I maintain safe distance when following and leave safe distance after passing, never pass on the right, and always signal turns and lane changes at least four seconds beforehand... although I'd skip that to evade an accident ahead =). If someone is tailgating me, I change lanes, and drop as much speed as I can to let them pass. I take the defensive driving class every three years just to improve my odds--
...
And I will cheerfully slow to *half* the speed limit to give a cyclist the clearance they need to be comfortable.
Jesus, how do you ever get anywhere at all with all of those precautions?:)
Magnetic polarization reversal had long been theorized to be the cause of the variations seen in the magnetism of the ocean floor. Eventually someone (I guess the guy in the GP's post) came up with a computer model that predicted this fairly accurately. I used to work for an oceanographic research group and we did some magnetic surveys of the ocean floor on the Juan de Fuca Ridge. Neat stuff.
The Fiero was marketed as a commuter car because Pontiac wasn't allowed by Chevy to market it as a sports car (at GM that honor goes exclusively to the Corvette).
What about the Camaro and Firebird/Trans Am? Those both have to be considered sports cars. But otherwise I agree with you, Forbes should stick to financial press.
Without these cards, they have only store-by-store info.
Why does the store need for you to produce a card (or any sort of identifier) to compile the specifics of your purchases? It would seem to me that the cash register information easily identifies the beginning and end of a specific customer's purchases.
begin transaction
Scan item 1
Scan item 2 ...
Scan item N
accept payment
end of transaction
If all the store was interested in was the list of items that each customer purchased, it seems like they already have that information. It would therefore seem that the 'discount cards' are likely used to associate a specific customer with that list of items.
What's harder - developing some hardware control mechanism or using off-the-shelf parts and a thick metal box with a little shielded exit port for an IDE cable?
The latter.:)
Seriously though, people often underestimate the difficulty in building good low frequency magnetic shields. Electric fields are pretty easy to shield, all you need is a good conductor. For a magnetic shield, you need a material with a high permeability. There are several such materials, but they're far harder to work with. For instance, the permeability of these materials can change substantially with temperature. They're also sensitive to shock.
It's not an unsolveable problem, but not a trivial one either.
Thanks for the replies so far. I left some of the design details out of the original submission to keep the discussion focused, but I guess that's hard to do.
Most people have suggested USB/FW/FC hard drives instead of ethernet. The 2 problems I can see with this solution are:
1) The 20 feet distance I mentioned in the original submission is a minimum value. It may have to be further away than that. I'll look into what the real distance limitations of these technologies are. Obviously FibreChannel still stands as a possible option over longer distances.
2) One of the reasons I chose ethernet was that it gives me the option of putting something else besides a dumb storage device at the end of the wire (i.e. laptop). The laptop can do double-duty as a storage device and a display so that an operator can look at the data in real time. I guess it's possible to have the same functionality if the sensor is a Firewire device, but not that many laptops come with Firewire ports yet. Ethernet is far more universal.
Granted, ethernet's not the lowest power transport mechanism out there, but it suits our needs pretty well. ANOTHER detail I left out is that this sensor will eventually be attached to a generic data acquisition system made by another company. We haven't settled on an interface mechanism yet but I'd bet they'll prefer ethernet to USB/FW.
It looks like I'll be rolling my own. I'll probably use a single board computer running embedded linux and a 2.5" hard drive. Other than some packaging there's not much design to do, other than maybe writing some software if I want to do something more sophisticated than acting as an FTP server.
Hmm..this thing is interesting. Looks like it might be too "interactive" for unattended use but the idea is close. I'll take a look at it some more and see if I can get some more detailed info.
have a look at the latest iSCSI stuff. It may be what you need. Sure you'll have to write a driver but at least you'll have stock parts available.
It's on my list of roll-my-own candidates.
Also why can't you put the HD in a faraday cage? Or doesn't that block magnetic fields?
Most Faraday cages are designed to attenuate electric fields only. You can build a magnetic field shield out of high permeability materials such as Mu-metal, but that's a PAIN IN THE ASS. Especially when you want to start passing ribbon cable through it.
Good question. This is the most off-the-shelf and flexible solution there is, but it's not the most asthetically pleasing solution. During development this is exactly what we'll use, but eventually this has to be something that's relatively compact and can fit into a weatherproof box. And besides, the laptop consumes a lot more power than we'd like. We can just add more batteries as you suggest, but that gets to be a pain when the batteries get REAL big.
But the problem didn't require a NAS. Your problem was "too much interference so need the hard drive 20 feet away". You could solve this problem with Fibre Channel.
Several people have suggested Fibre Channel drives. We've discussed the pros and cons of using a fiber link but I kinda lost that battle for a few reasons. However, I didn't have a specific solution in mind when we were discussing it. I'll take a look at the options and see what it would take to implement on the data collection side of things.
I'm not sure I understand the issues here. If your sensor gadget has an ethernet port, can't you just run a TP cable to some place where there is power? Or are you looking to have a complete, self-contained system?
Complete, self-contained system that just happens to be in two parts (data collection lump and data storage lump). This system is used for remote, unattended outdoor data collection (remote being the key word). AC power isn't an option at all since even it was available, it creates so much interference as to make the sensor useless.
If that's the case, you've made the usual "Ask Slashdot" mistake of phrasing the question in terms of specific and non-essential technology.
Perhaps, but rather than going into a long diatribe describing every design decision that led us to "ethernet hard drive", I figured it would be best to keep it simple.
Instead of an Ethernet port, your device should have an old-fashioned RS-232 port.
RS-232 isn't fast enough by a long shot or else that's exactly what I'd use. I expect the data rate to be in the ballpark of 1Mbps.
Of course, you'd need to add USB capabilities to your sensor(s) and you wouldn't really FTP to the drive (I don't believe), but this would be a fairly cheap and modular way to solve your problem.
Several people have suggested USB or FW. I have concerns regarding the cable length issue though, which is the primary reason I've been looking for ethernet solutions thus far. I must admit I haven't thoroughly researched how long a USB/FW link can be, but I also chose ethernet for a couple of other reasons I didn't spell out in the original submission.
When I mentioned 20 ft as the distance from the sensor to the storage device, that's really a minimum. I think there's a chance it will have to be further since we're increasing the sensitivity of the sensors right now. The other attractive thing about ethernet is that rather than plugging the sensor into this storage device, you can plug it into a laptop, which means I can write some software to watch the data in real time. I don't think that's an option with USB, but I'm not all that familiar with USB protocols.
Yeah, I stumbled across Linuxdevices.com when I was searching for SBC's. I may very well choose one of the boards mentioned there if I end up building my own solution.
The slowest Seagate 20GB drive i found consumes 24 watts
Toshiba's 1.8" 20GB drive consumes 1.4W during reading or writing. I imagine the 2.5" drives aren't much worse.
If it's a power availability problem - spacecraft maybe?... you're not looking for a COTS solution... there would be none. If it's a power DENSITY problem... consolidate all the drive needs in a server away from the environment you're measuring.
It's not for spacecraft, thankfully. That would open up an entirely new set of problems. This is just for outdoor, unattended data collection. We can bring all the power we need, but up until now we've been lugging around 12V, 100Ah marine batteries. That gets old real quick when you're lugging those through the mountains.:)
Incidentally, just so you don't embarrass yourself in front of your boss... wireless, at least in the most-commonly used sense, uses both electric and magnetic fields, most commonly at 2.4 or 5 GHz.
Yeah, I realize that. These sensors are strictly low frequency (.01Hz-50kHz). RF isn't an issue, but a transmitter will have some unknown broadband interference we may pick up even at these low frequencies. That's why we're trying to avoid wireless for now.
Frankly, I think I'd take the DC outputs of the sensors and route them as the center conductors of good-quality coax, and bring them all to a bank of A/D converters somewhere away from the environment being measured.... can you give some kind of detail about the environment?
It's funny you should suggest that, because that's what we do at the present. The problem is that there are 8 analog outputs from the sensor, and stringing 8 coaxes back to a data acquisition node is a pain in the ass. We're moving the A/D's into the sensor housing just to avoid this. Right now we're the only people who use this sensor so it's not a big deal, but we've also got people who want to buy some of them so we're trying to make it as clean as possible.
Yeah, that's basically my plan if I have to roll my own. I'll probably use a PC-104 CPU board with an ethernet and IDE interface already on it. I've found a couple of those that come in around 2-3 watts. That will suffice, but a lot of that power (well, some of it anyway) is wasted by things like VGA and USB interfaces that I'd rather just not have. Building a custom board starting from the controller is easy enough, but far more costly than buying something off the shelf.
Actually the Ipod uses a non-cheap, not-so-mass-produced 1.8" hard drive
...
How you're gonna hook it up a fastethernet without a PC in between is another story:-)
I have an iPod myself and that was the first thing that came to mind. But, like you say, the "PC" inbetween is the rub.
I thought Firewire and USB2 cables were much more limited in their maximum length. I'll look into that option some more. I don't relish the idea of building a Firewire interface into the sensor but I guess it's an option.
How do you propose to do wireless without EM interference? Sound waves? Gravity waves?
Our sensors work exclusively in the low frequency regime (0.01Hz-50kHz) so the RF interference from a transmitter isn't necessarily a killer. There will be some issues to overcome, for sure, which is why we're putting that part off until later.
A microdrive or solid state disk are good options for the storage device, but there still needs to be an intelligent device to act as an interface between the ethernet line and the disk. That's what I'm having trouble finding.
One of the proposed laws gives the consumer $10 for successfully sueing a spammer. Gimme a break, who's got the time to go to court for $10? Another of the proposed laws awarded the spamee $5000 (or was it $2000?) if they had registered on the no-spam list but gets spammed anyway. That would certainly be more of a deterrant, but it doesn't address the problem of finding the spammer to begin with.
While it's good to see someone trying to do something about the problem, this ain't it.
More likely, they'll break in, trash the joint, and steal the cool looking robot thingy. It'll be found stripped for parts in an alley in Mexico (or sold into prostitution, one of othe two).
Slashdot Mirror
And I will cheerfully slow to *half* the speed limit to give a cyclist the clearance they need to be comfortable.
Jesus, how do you ever get anywhere at all with all of those precautions? :)
Magnetic polarization reversal had long been theorized to be the cause of the variations seen in the magnetism of the ocean floor. Eventually someone (I guess the guy in the GP's post) came up with a computer model that predicted this fairly accurately. I used to work for an oceanographic research group and we did some magnetic surveys of the ocean floor on the Juan de Fuca Ridge. Neat stuff.
...that "smell-ya-later" would replace "goodbye".
Pffft. All you had to do was open the refrigerator/freezer door. Duh!
What about the Camaro and Firebird/Trans Am? Those both have to be considered sports cars. But otherwise I agree with you, Forbes should stick to financial press.
Why does the store need for you to produce a card (or any sort of identifier) to compile the specifics of your purchases? It would seem to me that the cash register information easily identifies the beginning and end of a specific customer's purchases.
begin transaction
...
Scan item 1
Scan item 2
Scan item N
accept payment
end of transaction
If all the store was interested in was the list of items that each customer purchased, it seems like they already have that information. It would therefore seem that the 'discount cards' are likely used to associate a specific customer with that list of items.
The latter. :)
Seriously though, people often underestimate the difficulty in building good low frequency magnetic shields. Electric fields are pretty easy to shield, all you need is a good conductor. For a magnetic shield, you need a material with a high permeability. There are several such materials, but they're far harder to work with. For instance, the permeability of these materials can change substantially with temperature. They're also sensitive to shock.
It's not an unsolveable problem, but not a trivial one either.
Most people have suggested USB/FW/FC hard drives instead of ethernet. The 2 problems I can see with this solution are:
1) The 20 feet distance I mentioned in the original submission is a minimum value. It may have to be further away than that. I'll look into what the real distance limitations of these technologies are. Obviously FibreChannel still stands as a possible option over longer distances.
2) One of the reasons I chose ethernet was that it gives me the option of putting something else besides a dumb storage device at the end of the wire (i.e. laptop). The laptop can do double-duty as a storage device and a display so that an operator can look at the data in real time. I guess it's possible to have the same functionality if the sensor is a Firewire device, but not that many laptops come with Firewire ports yet. Ethernet is far more universal.
Granted, ethernet's not the lowest power transport mechanism out there, but it suits our needs pretty well. ANOTHER detail I left out is that this sensor will eventually be attached to a generic data acquisition system made by another company. We haven't settled on an interface mechanism yet but I'd bet they'll prefer ethernet to USB/FW.
It looks like I'll be rolling my own. I'll probably use a single board computer running embedded linux and a 2.5" hard drive. Other than some packaging there's not much design to do, other than maybe writing some software if I want to do something more sophisticated than acting as an FTP server.
Hmm..this thing is interesting. Looks like it might be too "interactive" for unattended use but the idea is close. I'll take a look at it some more and see if I can get some more detailed info.
It's on my list of roll-my-own candidates.
Also why can't you put the HD in a faraday cage? Or doesn't that block magnetic fields?
Most Faraday cages are designed to attenuate electric fields only. You can build a magnetic field shield out of high permeability materials such as Mu-metal, but that's a PAIN IN THE ASS. Especially when you want to start passing ribbon cable through it.
Good question. This is the most off-the-shelf and flexible solution there is, but it's not the most asthetically pleasing solution. During development this is exactly what we'll use, but eventually this has to be something that's relatively compact and can fit into a weatherproof box. And besides, the laptop consumes a lot more power than we'd like. We can just add more batteries as you suggest, but that gets to be a pain when the batteries get REAL big.
Several people have suggested Fibre Channel drives. We've discussed the pros and cons of using a fiber link but I kinda lost that battle for a few reasons. However, I didn't have a specific solution in mind when we were discussing it. I'll take a look at the options and see what it would take to implement on the data collection side of things.
Complete, self-contained system that just happens to be in two parts (data collection lump and data storage lump). This system is used for remote, unattended outdoor data collection (remote being the key word). AC power isn't an option at all since even it was available, it creates so much interference as to make the sensor useless.
If that's the case, you've made the usual "Ask Slashdot" mistake of phrasing the question in terms of specific and non-essential technology.
Perhaps, but rather than going into a long diatribe describing every design decision that led us to "ethernet hard drive", I figured it would be best to keep it simple.
Instead of an Ethernet port, your device should have an old-fashioned RS-232 port.
RS-232 isn't fast enough by a long shot or else that's exactly what I'd use. I expect the data rate to be in the ballpark of 1Mbps.
Several people have suggested USB or FW. I have concerns regarding the cable length issue though, which is the primary reason I've been looking for ethernet solutions thus far. I must admit I haven't thoroughly researched how long a USB/FW link can be, but I also chose ethernet for a couple of other reasons I didn't spell out in the original submission.
When I mentioned 20 ft as the distance from the sensor to the storage device, that's really a minimum. I think there's a chance it will have to be further since we're increasing the sensitivity of the sensors right now. The other attractive thing about ethernet is that rather than plugging the sensor into this storage device, you can plug it into a laptop, which means I can write some software to watch the data in real time. I don't think that's an option with USB, but I'm not all that familiar with USB protocols.
Yeah, I stumbled across Linuxdevices.com when I was searching for SBC's. I may very well choose one of the boards mentioned there if I end up building my own solution.
Toshiba's 1.8" 20GB drive consumes 1.4W during reading or writing. I imagine the 2.5" drives aren't much worse.
If it's a power availability problem - spacecraft maybe?... you're not looking for a COTS solution... there would be none. If it's a power DENSITY problem... consolidate all the drive needs in a server away from the environment you're measuring.
It's not for spacecraft, thankfully. That would open up an entirely new set of problems. This is just for outdoor, unattended data collection. We can bring all the power we need, but up until now we've been lugging around 12V, 100Ah marine batteries. That gets old real quick when you're lugging those through the mountains. :)
Incidentally, just so you don't embarrass yourself in front of your boss... wireless, at least in the most-commonly used sense, uses both electric and magnetic fields, most commonly at 2.4 or 5 GHz.
Yeah, I realize that. These sensors are strictly low frequency (.01Hz-50kHz). RF isn't an issue, but a transmitter will have some unknown broadband interference we may pick up even at these low frequencies. That's why we're trying to avoid wireless for now.
Frankly, I think I'd take the DC outputs of the sensors and route them as the center conductors of good-quality coax, and bring them all to a bank of A/D converters somewhere away from the environment being measured.... can you give some kind of detail about the environment?
It's funny you should suggest that, because that's what we do at the present. The problem is that there are 8 analog outputs from the sensor, and stringing 8 coaxes back to a data acquisition node is a pain in the ass. We're moving the A/D's into the sensor housing just to avoid this. Right now we're the only people who use this sensor so it's not a big deal, but we've also got people who want to buy some of them so we're trying to make it as clean as possible.
Yeah, that's basically my plan if I have to roll my own. I'll probably use a PC-104 CPU board with an ethernet and IDE interface already on it. I've found a couple of those that come in around 2-3 watts. That will suffice, but a lot of that power (well, some of it anyway) is wasted by things like VGA and USB interfaces that I'd rather just not have. Building a custom board starting from the controller is easy enough, but far more costly than buying something off the shelf.
...
How you're gonna hook it up a fastethernet without a PC in between is another story :-)
I have an iPod myself and that was the first thing that came to mind. But, like you say, the "PC" inbetween is the rub.
I thought Firewire and USB2 cables were much more limited in their maximum length. I'll look into that option some more. I don't relish the idea of building a Firewire interface into the sensor but I guess it's an option.
Our sensors work exclusively in the low frequency regime (0.01Hz-50kHz) so the RF interference from a transmitter isn't necessarily a killer. There will be some issues to overcome, for sure, which is why we're putting that part off until later.
A microdrive or solid state disk are good options for the storage device, but there still needs to be an intelligent device to act as an interface between the ethernet line and the disk. That's what I'm having trouble finding.
One of the proposed laws gives the consumer $10 for successfully sueing a spammer. Gimme a break, who's got the time to go to court for $10? Another of the proposed laws awarded the spamee $5000 (or was it $2000?) if they had registered on the no-spam list but gets spammed anyway. That would certainly be more of a deterrant, but it doesn't address the problem of finding the spammer to begin with. While it's good to see someone trying to do something about the problem, this ain't it.
More likely, they'll break in, trash the joint, and steal the cool looking robot thingy. It'll be found stripped for parts in an alley in Mexico (or sold into prostitution, one of othe two).