Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

Comment Re:Surge protectors *must* be voltage specific (Score 1) 106

limiting that rush has less to do with protecting the PSU and more to do with not tripping crappy breakers if you plug them in coincident with the peak of the mains sine wave.

This has not been true since breakers were introduced. One of the principle advantages to the breakers was a magnetic *and* thermal capability. This allows breakers ( even crappy ones ) to handle inrush of massive machinery.

If you are feeling brave, you can try the following experiment ( I already have with several different types of breakers ): Take a dead short (Large gauge wire like 8 AWG or 6 AWG. Hook it up to a power transistor controlled by a microprocessor. International Rectifier makes some 100A jobbers that can handle spikes of up to thousands of Amps for less than 10ms bursts. Set up a zero cross detect on your microprocessor and turn the juice on through the dead short for 1 full wave of 120VAC. Hook up an o-scope to your mains and check the line voltage as you turn this critter on.

What you will get is one full cycle of severely browned out mains (might drop as far as 10Vp-p. across a 1mOhm load ). For those astute at math, that would be 10,000Amps at 10Vp-p, or about 60 kWatts RMS. As long as you only do this for one full cycle and no more, you will neither blow the breaker, nor significantly heat any of you components. The reason for this is because breakers are mechanical devices and have a specified amount of time before they can "react" to overload. Even their thermal shutdown takes time. Long story short, a standard 15A wall breaker can dispense about 15kAmps for 10ms without tripping. That is far more than the transformer at the pole can handle, and its own internal resistance becomes the current limiting factor. A good line supply will provide you with about 10kAmps before the current limiting drops the line voltage significantly. A typical residential supply will only dump about 1000 Amps before significant drops in voltage. Either way, the breakers will not trip no matter what you do, as long as the entire incident is shorter than 10ms.

This design is very much on purpose. As I have seen, and what these experiments demonstrates is that in-wall wiring, breakers, and utility company equipment are not seriously threatened by short power spikes, so they have no reason to try to protect against them. The breakers in your house are there to prevent enough power from flowing through the wires in the walls to cause them to start fires.

The inrush can however severely damage switching power supplies. Many of them are built to just barely handle the voltages they are designed for. This is done to save money and weight in the extremely competitive mobile power supply market. This market is so competitive, and the margins are so thin, that they will do almost anything sketchy they can to shave pennies off the design. I have personally seen transformerless designs that were built assuming one of the two main lugs was neutral. If they are plugged into a reverse polarity outlet, they short the hot line to the ground directly... This is a common design used on many different power supplies, and is one of the primary reasons you should never open one of those supplies while it is powered. Even in the lab the damn things are dangerous.

Comment Re:We are infected with MBA-think (Score 1) 217

Why would anyone think there's a constant proper relationship between revenue and the number of employees across all businesses?

Because employees cost a pretty predictable amount of money. Your revenues from existing product lines are pretty predictable too. If you have a very disproportionate number of employees to the revenue your business model can expect, then yes you have too many employees. What he's saying is that Yahoo better figure out how to make their existing business model work with 3000 employees because if they don't, the ride comes to a magnificent halt when the employee paychecks start bouncing (aka corporate suicide).

Comment Re:Hard to Believe (Score 1) 217

And I remember Alta Vista now get off my lawn

There has never been anything else that could produce as well as alta vista when all you wanted was technical information. These days, Google has become so watered down with BS that the signal to noise ratio is absolute crap.

The simple truth is that there are two types of searches, and nobody has figured out how to deal with it. There are searches that will end in money changing hands, and there are searches that will result in information moving. There is almost never any overlap. Someone could make a *lot* of people happy by making a search engine that could tell the difference and skip the adds altogether when there is a very low chance of someone buying anything. Save the bandwidth, save the advertisers money, and improve their effectiveness by spewing adds only where there is value to do so, and gain market share by being the best of both worlds.

Comment Re:What could go wrong (Score 3, Insightful) 405

The duty cycle on rooftops is a lot better

True but with rooftop, you pretty much need one inverter per rooftop, which adds substantial amounts to the cost. With these paved roadways, you could probably get the equivalent of 10-20 rooftops with only one inverter, thus significantly reducing the cost of installation. Plus, under most circumstances, installers wont fall off the roadway, thus creating a *very* expensive insurance liability. The cost of liability insurance is a very large (~20%) part of the cost a given rooftop solar installation. Playing around on roofs tends to kill and maim people with rather frightening frequency.

Comment Re:sounds good on paper (Score 5, Informative) 90

In a mesh, more nodes = more paths.

It's also wireless, which means that more nodes = more total traffic one the limited bandwidth available. Doccis has the same limitation, but when you add a router to a congested wifi location, the congestion gets worse, not better. When you add routers to Doccis, the individual congestion clears up.

Mesh networks work reasonably well up to about 1000 - 2000 nodes per square mile. Above that, performance drops as a function of n*log(n). By the time you reach 10,000 nodes per square mile, performance on 802.11n drops to less average bandwidth than old school DSL. at 20,000 nodes per square mile, bandwidth has dropped to dialup speeds. New york city (including the boroughs) has over 25,000 residents per square mile. That means that the Mesh network cant handle more than 10% of the population before performance begins to drop under ideal conditions. Under real world conditions, It is likely to be half that, and will get worse over time as more and more bandwidth pollution is brought into service in the form of the IoT. The part that people don't understand about mesh networks, is that even with huge numbers of gateway routers, once more than a dozen nodes are within reach of each other, their traffic starts interfering with each other significantly due to multiple transmission collisions. Directional antennas help a little, but not nearly enough, especially when there are vastly more sources of wifi that are not part of the mesh that inject pockets of congestion and provide no services to the mesh (other peoples wifi routers who are time warner customers for example).

Mesh networks have their uses, but the operating envelope is a relatively narrow band of opportunity, which will not allow them to "take the city by storm".

Comment Re: Way to go Netflix (Score 4, Interesting) 249

If you aren't using VPN, this doesn't affect you at all.

Absolutely incorrect. What Netflix is talking about, is cross referencing the payment methods bill-to address, and using that to determine what country the customer lives in. The result will be, that when you log in, *your account* determines what content you get access to, not your IP address. Spoofing a bill-to address for payment is a great deal harder to do. Banks do not allow incorrect bill-to address' easily. Most people don't have the wherewithal to get an american billing address, and even if they could, it will cost them more in time and money than it is worth.

Comment Re: How very Republucan... (Score 2) 249

And as a consumer, it gets irritating watching a broken legal system strangle capitalism.

So... This is the result of to much government interference?

in any unregulated economic system, The parties will move towards consolidation, and eventually, monopolies and oligopolies. The end result is no more capitalism. In an unrestricted manner, capitalism will actively destroy itself, as a free market economy is not in the interests of the owners of the biggest companies, and they have the means to do something about it.

Pure capitalism cannot exist any more than pure communism can. At the end of the day, the only thing the masses have to protect themselves from the tyranny of the "capitalists" is the point of a gun, and the rule of law backed up by the point of a gun...

Comment Re:Solar panels made of sand (Score 1) 645

??? The best Silicon only options on that graph barely reach the worst of the Gallium based cells, and even then, they are theoretical designs that have monumental manufacturing costs due to the extreme materials handling requirements. What we need to make a solar breakthrough is silicon only cells that can break the 35% efficiency mark while being as cheap to make as the current commercial silicon only products (which run 25% at best).

The other way to improve them would be to increase the durability of the cells. Silicon only cells loose about 15% of their capacity per decade. After 30 years, you're getting a little more than half the power output you were getting when they are brand new. Compare that to coal and oil powered plants where you still get close to 100% power output at the end of the plant lifespan. The only long term advantage that Solar has is that you don't have to keep buying fuel, but as long as you keep having to replace the cells every 20 years, its not terribly better than every other non-renewable energy source.

As I said, the Gallium based cells will provide a meaningful contribution because they improve on the silicon variety in a number of important ways, but due to materials limitations, and lack of raw materials, solar will forever be a minority contributor to our global power grid. In 200 years, our power supply will most likely be around 10% solar, 5% wind, 5% hydroelectric, 20% geothermal, 10% coal / oil and 50% nuclear. In 400 years, it will likely be 90% nuclear because the energy consumption will have increased dramatically, but the supply of all but nuclear is fundamentally limited such that it will not be able to grow beyond a certain absolute maximum.

Comment Re:Solar panels made of sand (Score 1) 645

Solar panels are made of sand

That is like saying cars are made out of steel. It vastly oversimplifies things.

The first photocells were made from silicon wafers, the same process as used by the semiconductor industry, but modern PV cells use a variety of other materials, to increase efficiency to acceptable levels, such as Gallium, Copper and Indium. These higher efficiency cells are what are referred to by anyone talking about the future of PV cells, because they are the ones that can meet the efficiency targets necessary to make solar cheap enough to be truly competitive. The problem is that these materials are only cheap because the quantities needed for production of todays volumes are easily within the availability of the materials. If you ramp up high efficiency PV cell production, you quickly discover that several of the materials needed will run severely short of supply before even a significant fraction of global power demand is met.

While Solar will make a strong addition to the global power grid, it cannot, by itself, make more than a minority contribution.

Comment Re:That's exactly right (Score 1) 645

His point is that nuclear can't, in and of itself, decarbonize the electric sector. We simply don't have the capacity to build that many nuclear power plants simultaneously, nor do we have the fuel, nor do we have the money.

Like most kinds of systems, there are significant economies of scale to be had when building nuclear power facilities. The first one costs $10B+ to build. The second one only costs $8B because you can reuse much of the equipment and designs. If you are building 100 per year, the cost would probably drop to $5B each. This is easily within the budget of the US military alone, even if we assumed no private investment in nuclear power.

What is missing is neither the capacity, the money, nor the knowledge. What is missing is the political will do actually do something about the problem and stop wasting money dropping $1M bombs on third world countries. Our politicians are guilty of criminal mismanagement of the resources of this planet, and the only excuse they offer is: "It's the will of the people"; which unfortunately it is.

Slashdot Top Deals

"The following is not for the weak of heart or Fundamentalists." -- Dave Barry