It's interesting that Ion Torrent has been getting so much more media exposure than Illumina, the market leader. By all accounts, from actual users, the data quality is better. Getting low error rates may involve much more sampling. This is very much akin to the bit error-rate probability in digital communication, and the number of packets you need to transmit to reduce the error to acceptable levels (but in this case, there is no forward error correction).
Illumina uses fluorescent labeling which has a much better inherent signal-to-noise ratio than sensing-pH where you have to deal with surface effects from the wells, not to mention the homopolymer issues inherent to their chemistry..
The one allure that ion-torrent has, is that it doesn't involve any optics, and thus can potentially be made much cheaper (how cheap it is depends on their error rates though). I know that there are several companies that are trying to make sequencing cheaper. There is a lot of interesting work being done in the sequencing world. The top of the line DNA sequencer from 2010 was equivalent to 60,000 top-of the line instruments from 2006. The progress is currently beating moore's law.
Sequencing has enabled all sorts of medical diagnostics and research that were previously unavailable. It was quite exhilarating to to read in Journals like Nature about people that had late stage cancer being sequenced to find out that they were misdiagnosed with the wrong cancer. Just a few years ago, it was impossible to determine what type of cancer a cancer was before it spread. Once it has metastasized, it was all a guessing game. With sequencing you can know for sure, and give you the right medicine to address that cancer.
I've always considered biology to be hundreds of years behind physics and the other "hard sciences", because they never had the tools to deal with it.The CPU power, the RAM, the hard-disk space, even the cloud infrastructure are all needed to make DNA sequencer efficient. The last instrument I worked on was a low cost DNA sequencer that could yield a sequence in one day. At the end of a run, to do the basecalling and base alignment of the data, you would need significantly more horsepower than what was on the meager instrument. The cloud allows you access to a supercomputer the the short time that you need it, so the customer is not burdened by the huge computational complexity involved.
As the cost sequence drops (and continues to drop), whole new fields of research have opened up. Bioinformatics where biology and computer science meet is a pretty hot topic. We have a deluge of data, but we don't yet have all the good algorithms necessary to unlock all the secrets we wish to solve. The Rosetta stone of the 21st century. This is the biggest complaint I hear about from biochemists.. Making sense of the data. Data leads to knowledge leads to wisdom, but data is not knowledge.
I consider DNA sequencing to be an enabler, just like the steam-engine, or the electric light. It is now possible to look deeply at things we never could, like meta-genomics. Did you know that you have more bacteria in your body then all other cells in your body put together?..And did you know that you can't grow most of them on a petri dish? We have been to mars, but we don't even know the bacteria in our own gut. Meta-genomics is a form of "shot-gun sequencing".. In the lab you understand the biology by making millions of replicas of it in the petri dish.Not all bacteria grow on a petri-dsh . With shot-gun sequencing, you sequence enough sample so you can digitally reconstruct what organisms were there to begin with. This has enabled us to [begin to] understand the biochemical messaging between soil bacteria and the roots of plants, understand the biochemistry of food digestion to generate bio-fuels more efficiently, etc.. Interesting times...
the radio amateur comment was mainly reference to people that can buy expensive equipment used on the cheap, vs those that want to sell a product.
FYI: It's possible to create a radio that can transmit a signal all over the world with very little power. The trick is to use a very narrow band source. This makes you sensitive to drift in your oscillator, and is mainly due to temperature. If you had a portable narrow band transceiver, you can use the human body as the thermal regulation source to stabilize your oscillator.
You can get cheep VCTCXO's (voltage & temperature controlled oscillators) that have a temperature stability of better than 1ppm/C. You can study the root-allan- variance curve from your oscillator data-sheet to see how well it will work.
You can get some very very good VCTCXO's and rubidium atomic clocks second hand if you're a radio amateur guy. They used to be combined in cel towers with crystal oscillators. If you look at the root-allan-variance plot, quarts oscillators have poor long-term stability and excellent short term stability. A rubidium atomic clock has poor short term stability, and excellent long term stability. They need to combine them to get the best of both worlds.
Be sure to phase-lock all your test equipment to the same 10MHz source so that all your instruments will agree.
I have a radio astronomer friend that works on the CARMA sub-millimeter array doing signal processing/FPGA work. They use a (2) 26GSPS 3bit ADC (made by Hittite) for each orthogonal polarization for each antenna. They actually get >15 effective bits because they do averaging over 10 seconds (the SNR drops by the square-root of the number of samples). I am amazed that they can recognize hundreds of molecules, and subtitles of radio spectra with a 3bit ADC. There is a lot of interesting signal processing at CARMA. You can read about all kinds of interesting tricks to get high sideband rejection for your spectrometer bins, etc.
One trick employed by oscilloscope companies [at least in the past] is to simply have a bank of ADC's, and use precision delays on the clock inputs. One ADC would see the clock pulse, the second one would see the clock that was staggered by 100ps, the next stage after that would be staggered another 100ps, and so on. You can get passive LC delay lines that are quite precise.
History is the real judge about what the great inventions are; what captured the popular imagination; what inspired a new wave of scientists. It's hard, in the near-term to know what will be relevant/ what will be funded / what will be ultimately useful. There are great ideas and smart people working on them, but not everything pans out; not ever research group is successful. The methodology is important. Some discoveries lead to break-breakthroughs, but it's largely luck what pans out.
For 5 years, I worked at a DNA sequencing company in the R&D group. The biochemistry group created a breakthrough in the chemistry called (sequencing-by-synthesis) where once instrument from 2010 was equivalent to 60,000 top-of the line DNA sequencers from 2005. Instead of reading bases one at a time in a capillary-tube, it was able to capture a whole microscope sized swath at a time, where fixed segments of DNA stuck to the bottom of the flow cell like a kelp on an ocean floor, and every cycle through the machine, using clever biochemistry techniques, removed one base-pair from all the strands at once (with a density of ~700,000 clusters per mm^2)..It ends up being about a terabase worth of data per sample. The cost drop was phenomenal, and is dropping faster than moores law.
Sequencing has enabled all sorts of medical diagnostics and research that were previously unavailable. It was quite exhilarating to to read in Journals like Nature about people that had late stage cancer being sequenced to find out that they were misdiagnosed with the wrong cancer. Just a few years ago, it was impossible to determine what type of cancer a cancer was before it spread. Once it has metastasized, it was all a guessing game. With sequencing you can know for sure, and give you the right medicine to address that cancer.
I've always considered biology to be hundreds of years behind physics and the other "hard sciences", because they never had the tools to deal with it.The CPU power, the RAM, the hard-disk space, even the cloud infrastructure are all needed to make DNA sequencer efficient. The last instrument I worked on was a low cost DNA sequencer that could yield a sequence in one day. At the end of a run, to do the basecalling and base alignment of the data, you would need significantly more horsepower than what was on the meager instrument. The cloud allows you access to a supercomputer the the short time that you need it, so the customer is not burdened by the huge computational complexity involved.
As the cost sequence drops (and continues to drop), whole new fields of research have opened up. Bioinformatics where biology and computer science meet is a pretty hot topic. We have a deluge of data, but we don't yet have all the good algorithms necessary to unlock all the secrets we wish to solve. The Rosetta stone of the 21st century. This is the biggest complaint I hear about from biochemists.. Making sense of the data. Data leads to knowledge leads to wisdom, but data is not knowledge.
I consider DNA sequencing to be an enabler, just like the steam-engine, or the electric light. It is now possible to look deeply at things we never could, like meta-genomics. Did you know that you have more bacteria in your body then all other cells in your body put together?..And did you know that you can't grow most of them on a petri dish? We have been to mars, but we don't even know the bacteria in our own gut. Meta-genomics is a form of "shot-gun sequencing".. In the lab you understand the biology by making millions of replicas of it in the petri dish.Not all bacteria grow on a petri-dsh . With shot-gun sequencing, you sequence enough sample so you can digitally reconstruct what organisms were there to begin with. This has enabled us to [begin to] understand the biochemical messaging between soil bacteria and the roots of plants, understand the biochemistry of food digestion to generate bio-fuels more efficiently, etc.. Interesting times...
..I too own about half a dozen FPGA boards, and do agree that there is a learning curve for the Zed board, but I still think it's an amazing platform. Altera has announced that they have a version of their FPGA that also has a dual core A9 processor (that is just about to come out?) . Xilinx made a lot of changes all at once to support their new integrated CPU/FPGA environment. Altera mentioned that they're taking a more conservative approach. It's essentially one die, but two separate partitions, one for the FPGA, and one for the two 800MHz arm cores. You can use the ARM tools completely separately than the FPGA tools. Traditionally, the Altera tools are easier to use & less buggy than Xilinx. You also get ModelSim (good hardware simulation software) for free, which you do not get with Xilinx. Also, Altera (Quartus) tools work for all previous versions of their FPGA, so you won't have compatibility issues supporting your old designs, won't need to support an old development machine, etc. If you're like me my home projects, given family constraints, are on a much longer development cycle than my professional ones. At the very lest Quartus is a great tool to learn FPGA programming.
Everything I've ever worked on as a professional had a microprocessor and an FPGA. You can pick up a "Zed Board" with a dual-core Arm Cortex-A9 and a 85K luts worth of FPGA. You could learn fpga programming in addition to learning about microcontrollers. You can run linux on one core, or run "bare-metal" or Free-rtos in the other for all of your hard real-time needs. You have a very wide selection of things for you can try. The FPGA is a true parallel processor, and is great for processing multi-sensor inputs. A microcontroller time-slices between all of the tasks it needs to take care of. An FPGA can essentially be a hardware dedicated task.
"We live in an age of apologies. Apologies, false or true, are expected from the descendants of empire builders, slave owners, persecutors of heretics and from men who, in our eyes, just got it all wrong. So, with the age of 85 coming up shortly, I want to make an apology. It appears I must apologize for being male, white and European." --Alec Guinness
There are a lot of great [old] games out there. I have"Windows XP mode" virtual PC, but found that my video card does not have drivers for XP, so I can't play them.
Has anyone here had a similar experience? I personally have an Nvidia 690 card, but now I find that I must maintain a slew of older PC's so I can get my programs/hardware/"old" games to run.
If you do get a new video card, make sure that it has drivers for XP [or older] as well so that you can play your old games in a virtual machine.
I've read that Boeing was intent on outsourcing as much of the design as possible, and even had a catch-phrase: "the product is the process". I've read that in order to clean up the design, they needed to bring in more Boeing engineers. I wonder what extent this is true, and how much of their plane was designed by third parties?
With engineering, it's always hardest to get that last fraction of a percent nailed and verified; an exponential more effort for that last few percentage points.
I have a friend that designed electronics for aircraft, and understand that the cables, connectors, and electronics are subject to a very high standard for robustness, I find it rather shocking to hear that there was a fire. I wonder what kind of technical over-sight did they do with their contractors and their own engineers. I've always considered that Boeing was over-the-top with this sort of thing.
This is a cautionary tale for anyone that wants to outsource. How do you guarantee that your subcontractor has done a sufficient job? That subcontractor does not necessarily want to let you in on all of the engineering details so they can avoid being designed out. Assuming that they were given everything, I wonder if they had their own engineers review them.
I understand that if you print out the number of individual components in a modern fighter aircraft as a function of time, it would be linear on a log scale, meaning that the number of all components have grown exponentially. As subsystems become more complicated, they are increasingly designed by small teams of specialists. Outsourcing has it's merits. It's hard to be a generalist.
Most text books are very simple relative to real-world code, and a great many books out there are average or poor. A large percentage of the code that you find on the internet is also bad, but it's often better than what you can find in textbooks. I've been a programmer for more than 10 years. Most of the code I have been exposed to has been pretty elegantly written. Well written code is probably more common than poorly written code for software engineers in Silicon Valley, though my vantage point is rather limited. Employers don't pay through the nose for Spaghetti code.. Silicon Valley is, by and large, a meritocracy. Well written code begets higher titles, salaries, other ventures with the like-minded.. Well written code is easily recognizable by your colleagues.
I've always held that 90% of anything is bad with respect to books, movies, music etc. After 10 years, the good authors tend to stand out.
Nucear weapons, and technology are so destructive, that I don't want this information to be widely available. The mathematical and technical details (Jesse Beams, etc) behind the ultracentrifuges used in the Manhattan project were later made available to Abdul Qadeer Khan, who went on to make nuclear weapons in Pakistan. Should private companies like Urenco exist, that enrich uranium? What kind of security clearance did Abdul Qadeer Khan have when he worked at the Physics Dynamics Research Laboratory in Amsterdam? How hard is it to secure a national lab. It seems to work here, but it failed [in hind-site?] in Amsterdam..
It is my belief that the status quo should remain. The system works, though perhaps they should hire better security contractors to reign in recalcitrant nuns.
Could someone tell me why the Cern machine can't be upgraded to smash muons rather than protons? Are they too short lived (2.2us?) I just wonder if it's possible to use the existing infrastructure and not spend another 20+ billion dollars. In an era of shrinking research budgets, It's worthy to be smart about how we allocate our money.
What are the underlying problems?
rights aren't ‘rights’ if someone can take ‘em away; they’re privileges. That’s all we’ve ever had in this country: a bill of temporary privileges. And if you read the news, even badly, you know that every year the list gets shorter and shorter.
“
— George Carlin, “You Have No Rights” (via kristinovich)\
If bandwidth is an issue, use a higher frequency carrier.,. You can replace the feed-horn and RF section with a higher frequency antenna (like a log periodic antenna). You can also build smaller antennas, in very dry regions of the earth. You can get a lot of gain out of a small antenna at high frequencies...mm-wave astronomy has some a long way. CARMA (the combined array for research in millimeter wave astronomy) works up to 250GHz. I'm sure there is plenty of unlicensed bands in the ultra-high frequency regions.
Commercially, you can get an ADC with 4-bits of resolution (3 effective) ADC that's good out to 26GHz. These kinds of advances in ADC technology greatly simplify the RF requirements by eliminating a whole bank of local oscillators, mixers, amplifiers, splitters, and the like. If you read the CARMA website, you can see that they've greatly reduced the amount of electronics needed to capture all the bandwidth for both polarizations from each of their 26 antennas.
(BTW: CARMA effectively samples 4 bits at 20GHz, but they integrate from between 1 and 20 seconds for an astronomical target; this is one thing that allows them to get such stunning images with such a low effective number of bits )
It's way cheaper to invest in new antenna technology than to create such an insanely expensive alternative. The only advantage I could see would be for ultra quiet radio astronomy, where you have little thermal noise from the planet, and no radio-frequency-interference or atmospheric attenuation, which on earth is mostly inhibited by water-vapor at very high frequencies.
Slashdot Mirror
Illumina uses fluorescent labeling which has a much better inherent signal-to-noise ratio than sensing-pH where you have to deal with surface effects from the wells, not to mention the homopolymer issues inherent to their chemistry..
The one allure that ion-torrent has, is that it doesn't involve any optics, and thus can potentially be made much cheaper (how cheap it is depends on their error rates though). I know that there are several companies that are trying to make sequencing cheaper. There is a lot of interesting work being done in the sequencing world. The top of the line DNA sequencer from 2010 was equivalent to 60,000 top-of the line instruments from 2006. The progress is currently beating moore's law.
I've always considered biology to be hundreds of years behind physics and the other "hard sciences", because they never had the tools to deal with it.The CPU power, the RAM, the hard-disk space, even the cloud infrastructure are all needed to make DNA sequencer efficient. The last instrument I worked on was a low cost DNA sequencer that could yield a sequence in one day. At the end of a run, to do the basecalling and base alignment of the data, you would need significantly more horsepower than what was on the meager instrument. The cloud allows you access to a supercomputer the the short time that you need it, so the customer is not burdened by the huge computational complexity involved.
As the cost sequence drops (and continues to drop), whole new fields of research have opened up. Bioinformatics where biology and computer science meet is a pretty hot topic. We have a deluge of data, but we don't yet have all the good algorithms necessary to unlock all the secrets we wish to solve. The Rosetta stone of the 21st century. This is the biggest complaint I hear about from biochemists.. Making sense of the data. Data leads to knowledge leads to wisdom, but data is not knowledge.
I consider DNA sequencing to be an enabler, just like the steam-engine, or the electric light. It is now possible to look deeply at things we never could, like meta-genomics. Did you know that you have more bacteria in your body then all other cells in your body put together? ..And did you know that you can't grow most of them on a petri dish? We have been to mars, but we don't even know the bacteria in our own gut. Meta-genomics is a form of "shot-gun sequencing" .. In the lab you understand the biology by making millions of replicas of it in the petri dish.Not all bacteria grow on a petri-dsh . With shot-gun sequencing, you sequence enough sample so you can digitally reconstruct what organisms were there to begin with. This has enabled us to [begin to] understand the biochemical messaging between soil bacteria and the roots of plants, understand the biochemistry of food digestion to generate bio-fuels more efficiently, etc .. Interesting times...
..Could I interest you in a pair? I couldn't help it I've been a Zappa fan too long to let such an opportunity pass.
the radio amateur comment was mainly reference to people that can buy expensive equipment used on the cheap, vs those that want to sell a product. FYI: It's possible to create a radio that can transmit a signal all over the world with very little power. The trick is to use a very narrow band source. This makes you sensitive to drift in your oscillator, and is mainly due to temperature. If you had a portable narrow band transceiver, you can use the human body as the thermal regulation source to stabilize your oscillator.
You can get some very very good VCTCXO's and rubidium atomic clocks second hand if you're a radio amateur guy. They used to be combined in cel towers with crystal oscillators. If you look at the root-allan-variance plot, quarts oscillators have poor long-term stability and excellent short term stability. A rubidium atomic clock has poor short term stability, and excellent long term stability. They need to combine them to get the best of both worlds.
Be sure to phase-lock all your test equipment to the same 10MHz source so that all your instruments will agree.
I have a radio astronomer friend that works on the CARMA sub-millimeter array doing signal processing/FPGA work. They use a (2) 26GSPS 3bit ADC (made by Hittite) for each orthogonal polarization for each antenna. They actually get >15 effective bits because they do averaging over 10 seconds (the SNR drops by the square-root of the number of samples). I am amazed that they can recognize hundreds of molecules, and subtitles of radio spectra with a 3bit ADC. There is a lot of interesting signal processing at CARMA. You can read about all kinds of interesting tricks to get high sideband rejection for your spectrometer bins, etc. One trick employed by oscilloscope companies [at least in the past] is to simply have a bank of ADC's, and use precision delays on the clock inputs. One ADC would see the clock pulse, the second one would see the clock that was staggered by 100ps, the next stage after that would be staggered another 100ps, and so on. You can get passive LC delay lines that are quite precise.
For 5 years, I worked at a DNA sequencing company in the R&D group. The biochemistry group created a breakthrough in the chemistry called (sequencing-by-synthesis) where once instrument from 2010 was equivalent to 60,000 top-of the line DNA sequencers from 2005. Instead of reading bases one at a time in a capillary-tube, it was able to capture a whole microscope sized swath at a time, where fixed segments of DNA stuck to the bottom of the flow cell like a kelp on an ocean floor, and every cycle through the machine, using clever biochemistry techniques, removed one base-pair from all the strands at once (with a density of ~700,000 clusters per mm^2) ..It ends up being about a terabase worth of data per sample. The cost drop was phenomenal, and is dropping faster than moores law.
Sequencing has enabled all sorts of medical diagnostics and research that were previously unavailable. It was quite exhilarating to to read in Journals like Nature about people that had late stage cancer being sequenced to find out that they were misdiagnosed with the wrong cancer. Just a few years ago, it was impossible to determine what type of cancer a cancer was before it spread. Once it has metastasized, it was all a guessing game. With sequencing you can know for sure, and give you the right medicine to address that cancer.
I've always considered biology to be hundreds of years behind physics and the other "hard sciences", because they never had the tools to deal with it.The CPU power, the RAM, the hard-disk space, even the cloud infrastructure are all needed to make DNA sequencer efficient. The last instrument I worked on was a low cost DNA sequencer that could yield a sequence in one day. At the end of a run, to do the basecalling and base alignment of the data, you would need significantly more horsepower than what was on the meager instrument. The cloud allows you access to a supercomputer the the short time that you need it, so the customer is not burdened by the huge computational complexity involved.
As the cost sequence drops (and continues to drop), whole new fields of research have opened up. Bioinformatics where biology and computer science meet is a pretty hot topic. We have a deluge of data, but we don't yet have all the good algorithms necessary to unlock all the secrets we wish to solve. The Rosetta stone of the 21st century. This is the biggest complaint I hear about from biochemists.. Making sense of the data. Data leads to knowledge leads to wisdom, but data is not knowledge.
I consider DNA sequencing to be an enabler, just like the steam-engine, or the electric light. It is now possible to look deeply at things we never could, like meta-genomics. Did you know that you have more bacteria in your body then all other cells in your body put together? ..And did you know that you can't grow most of them on a petri dish? We have been to mars, but we don't even know the bacteria in our own gut. Meta-genomics is a form of "shot-gun sequencing" .. In the lab you understand the biology by making millions of replicas of it in the petri dish.Not all bacteria grow on a petri-dsh . With shot-gun sequencing, you sequence enough sample so you can digitally reconstruct what organisms were there to begin with. This has enabled us to [begin to] understand the biochemical messaging between soil bacteria and the roots of plants, understand the biochemistry of food digestion to generate bio-fuels more efficiently, etc .. Interesting times...
..I too own about half a dozen FPGA boards, and do agree that there is a learning curve for the Zed board, but I still think it's an amazing platform. Altera has announced that they have a version of their FPGA that also has a dual core A9 processor (that is just about to come out?) . Xilinx made a lot of changes all at once to support their new integrated CPU/FPGA environment. Altera mentioned that they're taking a more conservative approach. It's essentially one die, but two separate partitions, one for the FPGA, and one for the two 800MHz arm cores. You can use the ARM tools completely separately than the FPGA tools. Traditionally, the Altera tools are easier to use & less buggy than Xilinx. You also get ModelSim (good hardware simulation software) for free, which you do not get with Xilinx. Also, Altera (Quartus) tools work for all previous versions of their FPGA, so you won't have compatibility issues supporting your old designs, won't need to support an old development machine, etc. If you're like me my home projects, given family constraints, are on a much longer development cycle than my professional ones. At the very lest Quartus is a great tool to learn FPGA programming.
Everything I've ever worked on as a professional had a microprocessor and an FPGA. You can pick up a "Zed Board" with a dual-core Arm Cortex-A9 and a 85K luts worth of FPGA. You could learn fpga programming in addition to learning about microcontrollers. You can run linux on one core, or run "bare-metal" or Free-rtos in the other for all of your hard real-time needs. You have a very wide selection of things for you can try. The FPGA is a true parallel processor, and is great for processing multi-sensor inputs. A microcontroller time-slices between all of the tasks it needs to take care of. An FPGA can essentially be a hardware dedicated task.
"We live in an age of apologies. Apologies, false or true, are expected from the descendants of empire builders, slave owners, persecutors of heretics and from men who, in our eyes, just got it all wrong. So, with the age of 85 coming up shortly, I want to make an apology. It appears I must apologize for being male, white and European." --Alec Guinness
If you do get a new video card, make sure that it has drivers for XP [or older] as well so that you can play your old games in a virtual machine.
I've read that Boeing was intent on outsourcing as much of the design as possible, and even had a catch-phrase: "the product is the process". I've read that in order to clean up the design, they needed to bring in more Boeing engineers. I wonder what extent this is true, and how much of their plane was designed by third parties? With engineering, it's always hardest to get that last fraction of a percent nailed and verified; an exponential more effort for that last few percentage points. I have a friend that designed electronics for aircraft, and understand that the cables, connectors, and electronics are subject to a very high standard for robustness, I find it rather shocking to hear that there was a fire. I wonder what kind of technical over-sight did they do with their contractors and their own engineers. I've always considered that Boeing was over-the-top with this sort of thing. This is a cautionary tale for anyone that wants to outsource. How do you guarantee that your subcontractor has done a sufficient job? That subcontractor does not necessarily want to let you in on all of the engineering details so they can avoid being designed out. Assuming that they were given everything, I wonder if they had their own engineers review them. I understand that if you print out the number of individual components in a modern fighter aircraft as a function of time, it would be linear on a log scale, meaning that the number of all components have grown exponentially. As subsystems become more complicated, they are increasingly designed by small teams of specialists. Outsourcing has it's merits. It's hard to be a generalist.
Le mieux est l'ennemi du bien. (The best is the enemy of the good.) -- Voltaire
I've always held that 90% of anything is bad with respect to books, movies, music etc. After 10 years, the good authors tend to stand out.
Imagine trying to look inconspicuous with -that- thing trailing behind you -wry grin-. ,..Just about as bad as being backed by a regiment of bagpipers.
What's, obvious? ..and why?
It is my belief that the status quo should remain. The system works, though perhaps they should hire better security contractors to reign in recalcitrant nuns.
Could someone tell me why the Cern machine can't be upgraded to smash muons rather than protons? Are they too short lived (2.2us?) I just wonder if it's possible to use the existing infrastructure and not spend another 20+ billion dollars. In an era of shrinking research budgets, It's worthy to be smart about how we allocate our money. What are the underlying problems?
rights aren't ‘rights’ if someone can take ‘em away; they’re privileges. That’s all we’ve ever had in this country: a bill of temporary privileges. And if you read the news, even badly, you know that every year the list gets shorter and shorter. “ — George Carlin, “You Have No Rights” (via kristinovich)\
Commercially, you can get an ADC with 4-bits of resolution (3 effective) ADC that's good out to 26GHz. These kinds of advances in ADC technology greatly simplify the RF requirements by eliminating a whole bank of local oscillators, mixers, amplifiers, splitters, and the like. If you read the CARMA website, you can see that they've greatly reduced the amount of electronics needed to capture all the bandwidth for both polarizations from each of their 26 antennas.
(BTW: CARMA effectively samples 4 bits at 20GHz, but they integrate from between 1 and 20 seconds for an astronomical target; this is one thing that allows them to get such stunning images with such a low effective number of bits )
It's way cheaper to invest in new antenna technology than to create such an insanely expensive alternative. The only advantage I could see would be for ultra quiet radio astronomy, where you have little thermal noise from the planet, and no radio-frequency-interference or atmospheric attenuation, which on earth is mostly inhibited by water-vapor at very high frequencies.