (even a large "truck-bed" sized cooler).

Reply to
Rich

While certainly true, big (up to 1800 MW generators in NPPs) are long going to be traditional fixed speed 2 or 4 pole generators, However, wind power is essentially variable speed, so in general some semiconductors (at least on the rotor side) is needed to connect to a fixed frequency network. It might make more sense to connect individual wind turbines on a wind farm to a DC bus (or perhaps even using a series connected bus) and have a centralized DC/AC converter to feed the HV AC net.

Small to medium sized gas turbines run at speeds much above 3000/3600 rpm, so you need a mechanical gearbox to generate 50/60 Hz. A micro gas turbine could have a generator running up to 60000 rpm, so an electronic "gearbox" would simplify things to convert 1000 Hz to 50/60 Hz or DC. The fixed speed gas turbines are not good at partial speed, so if the speed could be varied, some energy optimization would be possible.

Having DC/DC converters all over the place, you do not need that much switches, Just stop the converter clock :-). Of course mechanical interconnects are needed for safety, but these can be implemented as mechanical switches not having to break the full load (or short circuit) current.

Unipolar HVDC lines were used with mercury and SCR based line commuting converters, but the return current through the ground or water has some environmental effects in addition to corrosion etc. problems in long metallic conductors such as pipelines. These are essentially point to point links and in order to change the energy transfer, you also had to change the polarity of the isolated HVDC line.

Modern transistor based systems, such as ABB HVDC Light are bipolar systems (essentially big VFDs) which would be more suitable for real networking, not just point to point bulk transfer.

Traditional AC networks rely on phase and frequency of natural (passive) power distribution through various paths. A distributed DC network need much more active coordination and fast data connections, but is now doable and offer some other significant advantages.

Absolutely.

Big iron transformers and rectifiers.

Semiconductor and mechanical switches in series, disabling the feeder DC/DC converter in fault condition.

In Finland, there is atconverts it back down to low-voltage DC. A single standardised DC

Exactly for that reason, there is a standardized 380 Vdc system (+/-190 V).

Reply to
upsidedown

I live in a country with that "horribly dangerous" 230/400 V mains.

I checked the list of electrocution deaths since 1980

formatting link
(try Google translations if you want to know the details).

I calculated 85 death on this 35 years related to 230/400 V low voltage distribution. In fact most of those events were in the

1980's. There is surprisingly many accidents involving 20 kV transmission and 25 kV railroad electrocutions.

Thus 85 deaths/35 years is about 2.5 deaths/year and for a population of 5 million, that is about 0.5 deaths/million people each year.

Can you provide similar figures for any country with the "safe" 100

-120 V world ?

Reply to
upsidedown

Since 110V AC is enough to kill you in serious electrocution situations, and 230V AC is just as survivable in common shock cases, I think it is unlikely that there is any significant difference in safety due to the voltage alone. The US figure I found on a brief google was for 2001 at

0.63 deaths per million, putting it in the same category as Finland. It seems electrocution rates are only significant in countries where you have plenty of electricity, but poor regulations (or poorly followed regulations) combined with general ignorance about electrical safety:

The majority of deaths and injuries due to electricity are indirect - overheating or short-circuits causing fires.

Mains electricity in the range 220-250V AC therefore makes most sense - it is not any more dangerous than lower voltages (unlike 400V AC, which is more dangerous), but is more efficient to transport than 110V AC and is therefore less likely to cause overheating and fires.

Reply to
David Brown

For /modern/ generators, I agree that converting into DC is a good thing

- it frees the generator from the constraint of having to run at a controlled speed and phase. I meant that the type of generator used was part of the historical reason for AC.

That's fine at the DC/DC converter point, but if you want a switch before that you need to combine a mechanical and a semiconductor switch for reliable circuit breaking (or a more complex arrangement for high-power loads). I don't mean that this is a bad idea, just that it is more complex.

I was just discussing the possibilities. Yes, bipolar systems are more common and have many advantages, at the expense of needing two cables.

Until there is a DC system at a voltage that matches UPS batteries, is low enough that you don't need certification to play with it (I think 50 V is the limit?), and comes as a common standard on normal PC's and low-end servers, then there is no standard.

Reply to
David Brown

Not really, all we need are four spacers to mount it offset from the wall. For a copper box, there will be enough conduction and nature air flow betw een the box and wall.

A cube on wall would look funny.

This space requirement is mostly driven by efficiency constraint, but perha ps for indoor mounting requirement. Outdoor mounting will have additional considerations including theft. Roof mounted solar panels are too difficul t to steal, but side wall mounted inverter is too easily taken away.

Reply to
edward.ming.lee

You are making the wrong assumption.

It should be the RMS of 220V on average.

200W / 170V = 1.29A 1.29A x 1.29A x 10R = 16.6W
Reply to
edward.ming.lee

Sorry, i am using the wrong number

2000W / 170V = 11A 11A x 11A x 10R = 121W
Reply to
edward.ming.lee

You are still using the wrong number.

Reply to
John S

Actually, it's 1210W. Good. They will burn out the resistor before the inverter.

Reply to
edward.ming.lee

On Thu, 24 Jul 2014 05:14:17 GMT, Jan Panteltje Gave us:

"electrotrics"

I think that when you drink, your lack of physics knowledge shows up here.

So, where did you take your advanced electrotric engineering courses at.

Reply to
DecadentLinuxUserNumeroUno

On Thu, 24 Jul 2014 05:21:22 GMT, Jan Panteltje Gave us:

We made inverters for use by Hollywood on the very first "steady cam" set-ups. They were pretty dense too. fit in a little experimenter's box. Something like 2 x 4 x 5 inches. I'll bet they were at least 500 Watts.

Everything is way more sophisticated now, and lighter. Then, there were real gyros in them. Now there are cantilevers and "bobs".

Reply to
DecadentLinuxUserNumeroUno

Do a Google search for "380 Vdc", there are a quite large number of hits.

The IEC Low Voltage Directive (LVD) defines anything below 50 Vac or

60 Vdc as Extremely Low Voltage (ELV). I am not sure if +/-60 Vdc is allowed.

Anything below 1000 Vac (phase to phase) or 1500 Vdc are classified as Low Voltage (LV). The bipolar poles must be within +/-900 Vdc from ground potential.

If I understand correctly in the USA, the limit between low voltage and medium voltage is 600 V. I am not sure if this refers to voltages between (anti)phases or just between phase and neutral or how DC systems should be treated.

Reply to
upsidedown

Show me any electrical device mounted on such spacers or that uses convection cooling from the BACK of the device. I don't think such a beast currently exists. Even if it did, a typical 3 year old brat, doing chin-ups on the device, will rip it off the wall. Cantilever packaging is always problematic.

Maybe, but with only a 35 C temperature differential between the case temperature and the ambient air, there's not going to be much air flow, especially through a constrained space, that will undoubtedly be crammed full of wires and cables[1]. Certainly, there won't be enough air flow to effectively cool this thing solely by convection. However, for testing, Google is going to demand that it be floor mounted, not wall mounted, making convective cooling a non-issue. As for conductive cooling, Google clearly specified that they are going to make an effort to not allow much thermal conduction through the spacers. See Pg 9:

A 3.4x3.4 inch cube would be small and unobtrusive. A slab on the wall would occupy more wall space. I can see where it would be beneficial for Google. Most visible space for advertising.

However, you're probably correct. The main advantage of a slab is that the heat conduction path between anything inside that produces heat and the outside cooling surface is much shorter with a slab than with a cube.

Please explain. How does miniaturization improve efficiency? That's one my key points in my rant. Somehow, the miniaturization of the inverter is suppose to revolutionize alternative energy power generation for the world. In my never humble opinion, all it does is create a very difficult design and packaging problem, with zero benefit to the typical home owner.

If Google wanted a specific form factor, they would have specified it.

Yeah, I know how that works. I installed an outdoor security IP camera at an auto repair shop about 2 weeks ago. Last weekend, the last thing it recorded was the thief setting up a ladder and stealing the camera. Argh.

Amazing. You mean the Enphase style micro inverters, mounted under the solar panels are a major target for thieves? It's possible, but I haven't seen much of that because of the difficulty of gaining access to rooftop installation. Sure, if it were ground mounted, there might be a problem, but those are usually enclosed by a fence and guarded by the family Chihuahua. Again reading between the lines and taking a guess(tm), if Google planned to have this device mounted outdoors, there would have been more stringent environmental specifications (i.e. waterproofing). As the specs stand right now, it's an indoor floor mounted device.

[1] Google also forgot about device monitoring. Most such devices have an ethernet or wireless connection to the internet, so that the owner and the vendor can monitor production. Also for alarms.
--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558
Reply to
Jeff Liebermann

Not this time, though that's usually the case with my sloppy math. Don't worry, I'm used to it.

No. There's no 10 ohm series resistor on the OUTPUT, just the INPUT.

I'm talking about the inverter INPUT, which is 450V DC through 10 ohms. If I assume 100% efficiency for the inverter, a 2,000 watt load will require 2,000 watts of INPUT power. At 450v DC, that's 4.44 amps. 4.44A though the specified 10 ohm series resistance will dissipate 198 watts somewhere. We can argue where, but that's still

198 watts of INPUT power totally wasted.
--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558
Reply to
Jeff Liebermann

Den torsdag den 24. juli 2014 17.32.02 UTC+2 skrev Jeff Liebermann:

no one going to add a 10 ohm resistor in there, but if that's the way the intended source behaves that is the way it is

-Lasse

Reply to
Lasse Langwadt Christensen

A preposition is not a word to end a sentence with, you bloody illiterate septic imbecile.

Reply to
Pomegranate Bastard

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.