WEll, I agree, they're cool looking. But to have your computing speed be limited by the de-ionization time of a gas is less than thrilling. You think a VAX-750 with ten users was slow....:)
There's the problem of deciding what's a real computer-- do you include things that can crunch numbers, but have a hard-wired program, or have a program, but it's on a loop of paper tape, or have a program, but it's wired onto a plugboard. The Harwell machine is programmable, but the program is on a loop of paper tape, making anything other than one simple loop very problematical.
Also its data storage is in a few cold-cathode Dekatrons, which are basically overachieving neon lights. They limit the counting-up speed to about 20,000 increments per second, just barely in the electronic realm, and much slower than anything using real vacuum tubes. And it uses a lot of mechanical relays, further limiting its speed and making it a very marginal computer in any modern sense of the word.
I don't quite get how this number is by itself significant of something or other.
Yes, it's a small number compared to the total number of base pairs. But let's remember, this is a digital code. Some bits are more significant than the average. You can't just say "the gene is 99.994% correct". Just one stop codon can break a whole gene. We are all walking around with one-bit errors in our genes that used to be able to make Vitamin C and an anti-HIV factor. Just one-bit errors but oh, very high consequences, like scurvy and AIDS.
Those poor Japanese get screwed every which way but loose. They have been conditioned to pay $60 for a melon, and now $21 billion for something that will never work. And they don't complain!
Think of this more as a big wet kiss for the Japanese space industry. Just like "Star Wars" was for our military-industrial complex.
There's no way in heck this will ever get within a factor of 100 of being practical or economical.
I guess my brain refused to see "polyfuse", because it's a very ridiculous solution, for many reasons:
(1) Those devices are thermal-- which means it takes a considerable fraction of a second for it to ramp up in resistance. The microprocessor can't tolerate power dips for more than a microsecond or so, so this solution is about half a million times too slow.
(2) PTC's are expensive-- like 60 cents in quantity.
(3) At best this is a band-aid-- it does not prevent the problem, it just keeps a fire from starting.
>Um, I'm afraid you just defeated the pulse-[b]width[/b] modulation.
Nope. I chose the time constant so it will faithfully pass any PWM going on. I'm assuming they're using a typical PWM frequency of 5 to 50 KHz. If it's below that, which is unlikely, one would trim the resistor value upwards.
BTW there is one possible issue with my R/C solution, where if it runs at an extreme duty cycle, might require one more 4 cent component. Funny that you have not mentioned that, being a PWM expert and all.
Thanks for the reply. It's heartwarming to see a company responding to potential customers.
You might consider a less fatal solution that does not involve the irreversible blowing of expensive fuses:
Between the BOOST output from the microprocessor and the gate of the MOSFET, put a 0.01uF capacitor in series, and a 100K ohm resistor from gate to ground. Total cost: about 15 cents. (The cheapest fuse I can find costs a bit more than that.)
With this kind of AC coupling, the ONLY way the MOSFET can get turned on is if the micro puts out a positive-going edge. The RC time constant is about 1/1000 of a second, so even if the output gets stuck high, the gate will go low within a millisecond.
Sometimes those old analog components are real lifesavers!
Pardon me, but is not the PWM software controlled?
And what happens if:
(1) The initialization code gets hung up for some reason before the PWM is programmed? You're okay if the outputs unconditionally default to LOW, but if they default to high or tri-state, kaboom.
(2) What if the code goes off into the boonies and starts executing random code? I've had this happen where the random code just happened to have some "write to interrupt controller and disk" op codes! Lost everything. But at least I did not start a fire.
Having done more than a bit of real-time programming, I've learned that it's a bad idea to start something in software that you had better be able to finish or something very bad happens.
One-sixth of a schmitt trigger and one resistor and one capacitor make a simple, totally bulletproof and unhangable oscillator. That would be a much better choice IMHO.
Using a microcontroller to supply the BOOST clock is a poor idea. If the software stumbles and leaves the BOOST line high, you have a dead short across the power supply.
Perhaps fortunately, this might drag the power to the CPU low enough to cause it to reboot, which might restore operation. Or the low voltage might cause it to hang.
Sometimes good old hardware is the right solution.
It was only "open source" because the code had to be hand-crafted and re-assembled for each particular configuration. You young kids expect softwar to be rife with XML configuration files, and virtual methods, and hooks. Back in those days the code had to fit into 4K addressable segments, so they could not AFFORD to even think of opening up a file and reading configuration info, or having a table of external procedure hooks. More likely the configuration constants were not even separate, they were convenient opcodes. For instance, if you knew a 707 at this airline always had 112 seats, you'd recall that the HCF opcode happened to be 112 decimal, so you'd compare the seat count against that opcode. All you kids with your fancy separate data! Also it was extreme luxury to have a procedure hook (or as you callem nowadys "virtual methods"). You see you could only call within the current 4K block, and any addresses you wished to pass had similar or worse restrictions. And there was darnlittle dynamc linking available in old IBM DOS, so you could not call anything that had not been linked in last week at the weekly build (which took hours).
When I do the math, a square kilometer sail weighing 150 kilograms can only weigh 0.15 grams per square meter. If the material is only 0.0025 cm thick, it would have to have a density of 0.006. It's hard to find anything solid that is that light.
And that's ignoring the non-negligible weight of whatever lashes the 150Kg payload to the square kilometer of sail.
And if this thing is going to pull 0.6G, you need some kind of structure that can transfer the force to the payload without collapsing the sail. Quite a trick.
Also at 0.1 AU from the Sun, the thing is going to get mighty hot.
It won't discharge very quickly because the ionosphere is a mighty poor conductor. A tower stuck into the ionosphere is only going to discharge the molecules it touches-- anything more than an inch away is effectively insulated.
It's not your fault, but many companies are compelled to keep info about ex customers just in case you're a nogoodnik and try to scam them again. They have no way of verifying your "new" id number of your choosing and even if they did they could not crosscheck it next time when you give a different type of Id or other DL number
>And what if your IC-based car has the transmission crap out at 10 years?
Trannies are a mature technology made by the millions so they're not very expensive. Very few cars get scrapped when the transmission goes out.
>In theory, the Volt's electric drivetrain should be really reliable and require almost no maintance, and the Volt's gasoline engine which only has to run occasionally and at a constant speed should likewise last a long time and require little maintaince*.
There's the rub-- GM has a poor record with any kind of rushed introduction of new technology. let me mention: The Corvette, The Corvair, the Vega, The X's, The Diesels, and now The Volt.
They tend to rush things into production as the marketing guys are in charge of the schedule, not the engineers.
The chances the Volt will be nice and trouble-free are not very good.
There's more than a small gremlin in this plan-- transport.
Cement and concrete are always made close to their destination, because, the stuff is, like, really, really heavy.
Now one suspects that the required chemicals for this new CO2 absorbing stuff are not equitably distributed.
So the places without the stuff would need to have the stuff trucked, barged, or railed in. That would send the price of this concrete through the roof. Not to mention releasing more CO2 from all the diesel engines pushing the stuff to its destination.
It might be more cost-effective to just mine the stuff, bake it, and just lay it out in the open to absorb CO2. Forget about making concrete out of it.
So we're going to pay $40K for this little bugger. Well, somebody might, but not anybody I know.
Let's do a little figuring. Let's say you can buy a comparable new IC engine'd car for $20K. Now how long will it take you to to spend $20K on gas? At $4 a gallon that's 5,000 gallons. At 30MPG that's good for 150,000 miles of driving. At a (high) 15,000 miles a year of driving that's ten years.
So after ten years you have just equaled the cost of the Volt. Except after ten years there are quite a few years left on the IC car but the Volt needs a new battery which costs much more than what the car is worth, so that effectively makes it worthless.
And if you put the $20K you did not spend in into the bank at 2.5%, you'd have the $20K plus another $2.56K in interest. Hmmm, let's see, would I rather have a reliable car and $25K at the end, or a unknown first-try car and $0?
Using 3.14159 and 58.44, those are particularly poor examples.
First, the value of pi can't be written down exactly, in fact the term "pi" is the shortest and best, so that's not a good example.
And the gm/M of table salt isn't too keen an example either-- that number is going to vary depending on the isotopic composition of the sodium and chlorine.
So maybe "pi" and "table salt" are already good semantic descriptors.
Thanks for the basic numbers. But your extrapolation is wildly like 10^10 times optimistic.
A battery needs two electrodes. There's no way to take a cubic anything of carbon black and put electrodes around each particle, AND have an electrical connection to each particle, all insulated from the opposite electrode.
Also running your proposed battery into a matched load would result in half the power being dissipated in the battery. 97.5 megawatts in one cubic meter is going to heat up in a millisecond and kill all the bacteria.
Scotty is right. This idea is ludicrous. Sending power as magnetic fields is a major fail. The near-field which they're touting as a panacea, it inevitably falls off as the cube of the distance. So you need a sending coil about as big across as the distance. You want to hang a 10-foot coil on the ceiling to power your laptop in a 10 ft radius?
And there won't be a single standards body that will approve pumping many watts of 30m waves into living spaces.
You should look up "the Dave Barry principle of using exaggeration as humor".
AFAIK France spent about 50 billion francs just on the Super-Phenix breeder, which ran for about 18 months over 20 years. Now decomissioned. I would not call that a success by any means.
The only other "French" breeder I can find is ONE Russian BN-600, which is not only tiny, it requires highly (25%) enriched uranium. Not exactly a cheap way to go.
365 volts at 1000 amps is about ten times the available power at the average house. In order to carry this off you'd need a major upgrade of the wires going to each house, plus some interlocks so only 10% of the houses can be charging at any time.
The charging rate of 365 kilowatts, assuming a battery of 90% charging efficiency, means the battery needs 36.5 kilowatts of cooling while charging. That's one HUGE fan, or a complex liquid cooling loop.
We don't know the temperature coefficient of the cells they are considering. If their temperature coefficient goes the wrong way, you can't charge the cells in their series configuration. Just one weak cell in a string and it would tend to run away thermally and wreck at least its string, or worse.
It's sad to see students at a major university being so clueless about basic energy equations.
Breeders have been tried, to the extent of about 20 billion dollars, over the last 40 years. All have failed. It's really hard to make something that can run with the very high neutron fluxes for years and years. There are only so many different materials and alloys to choose from and they all tend to fall apart after a while with 10^38 neutrons per cm^2 per second buzzing thru them.
In addition we may have passed the point of no return re breeders-- i.e. if we had breeders right now, there isn't enough uranium left to run the current bunch of reactors and breed any usable amount of new material.
There's also the slight problem of building plutonium-burning reactors and not losing a few kilos to the bad guys.
Slashdot Mirror
WEll, I agree, they're cool looking. But to have your computing speed be limited by the de-ionization time of a gas is less than thrilling. You think a VAX-750 with ten users was slow.... :)
There's the problem of deciding what's a real computer-- do you include things that can crunch numbers, but have a hard-wired program, or have a program, but it's on a loop of paper tape, or have a program, but it's wired onto a plugboard. The Harwell machine is programmable, but the program is on a loop of paper tape, making anything other than one simple loop very problematical.
Also its data storage is in a few cold-cathode Dekatrons, which are basically overachieving neon lights. They limit the counting-up speed to about 20,000 increments per second, just barely in the electronic realm, and much slower than anything using real vacuum tubes. And it uses a lot of mechanical relays, further limiting its speed and making it a very marginal computer in any modern sense of the word.
I don't quite get how this number is by itself significant of something or other.
Yes, it's a small number compared to the total number of base pairs. But let's remember, this is a digital code. Some bits are more significant than the average.
You can't just say "the gene is 99.994% correct". Just one stop codon can break a whole gene. We are all walking around with one-bit errors in our genes that used to be able to make Vitamin C and an anti-HIV factor. Just one-bit errors but oh, very high consequences, like scurvy and AIDS.
Those poor Japanese get screwed every which way but loose. They have been conditioned to pay $60 for a melon, and now $21 billion for something that will never work. And they don't complain!
Think of this more as a big wet kiss for the Japanese space industry. Just like "Star Wars" was for our military-industrial complex.
There's no way in heck this will ever get within a factor of 100 of being practical or economical.
>The OP said "polyfuse". These are self-resetting polymer fuses, not irreversible devices. Example: http://www.circuitprotection.com/polyswitch.asp [circuitprotection.com]
I guess my brain refused to see "polyfuse", because it's a very ridiculous solution, for many reasons:
(1) Those devices are thermal-- which means it takes a considerable fraction of a second for it to ramp up in resistance. The microprocessor can't tolerate power dips for more than a microsecond or so, so this solution is about half a million times too slow.
(2) PTC's are expensive-- like 60 cents in quantity.
(3) At best this is a band-aid-- it does not prevent the problem, it just keeps a fire from starting.
>Um, I'm afraid you just defeated the pulse-[b]width[/b] modulation.
Nope. I chose the time constant so it will faithfully pass any PWM going on. I'm assuming they're using a typical PWM frequency of 5 to 50 KHz. If it's below that, which is unlikely, one would trim the resistor value upwards.
BTW there is one possible issue with my R/C solution, where if it runs at an extreme duty cycle, might require one more 4 cent component. Funny that you have not mentioned that, being a PWM expert and all.
Thanks for the reply. It's heartwarming to see a company responding to potential customers.
You might consider a less fatal solution that does not involve the irreversible blowing of expensive fuses:
Between the BOOST output from the microprocessor and the gate of the MOSFET, put a 0.01uF capacitor in series, and a 100K ohm resistor from gate to ground. Total cost: about 15 cents. (The cheapest fuse I can find costs a bit more than that.)
With this kind of AC coupling, the ONLY way the MOSFET can get turned on is if the micro puts out a positive-going edge. The RC time constant is about 1/1000 of a second, so even if the output gets stuck high, the gate will go low within a millisecond.
Sometimes those old analog components are real lifesavers!
Regards,
A_H
It's thinking like this that gets my ulcer going.
What happens if:
You substitute a "compatible" microprocessor which just so happens to have been "enhanced" with the option to tri-state the I/O lines on powerup?
Your cat brushes against the acrylic case and sets off a 5500V static discharge, toggling one bit of the PC?
A nearby lightning strike sends a glitch down the AC line and by capacitive coupling toggles a bit in a register.
A drunk driver hits a power pole three miles away and the AC power glitches for just long enough to drop the supply voltage to scramble a register.
A weak cosmic ray strikes the chip, toggling some register bits.
--------------
Is the reasonable response to any of these events a dead short across the power supply, smoke, and possibly fire?
Cmon folks, try thinking just a teensy bit outside of "the CPU and program can never goof up" box.
Pardon me, but is not the PWM software controlled?
And what happens if:
(1) The initialization code gets hung up for some reason before the PWM is programmed? You're okay if the outputs unconditionally default to LOW, but if they default to high or tri-state, kaboom.
(2) What if the code goes off into the boonies and starts executing random code? I've had this happen where the random code just happened to have some "write to interrupt controller and disk" op codes! Lost everything. But at least I did not start a fire.
Having done more than a bit of real-time programming, I've learned that it's a bad idea to start something in software that you had better be able to finish or something very bad happens.
One-sixth of a schmitt trigger and one resistor and one capacitor make a simple, totally bulletproof and unhangable oscillator. That would be a much better choice IMHO.
Nice looking clock, but:
Using a microcontroller to supply the BOOST clock is a poor idea. If the software stumbles and leaves the BOOST line high, you have a dead short across the power supply.
Perhaps fortunately, this might drag the power to the CPU low enough to cause it to reboot, which might restore operation. Or the low voltage might cause it to hang.
Sometimes good old hardware is the right solution.
It was only "open source" because the code had to be hand-crafted and re-assembled for each particular configuration. You young kids expect softwar to be rife with XML configuration files, and virtual methods, and hooks. Back in those days the code had to fit into 4K addressable segments, so they could not AFFORD to even think of opening up a file and reading configuration info, or having a table of external procedure hooks. More likely the configuration constants were not even separate, they were convenient opcodes. For instance, if you knew a 707 at this airline always had 112 seats, you'd recall that the HCF opcode happened to be 112 decimal, so you'd compare the seat count against that opcode. All you kids with your fancy separate data! Also it was extreme luxury to have a procedure hook (or as you callem nowadys "virtual methods"). You see you could only call within the current 4K block, and any addresses you wished to pass had similar or worse restrictions. And there was darnlittle dynamc linking available in old IBM DOS, so you could not call anything that had not been linked in last week at the weekly build (which took hours).
oookay, so the stuff can be as dense as 0.012.
PET's density is 1.35 so it's still a bit over 100 times too heavy. And I doubt if PET can stand being 0.1 AU from the Sun for very long.
When I do the math, a square kilometer sail weighing 150 kilograms can only weigh 0.15 grams per square meter. If the material is only 0.0025 cm thick, it would have to have a density of 0.006. It's hard to find anything solid that is that light.
And that's ignoring the non-negligible weight of whatever lashes the 150Kg payload to the square kilometer of sail.
And if this thing is going to pull 0.6G, you need some kind of structure that can transfer the force to the payload without collapsing the sail. Quite a trick.
Also at 0.1 AU from the Sun, the thing is going to get mighty hot.
It won't discharge very quickly because the ionosphere is a mighty poor conductor. A tower stuck into the ionosphere is only going to discharge the molecules it touches-- anything more than an inch away is effectively insulated.
It's not your fault, but many companies are compelled to keep info about ex customers just in case you're a nogoodnik and try to scam them again.
They have no way of verifying your "new" id number of your choosing and even if they did they could not crosscheck it next time when you give a different type of Id or other DL number
>And what if your IC-based car has the transmission crap out at 10 years?
Trannies are a mature technology made by the millions so they're not very expensive. Very few cars get scrapped when the transmission goes out.
>In theory, the Volt's electric drivetrain should be really reliable and require almost no maintance, and the Volt's gasoline engine which only has to run occasionally and at a constant speed should likewise last a long time and require little maintaince*.
There's the rub-- GM has a poor record with any kind of rushed introduction of new technology. let me mention: The Corvette, The Corvair, the Vega, The X's, The Diesels, and now The Volt.
They tend to rush things into production as the marketing guys are in charge of the schedule, not the engineers.
The chances the Volt will be nice and trouble-free are not very good.
There's more than a small gremlin in this plan-- transport.
Cement and concrete are always made close to their destination, because, the stuff is, like, really, really heavy.
Now one suspects that the required chemicals for this new CO2 absorbing stuff are not equitably distributed.
So the places without the stuff would need to have the stuff trucked, barged, or railed in. That would send the
price of this concrete through the roof. Not to mention releasing more CO2 from all the diesel engines pushing the stuff
to its destination.
It might be more cost-effective to just mine the stuff, bake it, and just lay it out in the open to absorb CO2. Forget about
making concrete out of it.
So we're going to pay $40K for this little bugger. Well, somebody might, but not anybody I know.
Let's do a little figuring. Let's say you can buy a comparable new IC engine'd car for $20K.
Now how long will it take you to to spend $20K on gas? At $4 a gallon that's 5,000 gallons.
At 30MPG that's good for 150,000 miles of driving. At a (high) 15,000 miles a year of driving that's ten years.
So after ten years you have just equaled the cost of the Volt. Except after ten years there are quite a few years left
on the IC car but the Volt needs a new battery which costs much more than what the car is worth, so
that effectively makes it worthless.
And if you put the $20K you did not spend in into the bank at 2.5%, you'd have the $20K plus another $2.56K in interest.
Hmmm, let's see, would I rather have a reliable car and $25K at the end, or a unknown first-try car and $0?
Using 3.14159 and 58.44, those are particularly poor examples.
First, the value of pi can't be written down exactly, in fact the term "pi" is the shortest and best, so that's not a good example.
And the gm/M of table salt isn't too keen an example either-- that number is going to vary depending on the isotopic composition of the sodium and chlorine.
So maybe "pi" and "table salt" are already good semantic descriptors.
Thanks for the basic numbers. But your extrapolation is wildly like 10^10 times optimistic.
A battery needs two electrodes. There's no way to take a cubic anything of carbon black and put electrodes around each particle, AND have an electrical connection to each particle, all insulated from the opposite electrode.
Also running your proposed battery into a matched load would result in half the power being dissipated in the battery. 97.5 megawatts in one cubic meter is going to heat up in a millisecond and kill all the bacteria.
Let's not get too excited about this.
There's just one parameter of interest-- how much power?
One suspects it's not a lot.
I can see biologists getting excited about picoamps and millivolts per square centimeter.
But as a practical source of electricity, that would be much less practical and economical than the lemon, penny, and nickel battery.
Even if they evolved to producing 1,000 times as much electricity, we could still be talking about nanoamps, millivolts, and picowatts.
TFA, kiinda ludicrous.
First of all, how do you hook up an oscilloscope to a parking meter without disassembling it?
Then, what could you get from that that you could not get just by reading the card stripe with a $29 card reader?
One suspects this "black hat" just read a valid card on a card reader, swiped it in a parking meter, then re-read the card and noted the changes.
In any case, since it's unlikely that the parking meters are networked, all he had to do was clone a good card and he's set.
No oscilloscopes or trickery needed.
Scotty is right. This idea is ludicrous. Sending power as magnetic fields is a major fail. The near-field which they're touting as a panacea, it inevitably falls off as the cube of the distance. So you need a sending coil about as big across as the distance. You want to hang a 10-foot coil on the ceiling to power your laptop in a 10 ft radius?
And there won't be a single standards body that will approve pumping many watts of 30m waves into living spaces.
You should look up "the Dave Barry principle of using exaggeration as humor".
AFAIK France spent about 50 billion francs just on the Super-Phenix breeder, which ran for about 18 months over 20 years. Now decomissioned. I would not call that a success by any means.
The only other "French" breeder I can find is ONE Russian BN-600, which is not only tiny, it requires highly (25%) enriched uranium. Not exactly a cheap way to go.
Ridiculous recharging specs!
365 volts at 1000 amps is about ten times the available power at the average house. In order to carry this off you'd need a major upgrade of the wires going to each house, plus some interlocks so only 10% of the houses can be charging at any time.
The charging rate of 365 kilowatts, assuming a battery of 90% charging efficiency, means the battery needs 36.5 kilowatts of cooling while charging. That's one HUGE fan, or a complex liquid cooling loop.
We don't know the temperature coefficient of the cells they are considering. If their temperature coefficient goes the wrong way, you can't charge the cells in their series configuration. Just one weak cell in a string and it would tend to run away thermally and wreck at least its string, or worse.
It's sad to see students at a major university being so clueless about basic energy equations.
Breeders have been tried, to the extent of about 20 billion dollars, over the last 40 years. All have failed. It's really hard to make something that can run with the very high neutron fluxes for years and years. There are only so many different materials and alloys to choose from and they all tend to fall apart after a while with 10^38 neutrons per cm^2 per second buzzing thru them.
In addition we may have passed the point of no return re breeders-- i.e. if we had breeders right now, there isn't enough uranium left to run the current bunch of reactors and breed any usable amount of new material.
There's also the slight problem of building plutonium-burning reactors and not losing a few kilos to the bad guys.