Dude, how much clearer than "diffraction limited adaptive optics design" should it get for you to get it past your thick skull that this design is NOT atmospheric lensing limited. End of story right there. Shut up.
I have a Makita LXT drill and screwdriver and I've been beating the crap out of them, running the batteries down each time until the performance drops low enough to make me have to switch. No problems so far, even though I've had them 3 years or so. I'm drooling for the new brushless versions, but seriously I have no reason at all to replace what I have -- they work great in spite of using brushed DC motors.
With modern charge controllers, there is no overcharging of any sort -- a charged battery pack is not being charged anymore, that's it. Only after it self-discharges a bit -- enough so that it's detectable -- will it be topped up. The problem you have of course is that the battery pack is still treated as a whole and while charge controllers in mobile device battery packs will detect voltage on individual cells, still very few have the electronic bypass switches needed for cell charge equalization and cell bypass. Never mind that most power supplies are not designed to properly work with a battery that has a dead cell in it -- even if it's entirely technically feasible.
The problem with most batteries is that the designs are cheap and the cells are not individually controlled with bypass switches. Then you have a pack with just one bad cell and the whole pack is "bad". All the rechargeable battery pack failures I've seen were of this kind: one weak cell with all the others having lots of life in them. The bad cell issue is self-exacerbating: as soon as a cell has higher internal resistance than the others, it will always get overcharged and over-discharged, accelerating the deterioration that leads to further raise of internal resistance.
Voltage monitoring should be just one aspect of charge determination. Properly done charge monitoring integrates the electrical charge actually delivered to or taken out of the battery -- as in taking the time integral of the current. The cell voltage should only used together with other indicators (cell temperature!) to determine each cell's health and charge/discharge endpoints (fully charged and fully discharged). The cell voltage should not figure in normal battery "% remaining" indications -- those solely base on the charge taken out of the battery, and the estimation of the 100% charge capacity.
I think it has been now sufficiently demonstrated that the larger the lack of understanding of something, the more activism it generates. People who have no clue about basic high-school level genetics, thermodynamics and nuclear physics make for very good activists. The more clueless they are, the more vocal.
I really wonder what kind of an agricultural disaster could emerge from a presumably safe for human consumption ham sandwich or an orange... Isn't this just government propaganda from an agency that frankly said doesn't know what to do with itself anymore?
Run a farm, pack it in tubes, sell, profit?:) And it's "all natural" of course. Because we all know that natural things are teh bestest n' green n' all.
I'm sure it's someone's bucket of fun to work on dusty office desktop machines:/ When I do an upgrade, I purchase a supported brand-name configuration from Dell so that I only need a single Windows image disk for all machines, and I'm done. Those configurations receive much more testing than off-the-shelf mix-and-match CPU and motherboards. Build your own works when you have one desktop, or perhaps a couple in your home. For office environments where your time is worth much more than that of a janitor, you really don't want to touch the hardware unless it breaks down.
Why? Because sometimes reverse-engineering something that has been shown to work is cheaper than going through the entire development process from scratch. The fact that F-1 engines were built manually and designed without the use of electronic computers doesn't mean that they are much worse for it. Why the heck do you think a lot of Chinese products were reverse-engineered western designs?
It is certainly possible for machines to have capability to repair themselves. For robotic space exploration, though, it's simply cheaper to design a machine that can tolerate some likely failures while still completing the mission, while taking the chance that some other failures will end the mission. Remember: those kilograms delivered safely to Mars's surface are very expensive.
What were those "spectacular" failures of the Surveyor program? The 5 out of 7 missions were successful. Those were the IIRC the first U.S. lander missions to the moon, BTW. Sure the Pioneer program had a more dismal record.
The GPS doesn't need to run continuously. Once you're tracking via the camera and you've got a lock, the GPS is pretty useless -- camera supplies much better data then. The camera power is probably negligible compared to the power needed to run the GPU full-tilt.
Number of users is fairly irrelevant. You only need to slowly keep "nearby" google street view images downloaded and kept in a cache in the phone. All the processing is done in the phone -- well, perhaps the street view stuff could be pre-crunched, perhaps already run through a transform that would need to be done anyway before doing matching, but that's it. There's no big demand for server resources. I've seen rather advanced fiducial tracking on iPad in real time and it was done generically -- the "fiducial" could have been, say, your face:)
There's something to be said for T5 tubes in high-bay fixtures. You turn them on for the first time and it feels like going outside into full sunlight. Sure they get a tad dimmer after a while, but it's still a big win over legacy fluorescent.
There's a consensus of sort that power supplies are often the most underengineered things out there in any electronic device. Well, guess what, in a CFL or a LED the entire electronics are the power supply, there's nothing else. When a CFL fails, it's not because the bulb has failed, it's because the power supply is dead. It's certainly possible to engineer a power supply that will last, but such know-how is rare and expensive, and engineering management often doesn't understand that it takes real effort to make a long-lasting power supply. You have to qualify every single part, pretty much -- there's no such thing as letting the purchasing loose to get the best deal. If you want to make a CFL or a LED lamp that will last as long as the life of the light-emitting element, you need to do proper design, then qualify sample parts, then do extensive testing on prototypes, then purchase a batch of parts for a production run, then re-qualify all of those parts again, then have the boards assembled, then qualify the board assemblies, and only then you ship. That's what it takes to get a quality product out. That's what it takes to get a lamp out that will be so old by the time it gets replaced that the house might have changed owners a bunch of times in the meantime. Guess how it's done in real life on consumer CFL/LED bulbs, LOL.
Try it out. There's a point when things are CPU-bound in bursts. I'd say there's no point in upgrading if you're not getting a top-of-the-line desktop CPU architecture. i7 doesn't cost all that much. You don't need to get the fastest grade. An i7 with 8G of RAM and a ~100G SSD is what I'll upgrading the office machines with. They'll be capable enough until windows 7 goes out of extended support.
Pattern recognition only works if you've seen the place before and you knew well what the camera position and orientation was. For general purpose applications it'll work for from-the-street looking around when Google Street View data is available, but that's about it. You'd have to have fiducial marks or otherwise pre-train the system for a particular situation -- then it's no more a general purpose heading sensor anymore.
First of all, I'm not breaking anything. Secondly, your agent doesn't give a damn -- at least not when the house is not occupied any longer and there's nothing there but perhaps prop furniture. I have yet to see an agent who is, in the long run, not just a bored-like-a-teenager money-oriented salesperson. Besides, your agent doesn't necessarily do the showing. In fact, when I was looking for a house and visiting places, the sellers' agent was never present. My agent was opening the places and showing them to me. I'm not some knucklehead who'll just go and randomly destroy things, thank you very much. I'm just trying to make sure things are what they seem to be. My favorite trick is to try and push the screwdriver into the wood on the underside of first floor bathroom subfloor, around the toilet flange -- when there's access from the unfinished basement or crawl space. It usually lets me know right away if there are or were substantial leaks that have rotted the floor out. It's incredible how common of a problem it is -- it can progress with little obvious signs for years, while incurring thousands of U.S. dollars worth of damage if you were to hire a contractor to fix it. Even if you do it yourself it can cost ~$1500 if it's a bathroom with a shower stall, assuming you have to replace the floor, some walls, and the shower stall and toilet. With permits it might be more like $2000.
I'll be damned, RTGs indeed. Well, they've got a nice water-cooled thermal sink, so it's almost an ideal place to use an RTG. No need to deal with radiators for cooling like you have to on space probes. One learns something new every day, thanks!
Yep -- they were driving on an "empty road". The conscious brain was monopolized by the conversation or texting and didn't even have a chance to notice the person on the road. Part of the problem is that our learned driving doesn't normally include obstacle avoidance -- we have no reflexes to avoid a person on the road, thus without the conscious attention paid to it, we just don't.
It's interesting: obstacle avoidance is not something you do every day, so you never learn it enough to get your automatic brain to take over. Alas, this is something one could learn in a simulator, and would need to stay current on. Then you could be distracted and would swerve past humans just fine -- although it'd require way more training to make such swerves properly avoid other traffic. Again, you'd need to train in a simulator and would need to experience a large and controlled number of on-the-road situations. In normal driving, we rely on our conscious attention to do all the planning - how to fit in between other cars, what speed changes to execute, etc. We could relegate all of that to our parallel, real-time pre-conscious processing part of the brain, but it's impossible to do merely sitting in a car on a road. It'd cost way too much in junked cars:/
Also remember that the fast, pre-conscious part of the brain lacks any reasoning capabilities -- it cannot infer, it cannot bring up memories, etc. It doesn't make sense of anything, it just follows what matches best the situations it has experienced so far. It sure can do what appears to be "logical" processing (say boolean operations work if you learn them) -- but it cannot apply reasoning of any sort. So, unfortunately, any maneuvers thus executed would not necessarily make any sense, nor would the necessarily give you any longer-term advantage. They might be OK to avoid a bunch of cars/obstacles this very second, just to be rolled over by a truck that was a couple hundred feet back but had no way to brake in time -- that's something you'd figure out quickly if you looked in the rearview mirror and let your conscious attention figure it out. You'd need to train (and re-train) for this very situation in order to have it available as an automatic skill -- you see where this goes, there's an exponential explosions of possible scenarios, rather quickly it's impossible to train for anything more since in real life it will surely occur with some other modifying conditions that will make the learned responses obsolete.
Things such as lane maintenance and following distance maintenance are being re-trained constantly whenever you drive and pay attention -- the conscious/attentive brain is tuning the behavior of the preconscious. If you consistently drive without paying attention, your automated actions slowly get detuned.
There is also the observation that people are thoroughly confused when they slam on the brake but accelerate instead. They were not paying attention, the automation in the brain did the "slam the brake" reaction, but it missed the brake and slammed accelerator instead. The conscious attention, once it gets to it, still has to reconcile the presupposed model of the situation (foot on the brake) with what's really going on (WTF am I going faster?). For some reason we don't understand yet, it takes long to do such reconciliations -- it gets worse the farther out the old model is with relation to reality. When you're sufficiently confused, things take long enough that when it happens to a fighter jet pilot, they'll end up underground before they figure it out. Even for commercial jet pilots it can be deadly. That's most likely what killed everyone onboard AF447, and that's what almost, almost did in China Airlines flight 006.
Slashdot Mirror
Dude, how much clearer than "diffraction limited adaptive optics design" should it get for you to get it past your thick skull that this design is NOT atmospheric lensing limited. End of story right there. Shut up.
I have a Makita LXT drill and screwdriver and I've been beating the crap out of them, running the batteries down each time until the performance drops low enough to make me have to switch. No problems so far, even though I've had them 3 years or so. I'm drooling for the new brushless versions, but seriously I have no reason at all to replace what I have -- they work great in spite of using brushed DC motors.
With modern charge controllers, there is no overcharging of any sort -- a charged battery pack is not being charged anymore, that's it. Only after it self-discharges a bit -- enough so that it's detectable -- will it be topped up. The problem you have of course is that the battery pack is still treated as a whole and while charge controllers in mobile device battery packs will detect voltage on individual cells, still very few have the electronic bypass switches needed for cell charge equalization and cell bypass. Never mind that most power supplies are not designed to properly work with a battery that has a dead cell in it -- even if it's entirely technically feasible.
The problem with most batteries is that the designs are cheap and the cells are not individually controlled with bypass switches. Then you have a pack with just one bad cell and the whole pack is "bad". All the rechargeable battery pack failures I've seen were of this kind: one weak cell with all the others having lots of life in them. The bad cell issue is self-exacerbating: as soon as a cell has higher internal resistance than the others, it will always get overcharged and over-discharged, accelerating the deterioration that leads to further raise of internal resistance.
Voltage monitoring should be just one aspect of charge determination. Properly done charge monitoring integrates the electrical charge actually delivered to or taken out of the battery -- as in taking the time integral of the current. The cell voltage should only used together with other indicators (cell temperature!) to determine each cell's health and charge/discharge endpoints (fully charged and fully discharged). The cell voltage should not figure in normal battery "% remaining" indications -- those solely base on the charge taken out of the battery, and the estimation of the 100% charge capacity.
We need a refinery for, it then :) At least it will smell like poop, not like petrochemicals.
I think it has been now sufficiently demonstrated that the larger the lack of understanding of something, the more activism it generates. People who have no clue about basic high-school level genetics, thermodynamics and nuclear physics make for very good activists. The more clueless they are, the more vocal.
I really wonder what kind of an agricultural disaster could emerge from a presumably safe for human consumption ham sandwich or an orange... Isn't this just government propaganda from an agency that frankly said doesn't know what to do with itself anymore?
This may be perhaps the most overlooked post of the week.
their poop is profuse and a lot like epoxy glue
Run a farm, pack it in tubes, sell, profit? :) And it's "all natural" of course. Because we all know that natural things are teh bestest n' green n' all.
Do those worms' larva or eggs survive cooking, though? I think all it takes is making sure you cook them well and that's it. I love snails.
I'm sure it's someone's bucket of fun to work on dusty office desktop machines :/ When I do an upgrade, I purchase a supported brand-name configuration from Dell so that I only need a single Windows image disk for all machines, and I'm done. Those configurations receive much more testing than off-the-shelf mix-and-match CPU and motherboards. Build your own works when you have one desktop, or perhaps a couple in your home. For office environments where your time is worth much more than that of a janitor, you really don't want to touch the hardware unless it breaks down.
Why? Because sometimes reverse-engineering something that has been shown to work is cheaper than going through the entire development process from scratch. The fact that F-1 engines were built manually and designed without the use of electronic computers doesn't mean that they are much worse for it. Why the heck do you think a lot of Chinese products were reverse-engineered western designs?
It is certainly possible for machines to have capability to repair themselves. For robotic space exploration, though, it's simply cheaper to design a machine that can tolerate some likely failures while still completing the mission, while taking the chance that some other failures will end the mission. Remember: those kilograms delivered safely to Mars's surface are very expensive.
What were those "spectacular" failures of the Surveyor program? The 5 out of 7 missions were successful. Those were the IIRC the first U.S. lander missions to the moon, BTW. Sure the Pioneer program had a more dismal record.
The GPS doesn't need to run continuously. Once you're tracking via the camera and you've got a lock, the GPS is pretty useless -- camera supplies much better data then. The camera power is probably negligible compared to the power needed to run the GPU full-tilt.
Number of users is fairly irrelevant. You only need to slowly keep "nearby" google street view images downloaded and kept in a cache in the phone. All the processing is done in the phone -- well, perhaps the street view stuff could be pre-crunched, perhaps already run through a transform that would need to be done anyway before doing matching, but that's it. There's no big demand for server resources. I've seen rather advanced fiducial tracking on iPad in real time and it was done generically -- the "fiducial" could have been, say, your face :)
There's something to be said for T5 tubes in high-bay fixtures. You turn them on for the first time and it feels like going outside into full sunlight. Sure they get a tad dimmer after a while, but it's still a big win over legacy fluorescent.
There's a consensus of sort that power supplies are often the most underengineered things out there in any electronic device. Well, guess what, in a CFL or a LED the entire electronics are the power supply, there's nothing else. When a CFL fails, it's not because the bulb has failed, it's because the power supply is dead. It's certainly possible to engineer a power supply that will last, but such know-how is rare and expensive, and engineering management often doesn't understand that it takes real effort to make a long-lasting power supply. You have to qualify every single part, pretty much -- there's no such thing as letting the purchasing loose to get the best deal. If you want to make a CFL or a LED lamp that will last as long as the life of the light-emitting element, you need to do proper design, then qualify sample parts, then do extensive testing on prototypes, then purchase a batch of parts for a production run, then re-qualify all of those parts again, then have the boards assembled, then qualify the board assemblies, and only then you ship. That's what it takes to get a quality product out. That's what it takes to get a lamp out that will be so old by the time it gets replaced that the house might have changed owners a bunch of times in the meantime. Guess how it's done in real life on consumer CFL/LED bulbs, LOL.
Try it out. There's a point when things are CPU-bound in bursts. I'd say there's no point in upgrading if you're not getting a top-of-the-line desktop CPU architecture. i7 doesn't cost all that much. You don't need to get the fastest grade. An i7 with 8G of RAM and a ~100G SSD is what I'll upgrading the office machines with. They'll be capable enough until windows 7 goes out of extended support.
Pattern recognition only works if you've seen the place before and you knew well what the camera position and orientation was. For general purpose applications it'll work for from-the-street looking around when Google Street View data is available, but that's about it. You'd have to have fiducial marks or otherwise pre-train the system for a particular situation -- then it's no more a general purpose heading sensor anymore.
See -- it can be done. With some rework :)
First of all, I'm not breaking anything. Secondly, your agent doesn't give a damn -- at least not when the house is not occupied any longer and there's nothing there but perhaps prop furniture. I have yet to see an agent who is, in the long run, not just a bored-like-a-teenager money-oriented salesperson. Besides, your agent doesn't necessarily do the showing. In fact, when I was looking for a house and visiting places, the sellers' agent was never present. My agent was opening the places and showing them to me. I'm not some knucklehead who'll just go and randomly destroy things, thank you very much. I'm just trying to make sure things are what they seem to be. My favorite trick is to try and push the screwdriver into the wood on the underside of first floor bathroom subfloor, around the toilet flange -- when there's access from the unfinished basement or crawl space. It usually lets me know right away if there are or were substantial leaks that have rotted the floor out. It's incredible how common of a problem it is -- it can progress with little obvious signs for years, while incurring thousands of U.S. dollars worth of damage if you were to hire a contractor to fix it. Even if you do it yourself it can cost ~$1500 if it's a bathroom with a shower stall, assuming you have to replace the floor, some walls, and the shower stall and toilet. With permits it might be more like $2000.
I'll be damned, RTGs indeed. Well, they've got a nice water-cooled thermal sink, so it's almost an ideal place to use an RTG. No need to deal with radiators for cooling like you have to on space probes. One learns something new every day, thanks!
Yep -- they were driving on an "empty road". The conscious brain was monopolized by the conversation or texting and didn't even have a chance to notice the person on the road. Part of the problem is that our learned driving doesn't normally include obstacle avoidance -- we have no reflexes to avoid a person on the road, thus without the conscious attention paid to it, we just don't.
It's interesting: obstacle avoidance is not something you do every day, so you never learn it enough to get your automatic brain to take over. Alas, this is something one could learn in a simulator, and would need to stay current on. Then you could be distracted and would swerve past humans just fine -- although it'd require way more training to make such swerves properly avoid other traffic. Again, you'd need to train in a simulator and would need to experience a large and controlled number of on-the-road situations. In normal driving, we rely on our conscious attention to do all the planning - how to fit in between other cars, what speed changes to execute, etc. We could relegate all of that to our parallel, real-time pre-conscious processing part of the brain, but it's impossible to do merely sitting in a car on a road. It'd cost way too much in junked cars :/
Also remember that the fast, pre-conscious part of the brain lacks any reasoning capabilities -- it cannot infer, it cannot bring up memories, etc. It doesn't make sense of anything, it just follows what matches best the situations it has experienced so far. It sure can do what appears to be "logical" processing (say boolean operations work if you learn them) -- but it cannot apply reasoning of any sort. So, unfortunately, any maneuvers thus executed would not necessarily make any sense, nor would the necessarily give you any longer-term advantage. They might be OK to avoid a bunch of cars/obstacles this very second, just to be rolled over by a truck that was a couple hundred feet back but had no way to brake in time -- that's something you'd figure out quickly if you looked in the rearview mirror and let your conscious attention figure it out. You'd need to train (and re-train) for this very situation in order to have it available as an automatic skill -- you see where this goes, there's an exponential explosions of possible scenarios, rather quickly it's impossible to train for anything more since in real life it will surely occur with some other modifying conditions that will make the learned responses obsolete.
Things such as lane maintenance and following distance maintenance are being re-trained constantly whenever you drive and pay attention -- the conscious/attentive brain is tuning the behavior of the preconscious. If you consistently drive without paying attention, your automated actions slowly get detuned.
There is also the observation that people are thoroughly confused when they slam on the brake but accelerate instead. They were not paying attention, the automation in the brain did the "slam the brake" reaction, but it missed the brake and slammed accelerator instead. The conscious attention, once it gets to it, still has to reconcile the presupposed model of the situation (foot on the brake) with what's really going on (WTF am I going faster?). For some reason we don't understand yet, it takes long to do such reconciliations -- it gets worse the farther out the old model is with relation to reality. When you're sufficiently confused, things take long enough that when it happens to a fighter jet pilot, they'll end up underground before they figure it out. Even for commercial jet pilots it can be deadly. That's most likely what killed everyone onboard AF447, and that's what almost, almost did in China Airlines flight 006.