Remember - JPEG is a compression standard. By definition it is a "lossy" comnpression. Picture quality loss remains TBD. Need to read the details.
This is a first generation UWB wireless interconnect. When the concept of UWB mas marketed around a few years ago, the claim was that it would be a low power RF communication method.
Low power at the antenna, yes, at the power supply, no.
However, the power consumed for all the signal processing in the receiver & transmitter is pretty huge. The channel bandwidth is 250MHz and uses OFDM modulation. The implication is gobs of juice to run an ADC to deal with that high bandwidth, and "must have" DSP to do all the signal processing. (OFDM requires rather fancy signal processing, which can not be implemented using a lower power analog method.)
The net result - The "low power of UWB" may be true at the antenna, but the electronics require huge amount of juice to get the job done. Consequently battery powered applications are no-go. Now you got this fancy new wireless standard and a limited use for it, with all the applications needing to be plugged into the wall.
IMHO? Poke a hole in the drywall at the floor, run the cables up thru the wall and into the display. You have to do that for the power cord anyhow, so why not? It's not like you are going to be moving the silly thing much after you install it!
UWB won't see the widespread use of WiFi or Bluetooth.
You want something that will surive multpath reflections without a lot of degeneration - that says over 100-200 MHz.
You want somthing that can sort out the signal and work with it after it has reflected off of a bunch of things and is getting received. If it is voice alone, then something like FM would work.
But then I just described a lot of the police radios already out there.
If you want it to be digital, then you need a multipath resilent modulation scheme. OFDM is where you go to do that.
Ideally, you want the capability for group communication, selective communication, and knowing the location of all the radio units at the control base station.
In a perfect world - Let's add the capability for everyone to communicate with the base station getting wiped out and no transponder/repeater dependency in a pinch. Barring those, lots of redundancy in the system, so if one gets wiped out, then another can take over.
If you are not aware of it, that has been around for for disaster communication for quite a while:
The distributed nature of the above, and all the redundancy of the multiple sources make it work, albeit not perfectly. Hm.... sounds like a terrorist network doesn't it?
Why does everyone love to point the finger at 802.11 for things it was never designed to do?
Internet methods for emergency communication in a burning building where the power plug has been pulled? Dependence on computer systems in these types of emergencies?
Please note that all this wonderous stuff happens in the next administration's budget.
NASA and space exploration is all about money, worse yet, NASA is just another beaurocratic organization of the federal government.
However, if some idiot says "lets go to the moon, so then we can get to Mars" then NASA will agree just to get the possible money to go do the job.
Why good god do we need to go land on an interim planet (um... dwarf planet? moon? gotta go see if the moon qualifies as a moon anymore, ever since Pluto had a mid-life crisis....:) so that we can then burn all the fuel and other expenditures necessary to launch from both the earth and the moon? But if that is what the president say we should do, then thats what we do, right?
Prior to that "visionary statement" nobody at NASA, JPL or any of the others looking at interplanetary exploration even mentioned going back to the moon as an interim step. Why bother? The goal is to get to Mars, not go harbor hopping to get there.
Due to current and temperature, power drive chips push the pin counts up with parallel paths on a number of connections.
Both to reduce IR drop in high current paths, and also to provide more thermal coupling to the PCB ground. (As well, special packages with good thermal conductivity under the chip package back to the PCB are in place)
That is the "Mighty" chip that went on later Fireball series drives. (See my post on the "Hitachi Combo" chip elsewhere.) Phillips fabricated the chip but a group inside Quantum specified it.
It was a POS, and the "pointy haired boss" of that group got his butt fired shortly after that happened.
I designed the power drive chip used in the Quantum Fireball... Hitachi fabricated the chip for us. Hitachi was not very cooperative in how they dealt with Quantum...
When the drive went into mass production it was failing in the lab on a regular basis. I went on record all over the company that the power drive transistors in the chip were not properly done. (then known internally as the Hitachi Combo chip, or better known internally as the "Hibachi" chip due to all the failures)
They were undersized quite a bit, because management was would not allow us to design any margin into it.
Management did not want to take the expense to get the chip properly sized & fabricated. When the field failures started to happen, and the fingerpointing started, I could point to all kinds of records showing that "you don't want to ship this product, or this is going to fail" - oh well, my CYA did not allow my boss to cut my throat, but after that he did not love me.
What happened there looks like he fried the power drive chip.
Lets see - All HDD PCB's have on it a power drive chip, that involves some rather large internal transistors for head positioning, and spindle rotation.
Durning fast seek situations, or spinning the drive up, these can dump a lot of current through them, on the order of 1A to 1.5A (talking 3 inch single platter drives here, YMMV)
That said, the power drive chip usually has some rather huge transistor arrays associated with controlling all that juice. Those power drive chips are generally done in either bipolar or DMOS silicon (DMOS, not CMOS, it is a power transistor process for large high voltage, high current transistors.)
Sometimes the current distribution across the transistor array is not balanced and you fry the transistors. (For the semiconductor folks - hot Vbe junction, without emitter resistance ballasting, to give current balalnce, leading to a a domino effect across multiple base-emitter junctions burning out)
What happens when the transistor fries, is that the chip inside the package gets hot enough that the plastic package above the chip melts, and then gassifies. Ka-boom!!! The gas blows a hole thru the top of the chip's package.
Getting on and off of a moon with a very low level of gravity is easier and cheaper than landing on something where you need to land and launch,while fighting the planets gravitational pull.
Except for the USB port modem, what you are describing is a cell phone from 5 years ago.
The modern cell phone has gone thru gobs and gobs of feature creep. The market wants more gadgetry, and if that is what sells, it will be provided.
I think the reason nobody has made (to my knowledge)the cell modem, is because they can cell (sell!) you a PCMCIA plug in and bill you for the modem service as an independent service. Verizon sucks $150 a month out of me instead of $70. You get the idea.
GaAs has all kinds of problems with defect density yield loss, this is another reason it is friendly to the RF PA (under 20 transistors) and not for the next Pentium-27 (2 zillion transistors):)
SiGe (from IBM & Jazz anyhow) is sold as BiCMOS, CMOS and SiGe Bipolars on the same chip. However BiCMOS tends to be a few generations in size behind CMOS.
SiGe (and Strained Silicon, but thats another story) does get used in specialized locations inside of big-iron (mainframe)computers, but not as a general purpose logic device.
It has been widely used in RF front ends, SerDes Data links, optical, and similar niche areas.
What you are describing is commonly known as a "software defined radio" -- Google it and you will find plenty on the subject.
Minor detail with respect to "Mr Nyquist" - If you want to do as you describe, you need an ADC that runs at (lets say RF at 2GHz to make the numbers easy) 4GHz sampling rate.
Ok, well that is do-able, although it takes ***amps*** of current and you need to really know what you are doing to get it to work.
Now a minor detail - due to the dynamic amplitude range of the signal, (on the order of 40dB to 90dB, a lot of cell phones require around 90) You need to be able to resolve a signal down in microvolt land.
Ooops!!!
I did all the math out once and it was something like a 24Bit converter running at 4GHz.
That, is not going to happen in the near future. Not with the present state of the art technology. Most ADC's that work at these frequencies are so specialized, nobody offers them as a standard off the shelf product, and I do not know of any with more than 6 bits of resoluton.
RF front ends will remain analog for the forseeable future.
GaAs is a FET, not a bipolar, although there are some GaAS bipolars as well.
Also the RF PA in a cell phone is generally GaAs although there are some exception. Never been a single transistor ever to my knowledge, the classic architecture is a three stage device.
WiFi sits in the ISM band at 2.7GHz Cellular service is in assorted bands all over 500MHz and under 3GHz Collision avoidance radar (now going into fancy new cars) is up around 40-60 GHz (not sure on the exact number, haven't designed one yet)
Any and all of that can be done with 0.18um CMOS, (excepth the radar) nothing fancier required.
For RF front end there are a lot of secialty transistors, SiGe, GaAs. IndP and others.
Magentic tape already existed at the time, no need for that.
The whole idea was "random access" not "serial access" - punch cards and mag tape you need to shuffle thru the pile of cards, or run down the tape end to end.
I met Reynold Johnson about 15 years back, (he died a while back) he ran the first design program developing this thing.
Some did not believe in it's viability back then. Somebody posted a picture of a bologna slicer on the side of the engineering prototype. The only thing in common between the original and the current methods are spinning disks. Everything else has changed in its approach.
They have been predicting the demise of the disk drive for 20 years. However the cost per byte (or mega,giga,tera,peta-byte) of magnetic storage stays ahead of the cost curve, and thus perserveres.
Encryption inherent to the drive was attempted over 10 years ago. There is no technology in the way of it, but it crashed and burned back then due to the fact that the world wants the HDD as a storage device, and the big brother stuff be kept up at a higher level in the system.
Sorta like the video telephone. Easy to do, but nobody really wanted it.
Watson does some fun and intersting stuff. They are pure research for EE and CS types. Having been part of IBM in a prior life, I spent a good amount of time working with people there.
I was one of the folks who took the "pie in the sky" ideas and tried to transfer them into (practical and profitable)reality.
Getting a job there however falls into two possible categories:
1. You walk on water in an applicable corner of academia, and want in. 2. You are a top tier Ph.D. fresh out of school and want in.
Many apply, few get accepted. You are getting paid to play with your favorite academic topic. I have seen some very capable talent there, at doing "proof of concept" work, with little (none really) attention to a final viable product outcome.
Practical products rarely come from here, but some innovative ideas do get created, and your "patent pile" gets impressive after a few years.
Few leave the place after getting a job there. Also, when IBM is bloodletting (read = layoffs) the staff here generally does not feel the pain. The reality of industry that most folks here are painfully aware of does not exist there.
Whoever comes up with a significant advance in battery technology will die a very rich person.
Li-Ion batteries have excellent amp-hour ratings for their size, but like all other batteries are still pretty limited.
Acceleration/Torque for electric cars is not a problem. High performance capabilities are there if you want them. However, you are playing battery energy against performance against distance, and all electrics, or fuel-electric hybrids have been designed to be "green" in their approach. (Any Hummer oweners want an environmentally aware vehicle?)
Right now the weakest link in many electronic systems is the energy source. A good solution there and you can be a very wealty person.
Slashdot Mirror
Remember - JPEG is a compression standard. By definition it is a "lossy" comnpression. Picture quality loss remains TBD. Need to read the details.
This is a first generation UWB wireless interconnect. When the concept of UWB mas marketed around a few years ago, the claim was that it would be a low power RF communication method.
Low power at the antenna, yes, at the power supply, no.
However, the power consumed for all the signal processing in the receiver & transmitter is pretty huge. The channel bandwidth is 250MHz and uses OFDM modulation. The implication is gobs of juice to run an ADC to deal with that high bandwidth, and "must have" DSP to do all the signal processing. (OFDM requires rather fancy signal processing, which can not be implemented using a lower power analog method.)
The net result - The "low power of UWB" may be true at the antenna, but the electronics require huge amount of juice to get the job done. Consequently battery powered applications are no-go. Now you got this fancy new wireless standard and a limited use for it, with all the applications needing to be plugged into the wall.
IMHO? Poke a hole in the drywall at the floor, run the cables up thru the wall and into the display. You have to do that for the power cord anyhow, so why not? It's not like you are going to be moving the silly thing much after you install it!
UWB won't see the widespread use of WiFi or Bluetooth.
http://standards.ieee.org/getieee802/download/802. 11-1999.pdf
please go and read the 802.11 standard. Wireless networking in the ISM bands is not what this is all about.
CB at 27MHZ with AM modulation?
Wrong frequency and wrong modulation method.
You want something that will surive multpath reflections without a lot of degeneration - that says over 100-200 MHz.
You want somthing that can sort out the signal and work with it after it has reflected off of a bunch of things and is getting received. If it is voice alone, then something like FM would work.
But then I just described a lot of the police radios already out there.
If you want it to be digital, then you need a multipath resilent modulation scheme. OFDM is where you go to do that.
http://en.wikipedia.org/wiki/COFDM
Above is a good overview.
Ideally, you want the capability for group communication, selective communication, and knowing the location of all the radio units at the control base station.
In a perfect world - Let's add the capability for everyone to communicate with the base station getting wiped out and no transponder/repeater dependency in a pinch. Barring those, lots of redundancy in the system, so if one gets wiped out, then another can take over.
If you are not aware of it, that has been around for for disaster communication for quite a while:
http://www.arrl.org/pio/emergen1.html
The distributed nature of the above, and all the redundancy of the multiple sources make it work, albeit not perfectly. Hm.... sounds like a terrorist network doesn't it?
Broadband communication using OFDM modulation is very resilient to multpath radio signals and localized fading problems.
Think lots of reflections and deadband issues.
That has nothing to do with video...
Why does everyone love to point the finger at 802.11 for things it was never designed to do?
Internet methods for emergency communication in a burning building where the power plug has been pulled? Dependence on computer systems in these types of emergencies?
I don't thinks so.
Please note that all this wonderous stuff happens in the next administration's budget.
:) so that we can then burn all the fuel and other expenditures necessary to launch from both the earth and the moon? But if that is what the president say we should do, then thats what we do, right?
NASA and space exploration is all about money, worse yet, NASA is just another beaurocratic organization of the federal government.
However, if some idiot says "lets go to the moon, so then we can get to Mars" then NASA will agree just to get the possible money to go do the job.
Why good god do we need to go land on an interim planet (um... dwarf planet? moon? gotta go see if the moon qualifies as a moon anymore, ever since Pluto had a mid-life crisis....
Prior to that "visionary statement" nobody at NASA, JPL or any of the others looking at interplanetary exploration even mentioned going back to the moon as an interim step. Why bother? The goal is to get to Mars, not go harbor hopping to get there.
Politics and science should not mix.
Getting off of my soapbox....
Due to current and temperature, power drive chips push the pin counts up with parallel paths on a number of connections.
Both to reduce IR drop in high current paths, and also to provide more thermal coupling to the PCB ground. (As well, special packages with good thermal conductivity under the chip package back to the PCB are in place)
That is the "Mighty" chip that went on later Fireball series drives. (See my post on the "Hitachi Combo" chip elsewhere.) Phillips fabricated the chip but a group inside Quantum specified it.
:)
It was a POS, and the "pointy haired boss" of that group got his butt fired shortly after that happened.
Ah, memories....
Um....
I designed the power drive chip used in the Quantum Fireball...
Hitachi fabricated the chip for us. Hitachi was not very cooperative in how they dealt with Quantum...
When the drive went into mass production it was failing in the lab on a regular basis. I went on record all over the company that the power drive transistors in the chip were not properly done. (then known internally as the Hitachi Combo chip, or better known internally as the "Hibachi" chip due to all the failures)
They were undersized quite a bit, because management was would not allow us to design any margin into it.
Management did not want to take the expense to get the chip properly sized & fabricated. When the field failures started to happen, and the fingerpointing started, I could point to all kinds of records showing that "you don't want to ship this product, or this is going to fail" - oh well, my CYA did not allow my boss to cut my throat, but after that he did not love me.
Time to get out of there....
What happened there looks like he fried the power drive chip.
Lets see - All HDD PCB's have on it a power drive chip, that involves some rather large internal transistors for head positioning, and spindle rotation.
Durning fast seek situations, or spinning the drive up, these can dump a lot of current through them, on the order of 1A to 1.5A (talking 3 inch single platter drives here, YMMV)
That said, the power drive chip usually has some rather huge transistor arrays associated with controlling all that juice. Those power drive chips are generally done in either bipolar or DMOS silicon (DMOS, not CMOS, it is a power transistor process for large high voltage, high current transistors.)
Sometimes the current distribution across the transistor array is not balanced and you fry the transistors. (For the semiconductor folks - hot Vbe junction, without emitter resistance ballasting, to give current balalnce, leading to a a domino effect across multiple base-emitter junctions burning out)
What happens when the transistor fries, is that the chip inside the package gets hot enough that the plastic package above the chip melts, and then gassifies. Ka-boom!!! The gas blows a hole thru the top of the chip's package.
Been there, done that.
IMHO ? Gravity.
Getting on and off of a moon with a very low level of gravity is easier and cheaper than landing on something where you need to land and launch,while fighting the planets gravitational pull.
Except for the USB port modem, what you are describing is a cell phone from 5 years ago.
The modern cell phone has gone thru gobs and gobs of feature creep. The market wants more gadgetry, and if that is what sells, it will be provided.
I think the reason nobody has made (to my knowledge)the cell modem, is because they can cell (sell!) you a PCMCIA plug in and bill you for the modem service as an independent service. Verizon sucks $150 a month out of me instead of $70. You get the idea.
Hm,for what you describe you need to wait until 802.16 (WiMax) deploys.
That will give you a lot of the longer distances that everyone is asking for.
People try to get 802.11 boxes to do a lot more than they were ever designed for.
Remeber the 802.11 standard(s) wereall set for (if I remember correctly) something like 30meters maxdistance.
hm, your "bedside manner" when dealing with IT, S/W and digital folks needs some refinement.
Oh, and BTW, if you think you are doing op-amps at these frequencies, then please go sit in the corner for a 10 minute time-out...
Andy:
:)
Excellent commentary - a couple of addendums -
GaAs has all kinds of problems with defect density yield loss, this is another reason it is friendly to the RF PA (under 20 transistors) and not for the next Pentium-27 (2 zillion transistors)
SiGe (from IBM & Jazz anyhow) is sold as BiCMOS, CMOS and SiGe Bipolars on the same chip. However BiCMOS tends to be a few generations in size behind CMOS.
SiGe (and Strained Silicon, but thats another story) does get used in specialized locations inside of big-iron (mainframe)computers, but not as a general purpose logic device.
It has been widely used in RF front ends, SerDes Data links, optical, and similar niche areas.
- Jerry
What you are describing is commonly known as a "software defined radio" -- Google it and you will find plenty on the subject.
Minor detail with respect to "Mr Nyquist" - If you want to do as you describe, you need an ADC that runs at (lets say RF at 2GHz to make the numbers easy) 4GHz sampling rate.
Ok, well that is do-able, although it takes ***amps*** of current and you need to really know what you are doing to get it to work.
Now a minor detail - due to the dynamic amplitude range of the signal, (on the order of 40dB to 90dB, a lot of cell phones require around 90) You need to be able to resolve a signal down in microvolt land.
Ooops!!!
I did all the math out once and it was something like a 24Bit converter running at 4GHz.
That, is not going to happen in the near future. Not with the present state of the art technology. Most ADC's that work at these frequencies are so specialized, nobody offers them as a standard off the shelf product, and I do not know of any with more than 6 bits of resoluton.
RF front ends will remain analog for the forseeable future.
Um, no...
GaAs is a FET, not a bipolar, although there are some GaAS bipolars as well.
Also the RF PA in a cell phone is generally GaAs although there are some exception. Never been a single transistor ever to my knowledge, the classic architecture is a three stage device.
Um, read the article. This is about Bipolar transistors using Flourine as a dopant.
Not CMOS, it is a different world.
As for switching speed being faster, well, yes, but at 90nm and down gate dielectric leakage plays into it in a big big ugly way...
Check your numbers -
WiFi sits in the ISM band at 2.7GHz
Cellular service is in assorted bands all over 500MHz and under 3GHz
Collision avoidance radar (now going into fancy new cars) is up around 40-60 GHz (not sure on the exact number, haven't designed one yet)
Any and all of that can be done with 0.18um CMOS, (excepth the radar) nothing fancier required.
For RF front end there are a lot of secialty transistors, SiGe, GaAs. IndP and others.
Magentic tape already existed at the time, no need for that.
The whole idea was "random access" not "serial access" - punch cards and mag tape you need to shuffle thru the pile of cards, or run down the tape end to end.
Heres a picture of the original production version:
e /storage_PH0350A.html
http://www-03.ibm.com/ibm/history/exhibits/storag
I met Reynold Johnson about 15 years back, (he died a while back) he ran the first design program developing this thing.
Some did not believe in it's viability back then. Somebody posted a picture of a bologna slicer on the side of the engineering prototype. The only thing in common between the original and the current methods are spinning disks. Everything else has changed in its approach.
They have been predicting the demise of the disk drive for 20 years. However the cost per byte (or mega,giga,tera,peta-byte) of magnetic storage stays ahead of the cost curve, and thus perserveres.
Encryption inherent to the drive was attempted over 10 years ago. There is no technology in the way of it, but it crashed and burned back then due to the fact that the world wants the HDD as a storage device, and the big brother stuff be kept up at a higher level in the system.
Sorta like the video telephone. Easy to do, but nobody really wanted it.
Watson does some fun and intersting stuff. They are pure research for EE and CS types. Having been part of IBM in a prior life, I spent a good amount of time working with people there.
I was one of the folks who took the "pie in the sky" ideas and tried to transfer them into (practical and profitable)reality.
Getting a job there however falls into two possible categories:
1. You walk on water in an applicable corner of academia, and want in.
2. You are a top tier Ph.D. fresh out of school and want in.
Many apply, few get accepted. You are getting paid to play with your favorite academic topic. I have seen some very capable talent there, at doing "proof of concept" work, with little (none really) attention to a final viable product outcome.
Practical products rarely come from here, but some innovative ideas do get created, and your "patent pile" gets impressive after a few years.
Few leave the place after getting a job there. Also, when IBM is bloodletting (read = layoffs) the staff here generally does not feel the pain. The reality of industry that most folks here are painfully aware of does not exist there.
Sounds like a nice environment, yes?
you need to go work for yourself. Anyone can start their own company.
Whoever comes up with a significant advance in battery technology will die a very rich person.
Li-Ion batteries have excellent amp-hour ratings for their size, but like all other batteries are still pretty limited.
Acceleration/Torque for electric cars is not a problem. High performance capabilities are there if you want them. However, you are playing battery energy against performance against distance, and all electrics, or fuel-electric hybrids have been designed to be "green" in their approach. (Any Hummer oweners want an environmentally aware vehicle?)
Right now the weakest link in many electronic systems is the energy source. A good solution there and you can be a very wealty person.