Rad-Hardened Computers vs. Shielding?

[mikejz]

I have been wondering something: Seeing that we are starting to see embedded computers with the equivalent power of the RAD6000 approaching the size of a single CF memory card--From a total cost, does the approach of using actual radiation shielding start to make more sense than the development of radiation hardened computers?

 

[thermionic]

<br /> I used to do rad-hard electronics for particle accellerators. It was pretty straight-forward at 1.2u and 0.8u. Low-temp processing, SOI, stabilized RAM cells, plenty of extra guard-rings, substrate ties. My hunch is that shielding would be prohibitively heavy unless the human-folk needed it. In which case it would be needed, I guess.

 

[frasercain]

Sounds like military sats. Those are EMP hardened.

 

[webtaz99]

You need both. No reasonable amount of shielding will stop cosmic rays, and cosmic rays produce a shower of secodary and tertiary radiation. So the ICs still need some rad hardening. <div class="Discussion_UserSignature"> </div>

 

[tap_sa]

<font color="yellow">"My hunch is that shielding would be prohibitively heavy unless the human-folk needed it."</font><br /><br />IOW it's much more mass-efficient to have a little larger IC transistors which don't fail even if gammaray hits them, than ordinary tiny stuff and try to make sure that no gamma will ever hit it (because it would certainly fry/flip) ?

 

[CalliArcale]

I definitely agree. It's also cheaper. Although rad-hardened chips obviously lag a bit behind chips meant for terrestrial use, the same trends apply to them, including speed and price. All the attendant problems of trying to totally shield your computer are going to far outweigh the cost of simply buying rad-hardened chips. <div class="Discussion_UserSignature"> <p> </p><p><font color="#666699"><em>"People assume that time is a strict progression of cause to effect, but actually from a non-linear, non-subjective viewpoint it's more like a big ball of wibbly wobbly . . . timey wimey . . . stuff."</em>  -- The Tenth Doctor, "Blink"</font></p> </div>

 

[tap_sa]

<font color="yellow">"rad-hardened chips obviously lag a bit behind chips meant for terrestrial use"</font><br /><br />Cheap source for alt.space rad-harnened CPUs; scavenge 80s PC boxes? <img src="/images/icons/smile.gif" /> <br /><br />The commercial ones cost a pretty penny, IIRC several hundred thousands bucks per chip.

 

[josh_simonson]

If there was enough demand, their price would come down to the levels of other mass produced ICs of comparable die area.

 

[mikejz]

I don't know why this old post came back--but here is my argument. Right now there are commercial embedded computers that have the power to run most spacecraft and are about 1/4 the size of cell phone. My argument is that to take that and encase it in, say 20 lbs of lead would still result in a significant cost savings. <br /><br />As another option, given the small size of these embedded computers, it would be possible to locate them in the center of a spacecraft's fuel tank--providing even more shielding.

 

[webtaz99]

We should start developing rad-hard programmable-architecture chips. There are going to be many different electronic-contol systems needed for space development, and reliability is at least as important as speed or "power". <div class="Discussion_UserSignature"> </div>

 

[thermionic]

>>20 lbs of lead would still result in a significant cost savings<br /><br />Amusingly enough, lead is an alpha emitter. Modern chip packaging is moving to no-lead solder, for this reason. There is often a design rule requiring that there not be a package ball under an on-chip RAM.<br /><br />By the way, 'soft errors' from radiation events is a big issue for memory chip design, and has become an issue for logic design since we went below 0.13u. There is a metric called 'critical charge', defining the magnitude of charge that will flip the state of a flipflop or RAM cell. It's in the range of 2-10fC these days. RAM arrays within chips are 'wrapped' in error-detection-correction (ECC) logic as a rule. System memory, such as a DRAM system, may have several levels of this. Electronic design would be easier if it weren't for radiation, even at the level of terrestrial consumer electronics.

 

[CalliArcale]

<blockquote><font class="small">In reply to:</font><hr /><p>We should start developing rad-hard programmable-architecture chips.<p><hr /></p></p></blockquote><br /><br />They already exist! I've seen FPGAs (Field Programmable Gate Arrays) used in spacecraft. They do have issues, like all FPGAs, but there are space-qualified ones out there. <div class="Discussion_UserSignature"> <p> </p><p><font color="#666699"><em>"People assume that time is a strict progression of cause to effect, but actually from a non-linear, non-subjective viewpoint it's more like a big ball of wibbly wobbly . . . timey wimey . . . stuff."</em>  -- The Tenth Doctor, "Blink"</font></p> </div>

 

[webtaz99]

I suspected as much, I just haven't seen any electronic trade rags for several years.<br /><br />"Programmable Architecture" is a slight step up from FPGAs. They use the same underlying technology, but PA is more "granular" (it involves slightly larger functional blocks). <br /><br />Does anybody remember bit-slice computer ICs? <div class="Discussion_UserSignature"> </div>