Form Dc 100c: Fill & Download for Free

Inverters use Power Transistors to switch the DC voltage at high frequencies (30khz) to convert them into AC voltage.Now-a-days IGBT (insulated-gate bipolar transistor) is used for switching at high efficiency at high voltage.The most common values for Maximum CE Voltage are 600 V and 1200 V. IGBTs with Maximum CE Voltage as high as 3300 V. Collector Current @ 100C can be between 3 A and 1200 A, with 20 A, 30 A and 40 A being the most common values.

Depends on how much current you are referring to?IF a 5 kW solar array were 500m away from the structure where the electricty will be consumed is, yes it probably is better to boost the voltage up to 1200 volts, feed the power via overhead conductors, and insulated poles, the 500m, then step down the voltage 10:1 to 120VAC.Even more so, if it was 10 kW or larger.Suppose you had a 20 kW array, and it nominal output voltge is 48 volts DC.20,000 divided by 50 is approximately 415 amperes.To have sufficient copper wires and have a voltage loss of only 5% would entail having the sizing of the wire being twice as big as a standard home’s typical 200Amp service.To reduce it further you could have three links of line to deliver that amount of power.The line losses would be 1,000 watts at a 5% loss factor.Delivered to the structure would be 20,000 less the 1000 watts burned up in heating the wires. to 90C or more.The power-inverter inside the structure would then convert the 19 kW up to 120 V. so there might be 1000 watts lost there. (if the power-inverter was 95% efficient). Less if it is more efficient say 98% so 500 lost to the inversion.Suppose you did it the other way, take the 20,000 watts, invert it to 120v ac, at a 2% loss factor, then transmit the AC to two Local 10:1 1200 to 120 with 10kva transformers.The first transformer would lose perhaps 100W.The Power cable at 1200 volts would need to be suspended 10 feet off the ground or better, and be insulated properly. for a 5% voltage drop, you could size the wire to 20 Amperes or 10 A.W.G, bare stranded copper. For durability , you may go to a higher gauge, say, 8 A.W.G.At the structure you would have an adjustable transformer , that would regulate the 1140 volts , and pump out 120V into the structure, and only use up 100W to deliver the full power of 20,000 watts less the first transformer of 100W, perhaps 90W for the second transformer, and 100W burnt up in heating the 8 A.W.G. wire to 90 or 100C.Delivered to the structure is 120V AC at 19,700 Watts approximately.Losses of 300W vs losses of 1,000W at 48 volts.And the cost of copper at 000 gauge (tripple-0) vs #8 A.W.G.

A lot of good answers already here, and I learned a few things reading them. Some items that no one pointed out, though:Most modern computers can deal with a lot of heat in a high temperature setting. It is safe to run a computer for indefinite periods of time at above 80c, which is surprisingly close to the boiling point of water (100c). The ambient temperature of the server room could safely (from a computer’s point of view) be over 50c.Server rooms are often targeted to just over 10c (that’s a big diff from 80c!!!), and the cost of cooling to that extreme is dramatically higher because the efficiency of cooling systems drops as the temperature differential between the outside air and the target temperature increases.Given that there is a 70c differential between what a server can run at safely, and what the server rooms are cooled to, it begs the question of “why are server rooms cooled to that level?” The answer is simple: Spinning (magnetic) hard drives. The failure rate (e.g. MTBF) of a hard drive is closely correlated to operating temperature, so to reduce server failures (caused by hard drive failures), the data centers were designed to run much colder than necessary.(Power supplies also fail faster in very hot environments, but they are solid state, and usually redundant in servers. Furthermore, server power supplies are now being designed for, tested, and rated for higher temperatures.)Take this with a grain of salt since I’ve never managed servers, but most of the hardware failures that I’ve seen in data centers were hard drives (maybe 80% of failures?) and power supplies — and most of those were probably caused by the capacitor plague from China: Capacitor plague - Wikipedia.Other hardware also has a higher failure rate at higher temperatures, but the failure rates for most solid state hardware are still very close to zero — well up into the 80c+ range. What this means is that a data center that can be run much warmer if the servers do not use spinning hard drives.Some data centers now segregate storage (servers still using spinning hard drives, e.g. EMC) from compute, and only chill the storage sections, using reduced cooling for the compute hardware (servers that typically have no local storage or only nominal amounts of flash, and absolutely no spinning hard drives).There’s significant potential heat savings by switching to DC power, i.e. having the power that comes into the server room already be DC. This eliminates the power supplies from each server (a source of power inefficiency, heat, and potential failure), and allows much more efficient battery backups (i.e. eliminating the conversion of battery power from DC back to AC, which then gets immediately converted back to DC by the server power supplies). By using many fewer / much larger AC-to-DC transformers, the efficiency of power conversion can go up somewhere in the 10–20% range (well into the 95%+ efficiency range, compared to around 80% for typical server power supplies).Conventional server racks and server cases are very poorly designed for heat removal. Horizontal loading, unnecessary enclosure, and tiny high-speed fans (which, as a moving part, also tend to eventually fail!) are all generally bad ideas when “moving heat away from” servers is the goal.Data center investments are large and long, and it takes time to cycle in new technology. There are many brilliant ideas out for reducing energy usage (and Intel has done much better since trying to destroy the earth with its unbelievably hot and inefficient P4-based Xeons), so in another 10–15 years we should see the typical power efficiency go way up (the “PUE” numbers).