IPB

Welcome Guest ( Log In | Register )

7 Pages V  < 1 2 3 4 5 > »   
Reply to this topicStart new topic
High-Temp Electronics For Venus Exploration, recent advances
tasp
post Apr 9 2013, 04:05 AM
Post #31


Member
***

Group: Members
Posts: 903
Joined: 30-January 05
Member No.: 162



However, there are alternatives.

For folks with Sirius satellite radio (to cite a terrestrial example of something you might have), the satellites are in highly elliptical polar orbits, with their high points over the northern hemisphere. This means those satellites spend very little time over the south pole, but quite a bit over the north pole (but at a great altitude)

So if you had a probe on the surface, and it was anticipated to last long enough for this to matter, you might want to put it poleward and have an orbiter in a path similar to the Sirius radio satellite.

{BTW, Sirius did not invent that technique, the Soviets used it extensively for many years prior}
Go to the top of the page
 
+Quote Post
stevesliva
post Apr 9 2013, 05:42 AM
Post #32


Senior Member
****

Group: Members
Posts: 1598
Joined: 14-October 05
From: Vermont
Member No.: 530



QUOTE (siravan @ Apr 8 2013, 05:38 PM) *
I think it is unreasonable to expect a full fledge high-temperature computer with tons of flash memory; at least early on.


Why? Once you have ICs working at your target temp, making them small/highly integrated is something we've figured out how to do really, really, well in the past 40 years since Voyager was state-of-the-art. Sure, you may be able to demonstrate a discrete component system at 600C while ICs only work at 250C. The thing is, though, once you have the basic IC components working at your target temp, I think it's entirely reasonable to assume you'll have a very competent microcontroller with decent volatile storage. Nonvolatile storage might have a different temp spec, but all that means is the power has to be on.

Given those 40 years since Voyager, I'm not sure that with the skills we've gained in highly integrated circuits that we're ever going to have any high-temp technology that's "IC" without it being good enough to make a pretty complex integrated microcontroller. Either you have a lander with relatively unprogrammable instruments transmitting all data in real time-- and perhaps that would be really cool if it was long-lived-- or you have enough integration to do C&DH on the lander. It's really hard to imagine an in-between.
Go to the top of the page
 
+Quote Post
siravan
post Apr 9 2013, 11:51 AM
Post #33


Member
***

Group: Members
Posts: 128
Joined: 10-December 06
From: Atlanta
Member No.: 1472



Making an IC is the difficult part. It is not clear if you have a high temperature semiconductor, it can be turned into an IC easily. I'm no expert by any mean, but my understanding is that the reason silicon ICs exist and work great is the favorable crystalline structure of Si and the existence of insulator bases that integrate well with Si (such as SiO2 and, in case of radiation hardened electronics, Al2O3=sapphire). I'm not sure if silicon carbide shares these favorable features. I guess that the earliest systems will be mainly based on discrete elements and few low density ICs.
Go to the top of the page
 
+Quote Post
dtolman
post Apr 9 2013, 02:07 PM
Post #34


Member
***

Group: Members
Posts: 129
Joined: 20-April 05
Member No.: 291



They've already started making Silicon Carbide IC's - so the barrier has been breached. My understanding is that the problem is making them durable for long term use.
This article is about 4 years old, so the state of the art has advanced since then, but I found it to be a good overview.
Go to the top of the page
 
+Quote Post
stevesliva
post Apr 9 2013, 04:56 PM
Post #35


Senior Member
****

Group: Members
Posts: 1598
Joined: 14-October 05
From: Vermont
Member No.: 530



QUOTE (siravan @ Apr 9 2013, 06:51 AM) *
I'm not sure if silicon carbide shares these favorable features. I guess that the earliest systems will be mainly based on discrete elements and few low density ICs.


That sort of gets at my point-- I tend to think if you have the technology to get reliable "low density ICs" at your spec high temp, you'll very soon have enough VLSI to get a damn decent microcontroller. If it involves wafers--SiC, sapphire, whatever-- and lithography, they sort of hit the ground running these days. Technologies mature with VLSI as a given. I tend to agree that if other technologies aren't mature, the board those ICs go on might be a trip back to times when things were handmade rather than on PCBs, but again, just my opinion, is that if you have anything proven in manufacture that you can call an IC, lithography is probably going to allow you to have a microcontroller with an instruction cache and data cache. Technologies exit the lab with that level of integration. [It's my opinion that] When there is a facility that starts making qualified SiC IC's, it will have the ability to enable microcontrollers from the get-go, so it's fun but not necessary to imagine system architectures that don't need LSI.
Go to the top of the page
 
+Quote Post
Robotbeat
post Jun 11 2013, 12:51 PM
Post #36


Junior Member
**

Group: Members
Posts: 20
Joined: 19-March 13
Member No.: 6897



The big problem, I am told, is memory. We can probably make devices with hundreds or perhaps a couple thousand transistors (enough for a microcontroller, the Intel 4004 had 2300 transistors), but we pretty much can only do SRAM right now, which limits us to maybe 100-200 bits (not even bytes) in the near term (next few years). And even that is difficult. I don't get the idea that it's just a few steps to a big VLSI.
Go to the top of the page
 
+Quote Post
JRehling
post Jun 11 2013, 04:50 PM
Post #37


Senior Member
****

Group: Members
Posts: 2530
Joined: 20-April 05
Member No.: 321



I wonder if a "two-brain" strategy might make sense for a Venus lander. One that has sensitive parts and stays alive for hours, then dies as expected, and one that has tough, robust, vacuum-tube-style electronics that transmits low-bandwidth data for months. In particular, if that could break down into imaging and gas chromatography that is done during descent and just after landing -- perhaps a laser-induced spectrometer -- and then long-term seismological monitoring.

It's encouraging to think that improvements in electronics might make this mission cheaper one day, if not now. The last and only time a US mission transmitted data from the surface of Venus was 1978. That's astonishing.
Go to the top of the page
 
+Quote Post
Robotbeat
post Jun 12 2013, 05:13 PM
Post #38


Junior Member
**

Group: Members
Posts: 20
Joined: 19-March 13
Member No.: 6897



Indeed, I had thought of such an approach. It makes certain things a lot easier, for instance the camera and as you said some kind of spectrometer. You'd also want probably a pressure probe, temperature probe, and anemometer (wind speed), too, to determine if seismic events are really just caused by the wind.
Go to the top of the page
 
+Quote Post
stevesliva
post Jun 13 2013, 05:37 AM
Post #39


Senior Member
****

Group: Members
Posts: 1598
Joined: 14-October 05
From: Vermont
Member No.: 530



QUOTE (Robotbeat @ Jun 11 2013, 08:51 AM) *
I don't get the idea that it's just a few steps to a big VLSI.


As an example, I don't necessarily think that once a technology is reliable, there's any hurdle between 4004 and 8085 these days. Took them 5 years in the 70s, but the lithography was all new. Today, there's no point in technology development where anyone would cut and run with a 4004 thinking they next step to the 8085 wouldn't be quickly surmountable. Sure, a Pentium (22 years) might be a leap. I don't think it serves much purpose to imagine it will take as many years to move up the Moore's law curve as it did in the 70s, when microns were considered really small geometries. There's not going to be a long period comparable to the 70s when we could make microcontrollers, but only laughably piddling ones. Plus, the thing is that even if it did take the whole 5 years it did in the 70s, that doesn't seem that long when you're talking about unmannedspaceflight! I think it's conceivable they could solve board reliability and instrument reliability before they solve IC reliability, and someone will send a system with discrete components. I just don't think it's that plausible we'll send one with a 4004-level IC when the wait for 8085 or 8086 will not be long at all.
Go to the top of the page
 
+Quote Post
tasp
post Jun 13 2013, 05:48 AM
Post #40


Member
***

Group: Members
Posts: 903
Joined: 30-January 05
Member No.: 162



I'm wondering if something like a corner reflector might be made of some high temperature (and chemical) resistant material, that could have it's dielectric properties, or reflection angle, (or something) that varies in a predictable way with temperature and or pressure.

The reflector could be dropped and then illuminated with a radio signal from an orbiter, and the return echo would be phase modulated, or whatever the device can be made to do to alter the signal, and then you would have a point on the surface that could me monitored for one or a very few parameters, for as long as the orbiter lasted.

A reflector on an anemometer would frequency modulate the reflection (if done properly) in proportion to the wind velocity. Two such reflectors on a bimetallic arm on a pivot would introduce a bias to the wind signal that would be temperature related.

Might be a way to get some very basic long term data from a few interesting sites. Just need to have a very simple, and robust device in the right location.
Go to the top of the page
 
+Quote Post
Robotbeat
post Jun 13 2013, 12:56 PM
Post #41


Junior Member
**

Group: Members
Posts: 20
Joined: 19-March 13
Member No.: 6897



QUOTE (stevesliva @ Jun 12 2013, 11:37 PM) *
As an example, I don't necessarily think that once a technology is reliable, there's any hurdle between 4004 and 8085 these days. Took them 5 years in the 70s, but the lithography was all new. Today, there's no point in technology development where anyone would cut and run with a 4004 thinking they next step to the 8085 wouldn't be quickly surmountable. Sure, a Pentium (22 years) might be a leap. I don't think it serves much purpose to imagine it will take as many years to move up the Moore's law curve as it did in the 70s, when microns were considered really small geometries. There's not going to be a long period comparable to the 70s when we could make microcontrollers, but only laughably piddling ones. Plus, the thing is that even if it did take the whole 5 years it did in the 70s, that doesn't seem that long when you're talking about unmannedspaceflight! I think it's conceivable they could solve board reliability and instrument reliability before they solve IC reliability, and someone will send a system with discrete components. I just don't think it's that plausible we'll send one with a 4004-level IC when the wait for 8085 or 8086 will not be long at all.

I think you're missing something, here... With high-temp electronics, we don't have access to Moore's Law. There are just one or maybe two places that really are even trying to do complex integrated circuits. There's very, very little financial incentive for improvement, it's essentially ALL gov't funded. The state of the art can improve, but it is a direct function of money spent, not purely time. This is markedly different from the situation with the early IC CPUs like the 4004, where you have a huge market for improvements. Nowadays, tens of billions of dollars are spent on fabs and improving the technology for making conventional integrated circuits. Definitely not the case for silicon-carbide circuits, nor is it likely to ever be so. The market is tiny, you can't simply wait for it!

That said, an 8-bit architecture may make more sense than a 4-bit even at the extreme limit of minimal transistor count.
Go to the top of the page
 
+Quote Post
stevesliva
post Jun 13 2013, 09:24 PM
Post #42


Senior Member
****

Group: Members
Posts: 1598
Joined: 14-October 05
From: Vermont
Member No.: 530



QUOTE (Robotbeat @ Jun 13 2013, 08:56 AM) *
I think you're missing something, here... With high-temp electronics, we don't have access to Moore's Law.


Yeah, but you have access to anything that trickles down from the silicon ecosystem. It's unlikely that you're going to end up with an IC made on a 50mm wafer with 10um geometries. Something like SiC wafers with W wiring... that will still get proven in a 150mm or 200mm research fab at submicron geometries. They aren't reinventing everything from the 1970 level.

The funding is otherwise a huge issue. But I think automotive applications will help. And for whatever reason, it seems LED makers are interested in SiC.
Go to the top of the page
 
+Quote Post
Robotbeat
post Jun 17 2013, 01:12 PM
Post #43


Junior Member
**

Group: Members
Posts: 20
Joined: 19-March 13
Member No.: 6897



Power draw is a huge issue, because you can't do CMOS. Memory draws a ton of power, so even if you can build a big chip, you would need a very large power source. Which is a pretty big problem on the surface of Venus, where very little light gets through, your cold end for a heat engine is already nearly as high as the hot end for MMRTG.
Go to the top of the page
 
+Quote Post
dtolman
post Jun 17 2013, 09:13 PM
Post #44


Member
***

Group: Members
Posts: 129
Joined: 20-April 05
Member No.: 291



Doesn't seem to me that high temperature memory is that far behind IC's. As someone mentioned before, TI is selling an off-the-shelf flash memory unit that is rated up to 210 C currently. I've seen some research documents that refer up to 300 C memory units in lab tests being run now, so the OTS max temperature should keep pushing up over the next few years.
EDIT: and I now see this note by Raytheon that they are working on a SiC based CMOS rated at 450C.
So why can't we do CMOS on Venus?

This might be of interest - a report by Honeywell on the challenges (and solutions) for building 250 C rated ICs, that goes into memory solutions as well.
--
IC related Bonus - found this paper describing the results of actual testing at 500C with a custom IC built by NASA. Think I missed this on my last review of papers on high temperature ICs.
Go to the top of the page
 
+Quote Post
Robotbeat
post Jun 18 2013, 12:18 PM
Post #45


Junior Member
**

Group: Members
Posts: 20
Joined: 19-March 13
Member No.: 6897



QUOTE (dtolman @ Jun 17 2013, 03:13 PM) *
Doesn't seem to me that high temperature memory is that far behind IC's. As someone mentioned before, TI is selling an off-the-shelf flash memory unit that is rated up to 210 C currently. I've seen some research documents that refer up to 300 C memory units in lab tests being run now, so the OTS max temperature should keep pushing up over the next few years.
EDIT: and I now see this note by Raytheon that they are working on a SiC based CMOS rated at 450C.
So why can't we do CMOS on Venus?

This might be of interest - a report by Honeywell on the challenges (and solutions) for building 250 C rated ICs, that goes into memory solutions as well.
The tech that works up to 250C is not usable to the ~500C needed for Venus (it /can't/ work, the tech simply stops producing gain at that point). The Raytheon CMOS SiC stuff is interesting, but a lot of this stuff has been "talked about" for a good couple decades without significant progress towards something actually usable. Goes on the list, though!
QUOTE
IC related Bonus - found this paper describing the results of actual testing at 500C with a custom IC built by NASA. Think I missed this on my last review of papers on high temperature ICs.

Ah, yes, Philip Neudeck... One of the people I've talked to on trying to see what will and won't work. This field is so small that a lot of the references are by the same people. Some people are really optimistic about getting a usable large-scale IC in a short time, others are quite pessimistic. The consensus currently is that memory is a really hard problem. You need Megabytes of memory, not just the 100 bits that might be doable in the next 5 years.

(And Mike K, also one of the authors of that paper, is probably the most interesting person I've ever met at NASA. He's on my list of most favorite people ever.)

And the question isn't about physical impossibility or not, but on state of the technology... If you want a mission that actually /happens/ and isn't a paper study, we can't be happy with a TRL of "well, physically it's possible." The people who decide whose mission to build will reject that every time.

...That said, prove me wrong!
Go to the top of the page
 
+Quote Post

7 Pages V  < 1 2 3 4 5 > » 
Reply to this topicStart new topic

 



RSS Lo-Fi Version Time is now: 31st October 2024 - 11:37 PM
RULES AND GUIDELINES
Please read the Forum Rules and Guidelines before posting.

IMAGE COPYRIGHT
Images posted on UnmannedSpaceflight.com may be copyrighted. Do not reproduce without permission. Read here for further information on space images and copyright.

OPINIONS AND MODERATION
Opinions expressed on UnmannedSpaceflight.com are those of the individual posters and do not necessarily reflect the opinions of UnmannedSpaceflight.com or The Planetary Society. The all-volunteer UnmannedSpaceflight.com moderation team is wholly independent of The Planetary Society. The Planetary Society has no influence over decisions made by the UnmannedSpaceflight.com moderators.
SUPPORT THE FORUM
Unmannedspaceflight.com is funded by the Planetary Society. Please consider supporting our work and many other projects by donating to the Society or becoming a member.