100m JTAG cable

That will not work,

Even an old PC has power supplies.

The first thing to pop near a radiation source is often the power transistors in the power supplies.

Then, there is destructive SCR latch up that any device may suddenly experience (unless the vendor has already tested it in a beam - which for a PC is laughable).

Basically, a PC represents a collection of the least tested, most unreliable components when it comes to neutron or proton beams. For heavy ions, they will fall like a house of cards, within minutes.

The above does not even cover operation without crashing. We have seen vendors leaving beam facilities who were unable to do any testing because stray particles upset not only upset their components, but upset the equipment they were trying to use to set up and measure (don't forget the power supply for the device under test!).

We also use long cables. The cables sometimes have to be specially constructed, as all signals require a ground next to them to eliminate cross talk, and coiling a cable (to store its slack) may add cross talk right back in (the cable must remain non-overlapping).

Our longest cable is 30 meters, however, and that appears to do what we need it to do. We did not require any buffers, drivers, etc. Just good SI engineering over a long, low loss cable (yes, we ran simulations).

The other issue is that any metals in whatever is near or in the beam may be activated (become radioactive), and storing a bunch of activated stuff is not something that you want to discuss with your plant safety people. The experiments themselves are left to "cool off" (most active isotopes to decay) for a couple of months. I would hate to add to that more stuff that I can't use, access, or move.

Since Xilinx is the second largest user of Los Alamos LANSCE beam time (Intel is the first), we have been there, done that. This also means that if you need any information on neutrons, protons, or heavy ions, Xilinx has all the data you need to design your system: on the ground, or in space.

Which as it turns out, is good, as anything from a lcd TV, to an automobile navigation system, to a cell base station or network router switch, will be subject to atmospheric neutrons from cosmic rays. Knowing that Xilinx FPGAs will not undergo destructive latch up, and are robust to upsets becomes just a matter of design, not one that will later back-fire on you, and get you fired.

Aust>> The board will be used to test radiation tolerance techniques, but

Take a look at an Eden PC from VIA. If it meets your requirements for a PC it can run Linux or Windose and doesn't need a SMPSU. It can also have a cardbus main memory instead of a HDD. They have all the features of a full PC, cost about NZ$200-300, They are also fanless so place in a large concrete/lead/steel box and go for it.

Simon

I appreciate the post,

And I am sure others do as well.

But, cosmic ray induced neutrons and protons go through up to 30 meters of concrete. Alphas are stopped by a piece of paper.

So, if you are near a particle beam, there are recoils, strays, etc. at

60 to 800 Mev, and shielding is not very effective. Or even when you are anywhere here on earth (except for a cave).

That is why our reliability engineers who work for the FPGA Lab wear a film badge, and have radiation effects training.

Now, as it turns out, their total dose is pretty small, and is less than the dose they get from the airplane flights to get there, but when they set off the contamination alarms (it happens), that doesn't help much!

Turns out Los Alamos has some pretty strong low pressure events, and the radon from the surrounding ground can come right out, bond to dust, and if it gets on your clothes: ding ding ding (red lights), and you find yourself in the back of the little white van off to the contamination center where there is a little wait to discover that pesky patch of dust where you bent down to straighten the cable...

Aust> Take a look at an Eden PC from VIA. If it meets your requirements for a PC

but, off course, at $200 you can afford to have a few sitting on the shelf just in case an electron, photon, neutron, quark, duck etc decides to go through a sensitive part of the chip.

Simon

PC

a

full

There's code there now. (I checked it out from SVN and submitted a patch to get it to build under FreeBSD)

--
Daniel O'Connor software and network engineer
for Genesis Software - http://www.gsoft.com.au
"The nice thing about standards is that there
are so many of them to choose from."
  -- Andrew Tanenbaum
GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C

I also submitted some patches, with the user visible ones:

- try do detect cable type

- Reconfigure after Flash programming More to come if patches are applied.

--
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------