On the other hand, designs with less energy loss will open up the potential of higher speeds, once the techniques get refined.
No and here is why.
For a CPU of a given complexity, a specific area is needed for transistors, routing, etc in a given process. If the process density goes up, then the power has to be lowered to maintain the same power/area because the area largely determines the thermal resistance and for the past few generations, high performance CPUs already operate with the junction temperature as high as is reliable. So power is proportional to chip area and higher density processes yield smaller chips so power has to be lower.
You can see this trend in Intel processors since about the Core2. The highest power models all have a power rating proportional to area and since more recent models are smaller, they have lower power ratings.
This is also why stacking memory on top of logic is not going to happen for anything except low performance logic.
The use of non-volatile SRAM which has an embedded lithium cell might be considered this when it is not replaceable. A lot of test equipment built starting in about 1990 has this "feature" even if it was implemented by replacing the previous solution of an external lithium cell, low power SRAM, and backup circuits while using the same printed circuit board.
More recent is the problem with aluminum electrolytic capacitors which have a well understood and predictable wear out mechanism. I have older ATX power supplies which have lasted well past their warranty date but newer ones from specific manufacturers reliably fail just out of warranty.
My understanding was that topping up a lithium battery was good; it was a more complete discharge that added wear cycles to it. Obviously this isn't true forever but better than lots of charge and deep discharge cycles.
The irreversible damage occurs at both high and low voltage. If the battery capacity is limited to between say the 80% and 20% charge points, then the total energy delivered for a given decrease in capacity goes up so many shallow discharges are better than fewer deep discharges if they occur toward the middle of the capacity curve. Maintaining the battery charge state closer to 100% lowers the useful lifetime (lithium metal plates out?) but this effect is usually small compared to capacity lost do to general use. It would be more important in standby applications.
Supercapacitors also degrade at high voltage; in their case, increasing the voltage by 0.2 volts lowers the operating life by 10 times. Offhand I do not know what the equivalent specification is for lithium secondary cells.
For Hubble someone forgot to account for change in shape of the lens due to gravity, because, you know, it's a complicated thing to make.
The problem with the Hubble mirror had nothing to do with gravity.
The custom precision null corrector used during the figuring step was assembled incorrectly. When the standard null correctors used for the initial grinding showed spherical aberration during final testing, they were ignored as being too inaccurate compared to the custom precision null corrector.
A continuity tester would not have worked well because the response time for the human to test all N^2 connections and look at the continuity tester was too long.
I am not sure what continuity tester you have experience with but all of my digital multimeters going back 20+ years produce a tone in addition to a reading on the display and respond within milliseconds. Running the test lead along a string of contacts or pins and listening works fine with them.
I think you might be referring to the CALEA requirements which also apply to WISPs. There is no CALEA requirement to maintain the records you suggest however most ISPs do for other reasons and if you provide voice over IP services, the FCC does require location data to support E991. With or without these records, CALEA requires ISPs to aid law enforcement in identifying and wiretapping targets.
That is the situation some WISPs have faced in contested areas. The incumbent wired ISPs buy up tower space preventing the competing WIPS from using it and then flood the ISM bands with generated traffic. Canopy gear was popular for this because it uses TDMA instead of CSMA. More direct measures like jamming critical nodes with a flood of traffic on the same channel from a directional antenna at long range are also used.
This is why you either start by enforcing the license through legal means or immediately after the company first dissembles. If legal means to enforce the license are unavailable for whatever reason, then just publicize the violation and move on.
A firearm does not have to inflict a death to successfully protect the user and a majority of self defense firearm incidents result in either no death or the firearm not even being fired.
As for the difference between acquaintances or family members and strangers, most interactions are with the former rather than the later and the term acquaintance is used very very broadly in crime measurement.
I made my post just to point out that thermocouples and thermisters are unsuitable for precision temperature measurement compared to RTDs and that has been the case for decades. But even with their complexities like cold junction compensation and lower accuracy, thermocouples are preferred to RTDs because of their temperature range and price. RTDs have the advantage of at least an order of magnitude better accuracy. They do make "precision" thermocouples by using purer metals but accuracy is still limited by the cold junction compensation and they cost as much as an RTD solution making them only suitable for when you need increased accuracy outside of the temperature range where an RTD can be used. I was not addressing the use of RTDs 100 years ago but I can.
The need for a stable reference resistance is not unique to RTDs and it is hardly a problem; we have been building resistors out of Manganin which are stable over temperature and time for more than a century using the same construction techniques used to build RTDs. I have a pair of ESI impedance bridges which use wrapped wire (probably Manganin) over mica resistors (ESI made these in house) built in the 1960s which are still accurate to 3 or 4 significant digits. It is not real clear how accurate the bridges are because they are as good or better than any 0.1% resistor (or capacitor) that I have tested them with and that is about the limit of their precision anyway.
An accurate RTD thermometer without amplification or stable voltage or current sources could have been made 100 years ago using a balanced bridge, Kelvin Varley divider, and galvanometer. As a matter of fact, my old ESI impedance bridges work in exactly this way for DC resistance measurements; no stable reference voltage or current is needed and no amplification is needed.
This would have been pretty cumbersome way to measure temperature because you would need to balance the bridge for every reading and consult a chart to remove the non-linearity of the RTD so if you only needed 0.1F resolution, a mercury thermometer would have been better if only because of its ease of use in an adverse environment.
I would think that if something like that were to happen, it would have happened already.
That is a great argument up until the first time that it happens and there has to be a first time.
That this *hasn't* happened in spite of how obviously vulnerable these places are to even the typical American mass shooter, let alone some group with any kind of advanced planning and better weapons leads me to believe it ain't gonna happen, either. Like maybe the threat isn't nearly as real as the media makes it out to be, or maybe it's actually harder than you think to pull it off.
The threat just is not that real but eventually someone is going to use or try to use a UAV in a terrorist act.
A thermistor, with its highly nonlinear R=f(temp), is difficult to use to make an accurate thermometer. A thermocouple is better, but you need the cold junction reference.
Neither thermistors nor thermocouples would be used for precision electronic temperature measurement. RTDs are much better even without non-linear signal conditioning which itself is trivial to implement even in the analog domain.
See the law of large numbers [wikipedia.org]. It is possible to get arbitrarily good estimates by combining sufficiently many fuzzy individual measurements. This is not an invention of some communist cabal of climate scientists, but was formalised by Bernoulli and Poisson in the 18th and 19th century. And it is, of course, used in every modern tracking radar system, wether to keep moving bodies apart or to bring them together.
I do not doubt that anthropomorphic climate change is occurring (*) but there are many physical measurements where this does not apply; it assumes errors are uncorrelated and not systematic.
I am more familiar with this in connection with precision data acquisition where someone suggests taking more averages to reduce noise and get a more accurate result. Often this returns a more accurate wrong result because non-linearity or some other systematic error is not reduced.
In the case of temperature measurements, how does the calibration curve of the instruments change over time? Do all instruments of the same design or production run have the same errors? Did the immediate environment where the temperature measurements were made change over decades?
(*) I just do not care because politics will trump any effort to control it and it will only be used as an excuse for rent seeking.
Provided that you also encrypt DNS and the IP address is not associated with only one domain name, if you use HTTPS then the only thing the ISP will know is what IP you connected to and anything traffic analysis would reveal. TLS (transport layer security) protects all of HTTP.
I have been using a repurposed Celeron 300A as my main router running FreeBSD for years without problems. It has ECC memory, boots from Compact Flash attached to an IDE port, and I can alter the number of ethernet ports as needed.
If I were to do this today, I would use one of the cheap AM1 motherboards which support ECC memory and PCI or PCIe network cards as needed. If that does not allow enough ethernet ports, then a VLAN switch can be used as a port expanded. The AM1 CPUs are much faster than necessary for this kind of application so they can be forced to operate at a lower clock speed and core voltage to reduce power. With some cleverness, passive cooling can be used except perhaps for the power supply fan.
The memory is not needed for buffering although practically any PC hardware is going to have way more memory than needed. Where many embedded routers fail is that they lack enough space for state tables to support the number of connections commensurate with their throughput. Last time I checked, on FreeBSD this amounted to 1000 connections per megabyte of memory which may seem like a lot but manufacturers are surprisingly parsimonious with consumer networking equipment so this is a problem. On my AT&T U-Verse modem/router thing, I had to configure the UDP and TCP state timeouts to their minimum values to prevent filling the state table.
My 18+ year old Celeron 300A with 384M of ECC RAM has been running various incarnations of m0n0wall and pfsense for 10+ years and the only downtime has been to replace fans, make hardware upgrades, the one time the ice machine upstairs sprung a leak and dripped into the chassis. I have to restart my God forsaken AT&T U-Verse modem/router thing every few weeks.
Controlling imports and exports are one of the enumerated powers of the federal government and this power is often used to control things which would otherwise be outside of their constitutional reach. Whether something is readily available on the other side of the border is irrelevant and it does not even matter if something is manufactured on both sides; import and export restrictions still apply.
Slashdot Mirror
No and here is why.
For a CPU of a given complexity, a specific area is needed for transistors, routing, etc in a given process. If the process density goes up, then the power has to be lowered to maintain the same power/area because the area largely determines the thermal resistance and for the past few generations, high performance CPUs already operate with the junction temperature as high as is reliable. So power is proportional to chip area and higher density processes yield smaller chips so power has to be lower.
You can see this trend in Intel processors since about the Core2. The highest power models all have a power rating proportional to area and since more recent models are smaller, they have lower power ratings.
This is also why stacking memory on top of logic is not going to happen for anything except low performance logic.
The use of non-volatile SRAM which has an embedded lithium cell might be considered this when it is not replaceable. A lot of test equipment built starting in about 1990 has this "feature" even if it was implemented by replacing the previous solution of an external lithium cell, low power SRAM, and backup circuits while using the same printed circuit board.
More recent is the problem with aluminum electrolytic capacitors which have a well understood and predictable wear out mechanism. I have older ATX power supplies which have lasted well past their warranty date but newer ones from specific manufacturers reliably fail just out of warranty.
The irreversible damage occurs at both high and low voltage. If the battery capacity is limited to between say the 80% and 20% charge points, then the total energy delivered for a given decrease in capacity goes up so many shallow discharges are better than fewer deep discharges if they occur toward the middle of the capacity curve. Maintaining the battery charge state closer to 100% lowers the useful lifetime (lithium metal plates out?) but this effect is usually small compared to capacity lost do to general use. It would be more important in standby applications.
Supercapacitors also degrade at high voltage; in their case, increasing the voltage by 0.2 volts lowers the operating life by 10 times. Offhand I do not know what the equivalent specification is for lithium secondary cells.
The problem with the Hubble mirror had nothing to do with gravity.
The custom precision null corrector used during the figuring step was assembled incorrectly. When the standard null correctors used for the initial grinding showed spherical aberration during final testing, they were ignored as being too inaccurate compared to the custom precision null corrector.
The new cars will be built with a modern emissions compliant engine.
So like dominant pole compensation?
I am not sure what continuity tester you have experience with but all of my digital multimeters going back 20+ years produce a tone in addition to a reading on the display and respond within milliseconds. Running the test lead along a string of contacts or pins and listening works fine with them.
Do you have a citation for this?
I think you might be referring to the CALEA requirements which also apply to WISPs. There is no CALEA requirement to maintain the records you suggest however most ISPs do for other reasons and if you provide voice over IP services, the FCC does require location data to support E991. With or without these records, CALEA requires ISPs to aid law enforcement in identifying and wiretapping targets.
That is the situation some WISPs have faced in contested areas. The incumbent wired ISPs buy up tower space preventing the competing WIPS from using it and then flood the ISM bands with generated traffic. Canopy gear was popular for this because it uses TDMA instead of CSMA. More direct measures like jamming critical nodes with a flood of traffic on the same channel from a directional antenna at long range are also used.
This is why you either start by enforcing the license through legal means or immediately after the company first dissembles. If legal means to enforce the license are unavailable for whatever reason, then just publicize the violation and move on.
A firearm does not have to inflict a death to successfully protect the user and a majority of self defense firearm incidents result in either no death or the firearm not even being fired.
As for the difference between acquaintances or family members and strangers, most interactions are with the former rather than the later and the term acquaintance is used very very broadly in crime measurement.
The US government would never do such an underhanded thing. Just ask Julian Assange.
I made my post just to point out that thermocouples and thermisters are unsuitable for precision temperature measurement compared to RTDs and that has been the case for decades. But even with their complexities like cold junction compensation and lower accuracy, thermocouples are preferred to RTDs because of their temperature range and price. RTDs have the advantage of at least an order of magnitude better accuracy. They do make "precision" thermocouples by using purer metals but accuracy is still limited by the cold junction compensation and they cost as much as an RTD solution making them only suitable for when you need increased accuracy outside of the temperature range where an RTD can be used. I was not addressing the use of RTDs 100 years ago but I can.
The need for a stable reference resistance is not unique to RTDs and it is hardly a problem; we have been building resistors out of Manganin which are stable over temperature and time for more than a century using the same construction techniques used to build RTDs. I have a pair of ESI impedance bridges which use wrapped wire (probably Manganin) over mica resistors (ESI made these in house) built in the 1960s which are still accurate to 3 or 4 significant digits. It is not real clear how accurate the bridges are because they are as good or better than any 0.1% resistor (or capacitor) that I have tested them with and that is about the limit of their precision anyway.
An accurate RTD thermometer without amplification or stable voltage or current sources could have been made 100 years ago using a balanced bridge, Kelvin Varley divider, and galvanometer. As a matter of fact, my old ESI impedance bridges work in exactly this way for DC resistance measurements; no stable reference voltage or current is needed and no amplification is needed.
This would have been pretty cumbersome way to measure temperature because you would need to balance the bridge for every reading and consult a chart to remove the non-linearity of the RTD so if you only needed 0.1F resolution, a mercury thermometer would have been better if only because of its ease of use in an adverse environment.
That is a great argument up until the first time that it happens and there has to be a first time.
The threat just is not that real but eventually someone is going to use or try to use a UAV in a terrorist act.
You might test considerably in advance of the actual event to see what kind of results can be expected and to gauge any law enforcement response.
Neither thermistors nor thermocouples would be used for precision electronic temperature measurement. RTDs are much better even without non-linear signal conditioning which itself is trivial to implement even in the analog domain.
Otisburg?
I do not doubt that anthropomorphic climate change is occurring (*) but there are many physical measurements where this does not apply; it assumes errors are uncorrelated and not systematic.
I am more familiar with this in connection with precision data acquisition where someone suggests taking more averages to reduce noise and get a more accurate result. Often this returns a more accurate wrong result because non-linearity or some other systematic error is not reduced.
In the case of temperature measurements, how does the calibration curve of the instruments change over time? Do all instruments of the same design or production run have the same errors? Did the immediate environment where the temperature measurements were made change over decades?
(*) I just do not care because politics will trump any effort to control it and it will only be used as an excuse for rent seeking.
A statite would be a fine idea except Robert Forward is dead:
https://en.wikipedia.org/wiki/...
Provided that you also encrypt DNS and the IP address is not associated with only one domain name, if you use HTTPS then the only thing the ISP will know is what IP you connected to and anything traffic analysis would reveal. TLS (transport layer security) protects all of HTTP.
Encrypt Absolutely Everything
I have been using a repurposed Celeron 300A as my main router running FreeBSD for years without problems. It has ECC memory, boots from Compact Flash attached to an IDE port, and I can alter the number of ethernet ports as needed.
If I were to do this today, I would use one of the cheap AM1 motherboards which support ECC memory and PCI or PCIe network cards as needed. If that does not allow enough ethernet ports, then a VLAN switch can be used as a port expanded. The AM1 CPUs are much faster than necessary for this kind of application so they can be forced to operate at a lower clock speed and core voltage to reduce power. With some cleverness, passive cooling can be used except perhaps for the power supply fan.
The memory is not needed for buffering although practically any PC hardware is going to have way more memory than needed. Where many embedded routers fail is that they lack enough space for state tables to support the number of connections commensurate with their throughput. Last time I checked, on FreeBSD this amounted to 1000 connections per megabyte of memory which may seem like a lot but manufacturers are surprisingly parsimonious with consumer networking equipment so this is a problem. On my AT&T U-Verse modem/router thing, I had to configure the UDP and TCP state timeouts to their minimum values to prevent filling the state table.
My 18+ year old Celeron 300A with 384M of ECC RAM has been running various incarnations of m0n0wall and pfsense for 10+ years and the only downtime has been to replace fans, make hardware upgrades, the one time the ice machine upstairs sprung a leak and dripped into the chassis. I have to restart my God forsaken AT&T U-Verse modem/router thing every few weeks.
Controlling imports and exports are one of the enumerated powers of the federal government and this power is often used to control things which would otherwise be outside of their constitutional reach. Whether something is readily available on the other side of the border is irrelevant and it does not even matter if something is manufactured on both sides; import and export restrictions still apply.