It's hard finding suitable conflat chambers and I wouldn't trust vibration isolation of a refurbished optical table so those parts most likely have to be new. That's 25K right there. If you got access to a lot of stainless steel tubing and are good with welding AND have a way to electropolish the chamber in the end, you can cut costs drastically but most people even at large universities don't have that. (BTW, I have first hand experience with old optical tables and their vibration characteristics).
If you have big, no huge, lathes and mills you can make your own conflat flanges. If you have proper equipment you can,ake your own bellows and feedthrus. Basically if you already have $300K worth of equipment, building a UHV appratus will cost you peanuts and a lot of time. Otherwse you will have a hard time scrounging many crucial parts. You can indeed get refurbed pumps cheap, maybe a manipulator if you are lucky. Certainly most electronics (it's very simple stuff and I have found ebay to be a good source though we only use such parts in few places where they dont really matter if they fail). Believe me, if you want a world-class STM, $100K in the US is a rather small estimate, unless you are an STM building company with a lot of equipment already in-house. 10^-6 is not gonna do it for most physics studies. Around middle 10^-9 you have a prayer of having a stable junction for a while. Look at the images from build-it-yourself sites: they all have tip changes, instabilities, and tell-tale artefacts at the edges of their images. You see that and any serious STM person will send you back for real data. I wouldn't even trust their images of gold steps if I hadn't known from many other sources they were there and they are supposed to look similar to what they see.
This is gettig offtopic but... Building a UHV system (10^-10 torr type vacuum) is expensive though not hard. Part of it is that you'll need quality metal seal flanges to connect parts of your system, another part is that you need an ion pump because turbo and mechnical pumps introduce too much vibration. Of course if you got the money you could also dump it all inside a liquid helium dewar and be done with cryopumping alone. If you are scavenging equipment then look for a mechanical, a turbo and an ion pump (you'll be using them in that sequence), also find enough conflat hardware to do the task and you'll need some way to prepare metal surfaces at least, so find a good sputter gun. And also design and build a way to manipulate samples in vacuum, i.e. either a bunch of vacuum manipulators or UHV compatible motors. Also dont forget UHV vacuum gauges. In all, the systems I built usually cost about $300K-$500K and you can cut corners and make do with $200K. If you scavenge you can maybe go down to $100K.
Dude, that's bogus. Go to a decent library and check out a book by Julian Chen "Introduction to Scanning Tunneling Microscopy" ASIN: 0195071506. It's the bible of STM. It will tell you more or less everything you need to know to build one except may electronics but it will give you plenty of pointers to that too. Most of the information on that site is either in the book or a design that can be made with info from the book rather trivially. They have done a fine job but their IP is not worth much even as a giveaway.
a. Please tell me you got an F for your efforts as an undergrad. I say this because the very essense of tunneling is that it falls of as an exponential, not as a power law.
b. I assume you were doing STM in air or else moving a setup downstairs would have taken at least a week to rebake the vacuum chamber. Now in air, you have a thin layer of water on the surface which surprisingly makes it easier to stabilize a junction.
c. Tips aren't quite some much of a problem, I agree with that, especially since very good tips can be bought commercially. And cutting a wire may work well for metals and semiconductors. Attaching a probe is usually easy because the only part you can't touch is the very end.
Responding to original poster:
d. There is no need to calibrate the piezo to be able to tunnel, that's what a feedback loop is for. You do have to have an idea of what parameters to use so it doesn't ring.
e. In fact an ordinary op-amp will do fine and a clean dc supply from any decent manufacturer will do the trick. Look at bio and chemistry literature dealing with patch clamp applications for good references on more sophisticated designs but it aint rocket science. The one hard part is to make sure you put your setup far from any 60 Hz source and have no ground loops or even no weak grounds anywhere in you setup.
Lastly, the hard part about STM is getting meaningful data. You typically get junctions that aren't so good and you need to be able to tell whether it is the tip or the surface. Generally to do that you need to do this for a few years and build up and internal reference for which type of crappy junction corresponds to which problem. Then getting a good junction and some data becomes easier.
Beetle design, also known for its creator as the Besocke design is not unique in allowing full 3D approach positioning. It does have many advantages like farly good rigidity, thermal compensation and ease of assembly. There are designs which are even better than slipstick, such as the good old inchworm design and its offshoots, especially ones designed for low temperature operation (some are patented). Even slip-stick isn't limited to beetles. Another design was patented early on by Lyding and is also thermally compensated, though harder to build than Besocke stuff. Bottom line, look around. There is no shortage of really cool designs to fit any need.
I am probably quite qualified to answer your question seeing as I do STM research for a living. Your second question is easier to answer, so I'll do that first: there are two ways, either you put down enough of them to assure there will be a molecule in the range of your scanner whereever you approach or you use some other technique like lithography to make small structures then another technique to deposit your molecules near those structures then (if you got the microscope that allows it) position your tip optically near the structure and spend days on looking around with STM until you find it.
Now you first question. STM can be used to find some structural info from large molecules. My lab has done some research on nanotubes and you can get atomic resolution on those and then determine their helicity. People have also imaged bio stuff and for some smaller molecules have seen the structure. Even DNA has been imaged. That said, STM is not a great structure probe, it is a great probe of electronic states.
Last word of warning: people rarely realize that STM in air is not going to tell you anything that you can rely on physics-wise. The reason is that all surfaces exposed to air are covered in a thin layer of water which makes the interpretation of data hard. What they show on that page is a toy though well-thought-out and maybe even useful to some. Seeing atomic steps on gold and "atomic" resolution on HOPG is not hard, just don't hold your breath for something like atomic resolution on gold, or silicon, or anything else really. For that you at least need a UHV system. Cheers.
I am guessing it will work the other way around. Mac people have a sense of good UI so they will drive this part of the app which will then be ported back to X11. At least that's the sensible way things should evolve and may be the strongest reason to port K?>*&):"(&^ stuff to the Mac.
And how do you propose we "ween ourselves from exploitative labor"? Oh, I know, how about getting rid of all illegal immigrants. Why, then there would be a labor shortage on the farms, wages would go up, people would get benefits, etc. You are advocating same thing buddy. And being very indignant about it to boot.
I would think it a good thing if it were for all US citizens and legal aliens. If such a database were implemented the Feds could require, e.g. all supermarkets and food vendors in general to collect DNA and on the spot verify immigration status. This would make it nearly impossible for illegal immigrants to buy food, clothes, medicine, etc. You could also then investigate anyone buying food in bulk and see if they sell it on black market which would be sure to form. You could then really open a hunt for illegal immigrants and clean up US.
Dunno about java servlets, but java applets were around seven years ago. So broadly speaking, web technologies were around seven years ago even by your metric.
It is ~30 fps on LAN and it is useable, as in voice and video are coherent, picture is abit shaky but not painful. You can't move fast though or else it'll be a blur. However we use it to talk coast-to-coast. On university-to-university network you get ~20 fps and the quality degradation is notable. Now you get a picture that is a bit retarded and when someone moves (even medium speed) it results in unhappiness. Put the sucker on cable modem and you get 10 fps. Now it is virtually unusable in the sense that you are not getting much more than voice and what you do get is painfully choppy and often artefacted. IMHO, anything below 15 fps is not even worth consideration.
Well now, SCO has shown itself to be a litiguous company and did things that make people VERY unhappy, like implying that would reopen the AT&T case. At this point the only assurance IBM would have of SCO shutting up for good is to drive them out of business. So it is unclear that IBM would settle the counterclaim. Besides, IBM is not the only company now with lawsuits against SCO. And some of these are in motion to set a precedent. But regardless, once the lawsuit vanishes so will the investors and so will SCO.
In particular, is there any system for math handwriting recognition. Something which could interpret definite integrals, norms of matrices, and ideally more sophisticated things like group-theoretical and topological notations.
Knowing when Moore's law will run out of steam is useful because this is the time when people will seriously consider better materials than silicon, this is when all this talk of better computing techniques (whether the simple tri-valued logic or quantum computing) will finally get the all out funding that better lithography tech now enjoys. I imagine that once you can no longer just turn the crank and push the process smaller, Intel will start to decline and streamlined instruction sets will become more important. Basically, once Moore's law is out we might be getting back to quality over quantity in chips. So Moore's law dying is something to look forward to if you are interested in cool but obscure technologies that are currently sidelined by silicon's immense scaling capacity.
I can't believe noone yet mentioned Stanislav Lem. One his more humorous stories dealt with a similar machine though one that could produce real poetry, meaningful, beautifully written and confroming to arbitrary constraints, like one where all words had to begin with same letter. If you read the story you know this invention will lead to no good.
On second thought, I just read the thread on the swedish grad student solving part of Hibert's problem 16 and I have to say: you are right about/. croud being loners. That was just sad.
Well, huh? Did we see the same movie? Romance? His GF dies in his hands and the guy just says "bye" and goes on to more ass-kicking... I was really expecting him to do the Romeo thing and maybe then have the machines fight Smith and when they'd both collapse leave humanity be. The scene where Trinity dies showed clearly that "romance" was merely a plot device to advance action. Of course, Keanu couldn't emote love even acting opposite his real GF, if he got one. The guy can express emotions (other than surprise) just about as well as a doorknob so I am not sure where you saw emotions in that movie. Maybe you have filled in the blanks because the script certainly calls for emotions but the final cut leaves a lot to be desired in that department. Action was fine, though everyone expected some imagination and this was basically a rehash of the first two movies. Zion was essentially a scaled up version of their ships. So yes this movie was packed with rather redundant action sequences.
So to recap: the Matrix pretended to be a love story but needed a different cast to have a hope of pulling it off. It has predictable script (who doubted that the ship was gonna make it to Zion?), it has wooden acting, it has derivative action, and it has a nice ironic ending. This does not an *awesome* movie make.
Quite frankly, I think it is rare that one needs to change all words with one format to another type of format and there is so many words that it is hard to do by hand. In those cases, I think the most reasonable approach would be the one your company has taken: have a developer code up a way to do this, put a GUI button/icon/widget to trigger the automatic processing and train users about the capabilities of said widget. I cannot see your html cleanup being any more efficient if every users were to roll up their own utility.
Curiously, all your explicit examples are ones where you need a better gui. A command line interface will not help you sum data in a column of a spreadsheet faster or easier. We are not discussing the need to know how to use the tools and knowing which tool is better for the job, we are merely discussing whether a command line is the optimal tool for most "regular" tasks.
Slashdot Mirror
It's hard finding suitable conflat chambers and
,ake your own bellows
I wouldn't trust vibration isolation of a
refurbished optical table so those parts
most likely have to be new. That's 25K right
there. If you got access to a lot of stainless
steel tubing and are good with welding AND have
a way to electropolish the chamber in the end,
you can cut costs drastically but most people
even at large universities don't have that.
(BTW, I have first hand experience with old
optical tables and their vibration characteristics).
If you have big, no huge, lathes and mills you
can make your own conflat flanges. If you have
proper equipment you can
and feedthrus. Basically if you already have $300K
worth of equipment, building a UHV appratus will
cost you peanuts and a lot of time. Otherwse
you will have a hard time scrounging many crucial
parts. You can indeed get refurbed pumps cheap,
maybe a manipulator if you are lucky. Certainly
most electronics (it's very simple stuff and I
have found ebay to be a good source though we
only use such parts in few places where they dont
really matter if they fail). Believe me, if you
want a world-class STM, $100K in the US is a
rather small estimate, unless you are an STM
building company with a lot of equipment already
in-house. 10^-6 is not gonna do it for most
physics studies. Around middle 10^-9 you have
a prayer of having a stable junction for a while.
Look at the images from build-it-yourself sites:
they all have tip changes, instabilities, and
tell-tale artefacts at the edges of their images.
You see that and any serious STM person will
send you back for real data. I wouldn't even
trust their images of gold steps if I hadn't
known from many other sources they were there and
they are supposed to look similar to what they
see.
This is gettig offtopic but...
Building a UHV system (10^-10 torr type vacuum)
is expensive though not hard. Part of it is that
you'll need quality metal seal flanges to connect
parts of your system, another part is that you
need an ion pump because turbo and mechnical
pumps introduce too much vibration. Of course
if you got the money you could also dump it all
inside a liquid helium dewar and be done with
cryopumping alone.
If you are scavenging equipment then look for
a mechanical, a turbo and an ion pump (you'll be
using them in that sequence), also find enough
conflat hardware to do the task and you'll need
some way to prepare metal surfaces at least, so
find a good sputter gun. And also design and build
a way to manipulate samples in vacuum, i.e.
either a bunch of vacuum manipulators or UHV
compatible motors. Also dont forget UHV vacuum
gauges. In all, the systems I built usually cost
about $300K-$500K and you can cut corners and make
do with $200K. If you scavenge you can maybe go down
to $100K.
Dude, that's bogus. Go to a decent library and
check out a book by Julian Chen "Introduction to Scanning Tunneling Microscopy"
ASIN: 0195071506.
It's the bible of STM. It will tell you more or
less everything you need to know to build one
except may electronics but it will give you
plenty of pointers to that too.
Most of the information on that site is either
in the book or a design that can be made with info
from the book rather trivially. They have done a
fine job but their IP is not worth much even as
a giveaway.
a. Please tell me you got an F for your efforts
as an undergrad. I say this because the very
essense of tunneling is that it falls of as an
exponential, not as a power law.
b. I assume you were doing STM in air or else
moving a setup downstairs would have taken at
least a week to rebake the vacuum chamber. Now
in air, you have a thin layer of water on the
surface which surprisingly makes it easier to
stabilize a junction.
c. Tips aren't quite some much of a problem, I
agree with that, especially since very good tips
can be bought commercially. And cutting a wire
may work well for metals and semiconductors.
Attaching a probe is usually easy because the
only part you can't touch is the very end.
Responding to original poster:
d. There is no need to calibrate the piezo to be
able to tunnel, that's what a feedback loop is for.
You do have to have an idea of what parameters to
use so it doesn't ring.
e. In fact an ordinary op-amp will do fine and a
clean dc supply from any decent manufacturer
will do the trick. Look at bio and chemistry
literature dealing with patch clamp applications
for good references on more sophisticated designs
but it aint rocket science. The one hard part is
to make sure you put your setup far from any
60 Hz source and have no ground loops or even
no weak grounds anywhere in you setup.
Lastly, the hard part about STM is getting
meaningful data. You typically get junctions that
aren't so good and you need to be able to tell
whether it is the tip or the surface. Generally
to do that you need to do this for a few years and
build up and internal reference for which type
of crappy junction corresponds to which problem.
Then getting a good junction and some data becomes
easier.
Beetle design, also known for its creator as the
Besocke design is not unique in allowing full
3D approach positioning. It does have many advantages
like farly good rigidity, thermal compensation and
ease of assembly.
There are designs which are even better than
slipstick, such as the good old inchworm design
and its offshoots, especially ones designed for
low temperature operation (some are patented).
Even slip-stick isn't limited to beetles. Another
design was patented early on by Lyding and is also
thermally compensated, though harder to build than
Besocke stuff.
Bottom line, look around. There is no shortage of
really cool designs to fit any need.
I am probably quite qualified to answer your
question seeing as I do STM research for a living.
Your second question is easier to answer, so I'll
do that first:
there are two ways, either you put down enough
of them to assure there will be a molecule in the
range of your scanner whereever you approach
or you use some other technique like lithography
to make small structures then another technique
to deposit your molecules near those structures
then (if you got the microscope that allows it)
position your tip optically near the structure
and spend days on looking around with STM until
you find it.
Now you first question. STM can be used to find
some structural info from large molecules. My lab
has done some research on nanotubes and you can
get atomic resolution on those and then determine
their helicity. People have also imaged bio stuff
and for some smaller molecules have seen the
structure. Even DNA has been imaged. That said,
STM is not a great structure probe, it is a great
probe of electronic states.
Last word of warning: people rarely realize that
STM in air is not going to tell you anything
that you can rely on physics-wise. The reason is
that all surfaces exposed to air are covered in a
thin layer of water which makes the interpretation
of data hard. What they show on that page is a toy
though well-thought-out and maybe even useful to
some. Seeing atomic steps on gold and "atomic"
resolution on HOPG is not hard, just don't hold
your breath for something like atomic resolution
on gold, or silicon, or anything else really.
For that you at least need a UHV system.
Cheers.
HomEostasis? Freudian slip?
EU is a bit weak right now after Beagle but I am
not sure whatcha talking about the Chinese. They
just had a major success and look to be OK.
I am guessing it will work the other way around.
Mac people have a sense of good UI so they will
drive this part of the app which will then be
ported back to X11. At least that's the sensible
way things should evolve and may be the strongest
reason to port K?>*&):"(&^ stuff to the Mac.
A flop so monstrous, people even forgot about /. comments to boot (AFAICT).
it when writing this article AND when writing
And how do you propose we "ween ourselves from
exploitative labor"? Oh, I know, how about getting
rid of all illegal immigrants. Why, then there would
be a labor shortage on the farms, wages would go up,
people would get benefits, etc.
You are advocating same thing buddy. And being very
indignant about it to boot.
And when you say smart you mean goodlooking, right?
Gotta remember my Brittish.
I would think it a good thing if it were for all
US citizens and legal aliens. If such a database
were implemented the Feds could require, e.g. all
supermarkets and food vendors in general to
collect DNA and on the spot verify immigration
status. This would make it nearly impossible for
illegal immigrants to buy food, clothes, medicine,
etc. You could also then investigate anyone
buying food in bulk and see if they sell it on
black market which would be sure to form. You
could then really open a hunt for illegal
immigrants and clean up US.
Dunno about java servlets, but java applets were
around seven years ago. So broadly speaking, web
technologies were around seven years ago even by
your metric.
It is ~30 fps on LAN and it is useable, as in voice and video are coherent, picture is abit shaky but not painful. You can't move fast though
or else it'll be a blur.
However we use it to talk coast-to-coast. On
university-to-university network you get ~20 fps
and the quality degradation is notable. Now you
get a picture that is a bit retarded and when
someone moves (even medium speed) it results in
unhappiness.
Put the sucker on cable modem and you get 10 fps.
Now it is virtually unusable in the sense that you
are not getting much more than voice and what you
do get is painfully choppy and often artefacted.
IMHO, anything below 15 fps is not even worth
consideration.
Well now, SCO has shown itself to be a litiguous
company and did things that make people VERY
unhappy, like implying that would reopen the AT&T
case. At this point the only assurance IBM would
have of SCO shutting up for good is to drive them
out of business. So it is unclear that IBM would
settle the counterclaim. Besides, IBM is not the
only company now with lawsuits against SCO. And
some of these are in motion to set a precedent.
But regardless, once the lawsuit vanishes so will
the investors and so will SCO.
In particular, is there any system for math
handwriting recognition. Something which could
interpret definite integrals, norms of matrices,
and ideally more sophisticated things like
group-theoretical and topological notations.
Knowing when Moore's law will run out of steam
is useful because this is the time when people
will seriously consider better materials than
silicon, this is when all this talk of better
computing techniques (whether the simple
tri-valued logic or quantum computing) will
finally get the all out funding that better
lithography tech now enjoys. I imagine that once
you can no longer just turn the crank and push
the process smaller, Intel will start to decline
and streamlined instruction sets will become more
important. Basically, once Moore's law is out
we might be getting back to quality over quantity
in chips. So Moore's law dying is something to
look forward to if you are interested in cool
but obscure technologies that are currently
sidelined by silicon's immense scaling capacity.
I can't believe noone yet mentioned Stanislav
Lem. One his more humorous stories dealt with
a similar machine though one that could
produce real poetry, meaningful, beautifully
written and confroming to arbitrary constraints,
like one where all words had to begin with same
letter. If you read the story you know this
invention will lead to no good.
Heh, I initially misread this as:
The United BibleWrapped America.
Works both ways.
Why is it built on top of that?
On second thought, I just read the thread on the /. croud being loners. That was just sad.
swedish grad student solving part of Hibert's
problem 16 and I have to say: you are right about
Well, huh? Did we see the same movie? Romance?
His GF dies in his hands and the guy just says
"bye" and goes on to more ass-kicking... I was
really expecting him to do the Romeo thing and
maybe then have the machines fight Smith and when
they'd both collapse leave humanity be. The scene
where Trinity dies showed clearly that "romance"
was merely a plot device to advance action.
Of course, Keanu couldn't emote love even acting
opposite his real GF, if he got one. The guy can
express emotions (other than surprise) just about
as well as a doorknob so I am not sure where you
saw emotions in that movie. Maybe you have filled
in the blanks because the script certainly calls
for emotions but the final cut leaves a lot to be
desired in that department.
Action was fine, though everyone expected some
imagination and this was basically a rehash of the
first two movies. Zion was essentially a scaled up
version of their ships. So yes this movie was
packed with rather redundant action sequences.
So to recap: the Matrix pretended to be a love story
but needed a different cast to have a hope of
pulling it off. It has predictable script (who
doubted that the ship was gonna make it to Zion?),
it has wooden acting, it has derivative action,
and it has a nice ironic ending. This does not
an *awesome* movie make.
Quite frankly, I think it is rare that one needs
to change all words with one format to another
type of format and there is so many words that it
is hard to do by hand. In those cases, I think the
most reasonable approach would be the one your
company has taken: have a developer code up a way
to do this, put a GUI button/icon/widget to trigger
the automatic processing and train users about
the capabilities of said widget. I cannot see
your html cleanup being any more efficient if every
users were to roll up their own utility.
Curiously, all your explicit examples are ones
where you need a better gui. A command line
interface will not help you sum data in a column
of a spreadsheet faster or easier. We are not
discussing the need to know how to use the tools
and knowing which tool is better for the job, we
are merely discussing whether a command line is
the optimal tool for most "regular" tasks.