Some CD players have a transistor-based output stage (which range from very cheap (e.g. in a discman) to extremely good), others use valves
It's generally accepted that transistors more faithfully amplify a signal which does not exceed the transistor's dynamic range (defined by its input voltages and biasing network). When that range is exceeded, most transistor amps will distort the signal with odd-order harmonics, whereas most vacuum tube amps distort with even-order harmonics, which are generally accepted to be more pleasing to the ear. If a CD player's internal amplifiers clip at all, it's been designed wrong. It's also generally accepted that playback is not a good place to add distortion - the sound produced at your ear should be the same as the sound produced at the ears of those who mixed and produced the CD. Let the guitarist put the warmth of his tube amp on the bits on the CD. Also, the most common type of cheap DAC is a 1-bit DAC, which works by turning the switch to charge a capacitor on and off for exact periods of time. There is no reason that a 1-bit DAC is going to introduce more distortion than a non-oversampled 16-bit DAC, assuming any reasonable level of engineering competence - it's much easier to ensure linearity with the former. In fact, the SACD format is practically built around writing the information on the disc so it can be directly fed into a 1-bit DAC.
Now, regardless of what is and isn't generally accepted, some people have tastes for certain things being distorted certain ways. If you are one of them, then please accept my apologies.
You compute the strengths and phases for each of the N antennas to create a pattern that has a null at the angles toward stations 2 through N but not on station one. You modulate that with the information you want to send to station 1.
You also compute another set of strengths and phases to put nulls on stations 1 and 3 through N but not on station 2 and modulate that with the information you want to send to station 2.
In the searching I've done, neither "steerable null" nor "MIMO" seems to imply these technical abilities, so I'm having a hard time understanding this process. Let me describe the simplest example I can think of so you can point out to me where I'm not getting it. Say you want to transmit the same simple carrier wave to two points at the same time. You are using two dipoles oriented vertically to transmit, one is directly north of the other, one wavelength away (just using one wavelength to make the math simple - if a different number is needed, let me know). One of the receivers is at 45 degrees (northeast) and the other is at 135 degrees (southeast). For the northeast station to receive a coherent signal, the signal needs to leave the southern antenna.707 wavelengths (or sqrt(2)/2*360 = 255 degrees) ahead of the northern antenna. For the southeast station to receive a coherent signal, the opposite is true. So if you just add the signals together, you're sending out two sine waves simultaneously out of both antennas. Two sine waves of the same frequency added together will always add to one single sine wave (which is how you get 208V from two 120V power phases), and if the two components are 105 degrees apart (360-255=105), that single sine wave will have 159% of the strength of one of the components (whereas if the signal power was not split between the antennas this would be 200%). Since it leaves both antennas at the same time, and they are one wavelength apart, the radiation pattern will have a 3db gain (not relative to the 159%) at the north, south, east, and west points with nulls in between each point. What am I missing?
- Multiply by the number of cells.
- Multiply that by the number of directaional-antenna sectors in each cell.
- And multiply yet again by the number of antennas in the steerable-null array in each sector.
While it is important to consider how to prevent one customer's traffic from affecting another, I was talking about the amount of bandwidth a single customer could potentially receive. Also, I believe the cellular companies are transmitting on non-overlapping channels with each sector antenna (at UHF and microwave frequencies, they are not directional enough to prevent overloading the other antennas/radios a few feet away), and they separate uplink/downlink frequencies quite a bit to help with this too.
it would seem that 700 Mhz spectrum would only give you 15 MB/s of available bandwidth
You're trying to compare two separate units. 15MB/sec is not an amount of bandwidth, it is a bitrate. Bitrates much, much higher than the bandwidth can easily be achieved if you have a high enough signal-to-noise ratio. For example, a "56k" modem can achieve 53000bps in 3000hz of bandwidth. Similarly, low bitrates can still be achieved even with signal-to-noise ratios much less than one (GPS does 50bps with signals less than one thousandth the strength of the noise floor).
To determine error-free bitrate, you need to know how much bandwidth you have, how much signal you have, how much noise you have, and also what the spectral efficiency of the modulation technique you are using is. The formula is called Shannon's Theorem.
In other words, once the FCC announces what the maximum allowable power is for this band, then you can start speculating on how much data you can pump through it.
Folks, picture this. Your next door neighbor dies. The next day, co-workers start dieing. Are you going to go back to work?
Because of the time delay. You'll be symptom-free for a day or two after the initial infection and your doomed coworkers won't die for several more days. Are you really likely to figure out that this is the next pandemic flu before your coworkers expose you to it?
Compared to just replacing the hard drive for $150. Hardware is cheap. Labor is not.
Your example makes sense, but what if you've already done that? Say your app is SQL-based and does some queries that are O(n^2) complex. You've already spent $20k on a bad-ass server with RAID10, a bunch of spindles, separate transaction log drives, and as much RAM as can fit. Now, a year later, there's more records in the system and performance sucks again. Where do you go from there? These disks don't go to 11. If you want to double the performance of that $20k box, you're likely going to spend not $40k but $200k.
Once you outgrow commodity parts, if you want a 2x speedup, you'll usually have to pay 10x for it. Or wait three years. The price/performance curve is deceptively shallow towards the bottom end.
Patience. The Asus EEE is due in a few weeks. It beats the pants out of this one.
Except for two things (for me anyway): a display readable in direct sunlight, and extended battery life (the presenter at LinuxFest Northwest earlier this year claimed he left an XO running for 24 hours once while it was displaying the camera's output on the screen).
For example, assume my cellphone has the best electronics available, how can I increase the signal to noise ratio? Standing at the focal point of a dish aimed at the nearest tower?
One way to do this is to use an array of smaller antennas, and change through software how signals are timed through each. Say I have two dipoles 30cm (1 nanosecond) apart. If I transmit something at exactly the same time from both, the receiving antenna broadside to the two transmitting antennas will see a 3db boost. If I delay the signal to the second antenna by 1ns, the receiving antenna in line with the transmitting antennas will see a 3db boost. If I delay some fraction of 1ns, the 3db boost will be observed at an angle. The same effect can occur on received signals, and more than two antennas can be used (which only becomes practical for portable devices at the higher microwave frequencies anyway). Wikipedia explains it better than I can.
There are 2 ways to increase the amount of data that can be sent.
There are actually four:
Increase the signal strength (using a directional antenna or amplifier)
Decrease noise (use higher-quality components, shut off interfering transmitters, use directional antennas)
Increase the signal bandwidth
Increase signal spectral efficiency (for example use OFDM instead of FSK)
Changing the carrier frequency has no effect, except that there's more room for higher-bandwidth signals at higher frequencies. 2.400-2.422GHz seems like a smaller chunk than 400-422MHz, but it can carry the same data.
The formula for how many bits you can send and receive error-free is the Shannon-Hartley theorem, and spectral efficiency is typically stated as a percentage of the theoretical.
Like you said, it's already put inside of every lithium battery made, that's not the problem
I admit I'm being a bit of a pedant here, but it sounds like what you want aren't lithium batteries, but lithium cells. I would guesstimate that the mass-produced cost of the safety circuitry is somewhere between $3 and $10, which would double or quadruple the cost of a cell. (a MAX1737 which only implements charge control, not discharge control, is $2.85 for lots of 1000, and requires supporting components. The LM3621 is less expensive at $1.40 but has the same limitations), and some room in each cell (reducing capacity and adding weight). You may find the R/C aircraft hobbyists' attempts to use lithium in their homemade battery packs interesting reading.
If my guesstimation is wrong, there's probably a profitable business waiting for you.
I'd rather see *standard* lithium-ion rechargeable battery sizes, so that manufacturers could just quit designing things for alkalines.
The holdup with that is that lithium-ion and lithium-polymer batteries require some specialized charging/discharging circuitry that would need to be placed inside the battery pack itself for safety (otherwise you get this if you overcharge them, drain them too low, or short them). The safety circuitry is expensive (especially if it needs to be made general purpose), so you only see the work for this done in products which require lithium's energy density.
You'd probably lose some due to weather, of course.
If you can make a UAV that climbs to ~100000 feet in the day and glides down slowly enough that it's still above ~50000 feet at dawn, then you'll avoid all weather (including clouds) except on takeoff and landing. I could see something like that staying up for months or years at a time.
Over her in the UK there was an attempt by motor manufacturers to claim that new car warranties were only valid if the cars were serviced by authorised (read overpriced) dealers.
They tried that in the US too. The result was the Magnuson-Moss Warranty Act. 15 USC 2302 (c) puts the onus on the manufacturer to prove that a problem was caused by a third party.
kflex tried to rush to market.. but i personaly think it was USR that made v.90 stick
That's rewriting history. USR promoted X2, Lucent/Rockwell promoted K56Flex. There was no interoperability. A year or so later, with poor sales and no clear market leader, they both compromised with the v.90 standard. USR equipment sold after that point typically supported X2 and v.90, Lucent/Rockwell equipment sold after that point typically supported K56Flex and v.90.
Sort of reminiscent of DVD+RW vs. DVD-RW, Bluray vs. HD-DVD, etc, etc. It seems that if you want everybody's product to follow a documented open standard, you should have the first implementation of it be done by an academic institution.
there is no reason to not release them to those that want them earlier, as well as a monthly package.
That depends on if the vulnerability was already public knowledge. Once a patch is released, it usually only takes someone a day or two to find out what was patched, how the unpatched version can be exploited, and how to adapt some existing worm to automatically exploit it.
15. A complex system that works is invariably found to have evolved from a simple system that works.
16. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over, beginning with a working simple system.
IMHO, John Gall's observations on political systems are incredibly apropos to technical systems.
When they are powered by 120 volts AC, they use a voltage doubler to generate the 340 volts DC.
If you hook 120V AC to a four-diode rectifier bridge, and hook the rectifier bridge to a suitable capacitor, and hook a voltmeter to that capacitor, you will measure 120*2*sqrt(2)=339.4V DC. If you power it from 240V AC, you get 678.8V DC, and while SCRs are commonly available for such voltages, designers will use an SCR on each leg to keep the negative output at ground potential.
There are lots of ways to design power supplies, and designers are free to use their own approaches, but that's probably the most common one. Most modern switching power supplies (ones without a 120/240 switch on the back) are rated to use any input voltage from 100 to 240V, without any hole in the middle of the range that would indicate a doubler being used.
Using a voltage higher then the AC peak made for an easier initial AC to DC conversion when doing power factor correction.
I would think the opposite would be true - instead of merely adding some inductors to the input to keep current flowing, you'd have to use a boost regulator or a transformer to bump up the voltage. I doubt their initial AC to DC conversion uses 120V AC - commercial power is typically delivered to office buildings as 480Y/277V 3-phase, and they probably take it straight from there.
Why would it matter whether you had 168 cells in series or if they were series-parallel for the same number of volt-amps?
Assuming the battery's internal resistance is the dominating factor, the maximum power a battery can deliver when shorted is proportional to the battery's total voltage.
A 168-cell battery can deliver power 7 times as fast as a 24-cell battery. Instead of a battery merely heating up quickly, it may catch fire, or instead of merely catching fire, it may explode violently.
Short circuits are never supposed to happen, but the fact that it could happen if I screwed up while working on or around it would definately make me nervous.
It would seem to me that the cost of HARDWARE would be lower if you weren't paying for a power supply for every dang computer in your data center.
The necessary copper gets expensive fast. A penny worth of it now costs 1.6 cents in raw materials (which, not coincidentally, is why pennies are now mostly zinc inside).
Conductor size is not related to whether the power is AC or DC or what frequency of AC it might be; it is related to current.
If you are specifying wires to pass a certain amount of current without significant heating, this is true. However, if you are specifying wires to pass a certain amount of current without significant voltage drop (as the NEC ampacity ratings do) then there is a slight difference, and that is root-mean-square vs. peak current (for sine waves, it's a ratio of 1 to sqrt(2)).
380VDC corresponds to the peak-to-peak voltage of 134.4V RMS AC, which is probably the maximum they felt any random unmodified switching power supply for 120V AC could take. Incredibly, the article doesn't say whether the implementation plan calls for unmodified power supplies - there's only a teaser of a hint:
At one point in the process, the power is switched to 380v DC power, so keeping it at the stage throughout the entire data center isn't a stretch.
Using DC also introduces the potential for galvanic corrosion wherever you have current passing through two different types of metal (for example a copper wire bolted to a lead battery terminal), and maintaining a 380V stack of lead-acid batteries would make me nervous (With a typical 2.26V lead-acid cell float voltage, that's 168 cells in series!)
Slashdot Mirror
It's generally accepted that transistors more faithfully amplify a signal which does not exceed the transistor's dynamic range (defined by its input voltages and biasing network). When that range is exceeded, most transistor amps will distort the signal with odd-order harmonics, whereas most vacuum tube amps distort with even-order harmonics, which are generally accepted to be more pleasing to the ear. If a CD player's internal amplifiers clip at all, it's been designed wrong. It's also generally accepted that playback is not a good place to add distortion - the sound produced at your ear should be the same as the sound produced at the ears of those who mixed and produced the CD. Let the guitarist put the warmth of his tube amp on the bits on the CD. Also, the most common type of cheap DAC is a 1-bit DAC, which works by turning the switch to charge a capacitor on and off for exact periods of time. There is no reason that a 1-bit DAC is going to introduce more distortion than a non-oversampled 16-bit DAC, assuming any reasonable level of engineering competence - it's much easier to ensure linearity with the former. In fact, the SACD format is practically built around writing the information on the disc so it can be directly fed into a 1-bit DAC.
Now, regardless of what is and isn't generally accepted, some people have tastes for certain things being distorted certain ways. If you are one of them, then please accept my apologies.
2400-2700ms minimum latency? Check!
In the searching I've done, neither "steerable null" nor "MIMO" seems to imply these technical abilities, so I'm having a hard time understanding this process. Let me describe the simplest example I can think of so you can point out to me where I'm not getting it. Say you want to transmit the same simple carrier wave to two points at the same time. You are using two dipoles oriented vertically to transmit, one is directly north of the other, one wavelength away (just using one wavelength to make the math simple - if a different number is needed, let me know). One of the receivers is at 45 degrees (northeast) and the other is at 135 degrees (southeast). For the northeast station to receive a coherent signal, the signal needs to leave the southern antenna .707 wavelengths (or sqrt(2)/2*360 = 255 degrees) ahead of the northern antenna. For the southeast station to receive a coherent signal, the opposite is true. So if you just add the signals together, you're sending out two sine waves simultaneously out of both antennas. Two sine waves of the same frequency added together will always add to one single sine wave (which is how you get 208V from two 120V power phases), and if the two components are 105 degrees apart (360-255=105), that single sine wave will have 159% of the strength of one of the components (whereas if the signal power was not split between the antennas this would be 200%). Since it leaves both antennas at the same time, and they are one wavelength apart, the radiation pattern will have a 3db gain (not relative to the 159%) at the north, south, east, and west points with nulls in between each point. What am I missing?
While it is important to consider how to prevent one customer's traffic from affecting another, I was talking about the amount of bandwidth a single customer could potentially receive. Also, I believe the cellular companies are transmitting on non-overlapping channels with each sector antenna (at UHF and microwave frequencies, they are not directional enough to prevent overloading the other antennas/radios a few feet away), and they separate uplink/downlink frequencies quite a bit to help with this too.
You're trying to compare two separate units. 15MB/sec is not an amount of bandwidth, it is a bitrate. Bitrates much, much higher than the bandwidth can easily be achieved if you have a high enough signal-to-noise ratio. For example, a "56k" modem can achieve 53000bps in 3000hz of bandwidth. Similarly, low bitrates can still be achieved even with signal-to-noise ratios much less than one (GPS does 50bps with signals less than one thousandth the strength of the noise floor).
To determine error-free bitrate, you need to know how much bandwidth you have, how much signal you have, how much noise you have, and also what the spectral efficiency of the modulation technique you are using is. The formula is called Shannon's Theorem.
In other words, once the FCC announces what the maximum allowable power is for this band, then you can start speculating on how much data you can pump through it.
Other companies don't even make you write a note.
Because of the time delay. You'll be symptom-free for a day or two after the initial infection and your doomed coworkers won't die for several more days. Are you really likely to figure out that this is the next pandemic flu before your coworkers expose you to it?
Your example makes sense, but what if you've already done that? Say your app is SQL-based and does some queries that are O(n^2) complex. You've already spent $20k on a bad-ass server with RAID10, a bunch of spindles, separate transaction log drives, and as much RAM as can fit. Now, a year later, there's more records in the system and performance sucks again. Where do you go from there? These disks don't go to 11. If you want to double the performance of that $20k box, you're likely going to spend not $40k but $200k.
Once you outgrow commodity parts, if you want a 2x speedup, you'll usually have to pay 10x for it. Or wait three years. The price/performance curve is deceptively shallow towards the bottom end.
Except for two things (for me anyway): a display readable in direct sunlight, and extended battery life (the presenter at LinuxFest Northwest earlier this year claimed he left an XO running for 24 hours once while it was displaying the camera's output on the screen).
One way to do this is to use an array of smaller antennas, and change through software how signals are timed through each. Say I have two dipoles 30cm (1 nanosecond) apart. If I transmit something at exactly the same time from both, the receiving antenna broadside to the two transmitting antennas will see a 3db boost. If I delay the signal to the second antenna by 1ns, the receiving antenna in line with the transmitting antennas will see a 3db boost. If I delay some fraction of 1ns, the 3db boost will be observed at an angle. The same effect can occur on received signals, and more than two antennas can be used (which only becomes practical for portable devices at the higher microwave frequencies anyway). Wikipedia explains it better than I can.
There are actually four:
Changing the carrier frequency has no effect, except that there's more room for higher-bandwidth signals at higher frequencies. 2.400-2.422GHz seems like a smaller chunk than 400-422MHz, but it can carry the same data.
The formula for how many bits you can send and receive error-free is the Shannon-Hartley theorem, and spectral efficiency is typically stated as a percentage of the theoretical.
I admit I'm being a bit of a pedant here, but it sounds like what you want aren't lithium batteries, but lithium cells. I would guesstimate that the mass-produced cost of the safety circuitry is somewhere between $3 and $10, which would double or quadruple the cost of a cell. (a MAX1737 which only implements charge control, not discharge control, is $2.85 for lots of 1000, and requires supporting components. The LM3621 is less expensive at $1.40 but has the same limitations), and some room in each cell (reducing capacity and adding weight). You may find the R/C aircraft hobbyists' attempts to use lithium in their homemade battery packs interesting reading.
If my guesstimation is wrong, there's probably a profitable business waiting for you.
Is your time travel machine for sale?
The holdup with that is that lithium-ion and lithium-polymer batteries require some specialized charging/discharging circuitry that would need to be placed inside the battery pack itself for safety (otherwise you get this if you overcharge them, drain them too low, or short them). The safety circuitry is expensive (especially if it needs to be made general purpose), so you only see the work for this done in products which require lithium's energy density.
If you can make a UAV that climbs to ~100000 feet in the day and glides down slowly enough that it's still above ~50000 feet at dawn, then you'll avoid all weather (including clouds) except on takeoff and landing. I could see something like that staying up for months or years at a time.
They tried that in the US too. The result was the Magnuson-Moss Warranty Act. 15 USC 2302 (c) puts the onus on the manufacturer to prove that a problem was caused by a third party.
Dear Microsoft,
How will my baby mulching machine be able to legally interoperate with your software?
This is very important to me and my colleagues, and I would appreciate it if you would address our concerns.
That's rewriting history. USR promoted X2, Lucent/Rockwell promoted K56Flex. There was no interoperability. A year or so later, with poor sales and no clear market leader, they both compromised with the v.90 standard. USR equipment sold after that point typically supported X2 and v.90, Lucent/Rockwell equipment sold after that point typically supported K56Flex and v.90.
Sort of reminiscent of DVD+RW vs. DVD-RW, Bluray vs. HD-DVD, etc, etc. It seems that if you want everybody's product to follow a documented open standard, you should have the first implementation of it be done by an academic institution.
That depends on if the vulnerability was already public knowledge. Once a patch is released, it usually only takes someone a day or two to find out what was patched, how the unpatched version can be exploited, and how to adapt some existing worm to automatically exploit it.
I hereby vow to refer to personal software firewalls as "cowbell".
15. A complex system that works is invariably found to have evolved from a simple system that works.
16. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over, beginning with a working simple system.
IMHO, John Gall's observations on political systems are incredibly apropos to technical systems.
If you hook 120V AC to a four-diode rectifier bridge, and hook the rectifier bridge to a suitable capacitor, and hook a voltmeter to that capacitor, you will measure 120*2*sqrt(2)=339.4V DC. If you power it from 240V AC, you get 678.8V DC, and while SCRs are commonly available for such voltages, designers will use an SCR on each leg to keep the negative output at ground potential.
There are lots of ways to design power supplies, and designers are free to use their own approaches, but that's probably the most common one. Most modern switching power supplies (ones without a 120/240 switch on the back) are rated to use any input voltage from 100 to 240V, without any hole in the middle of the range that would indicate a doubler being used.
I would think the opposite would be true - instead of merely adding some inductors to the input to keep current flowing, you'd have to use a boost regulator or a transformer to bump up the voltage. I doubt their initial AC to DC conversion uses 120V AC - commercial power is typically delivered to office buildings as 480Y/277V 3-phase, and they probably take it straight from there.
Assuming the battery's internal resistance is the dominating factor, the maximum power a battery can deliver when shorted is proportional to the battery's total voltage.
A 168-cell battery can deliver power 7 times as fast as a 24-cell battery. Instead of a battery merely heating up quickly, it may catch fire, or instead of merely catching fire, it may explode violently.
Short circuits are never supposed to happen, but the fact that it could happen if I screwed up while working on or around it would definately make me nervous.
The necessary copper gets expensive fast. A penny worth of it now costs 1.6 cents in raw materials (which, not coincidentally, is why pennies are now mostly zinc inside).
If you are specifying wires to pass a certain amount of current without significant heating, this is true. However, if you are specifying wires to pass a certain amount of current without significant voltage drop (as the NEC ampacity ratings do) then there is a slight difference, and that is root-mean-square vs. peak current (for sine waves, it's a ratio of 1 to sqrt(2)).
380VDC corresponds to the peak-to-peak voltage of 134.4V RMS AC, which is probably the maximum they felt any random unmodified switching power supply for 120V AC could take. Incredibly, the article doesn't say whether the implementation plan calls for unmodified power supplies - there's only a teaser of a hint:
Using DC also introduces the potential for galvanic corrosion wherever you have current passing through two different types of metal (for example a copper wire bolted to a lead battery terminal), and maintaining a 380V stack of lead-acid batteries would make me nervous (With a typical 2.26V lead-acid cell float voltage, that's 168 cells in series!)