Folks
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
As we won't now have an opportunity to discuss this in a telecon, can we have a clear line on this? I can just see that someone's gonna ask me this question at my presentation.
73, Howard G6LVB
Hello Howard!
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
As we won't now have an opportunity to discuss this in a telecon, can we have a clear line on this? I can just see that someone's gonna ask me this question at my presentation.
One configuration we discussed at one time was to put an "analog IF" inside the SDX. With the SDX operating with a nominal 10.7 MHz IF, we'd simply switch in a gain section with AGC, or switch it out and switch in the digital processing system. That way, the SDX device would contain both analog and digital transponder circuitry, under control of the CAN bus, and with little added mass or complexity.
The power-on default mode would be analog so in case of an AO-10-liek state, we'd have the highest probability of something working, even if it wouldn't be optimal.
73,
Lyle kK7P
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
Seriously, if we have 2 SDX's, aren't they backups for each other?
Matt
I think the issue is that people are worried about radiation damage to the SDX CPU. If the DSP dies so are we. We really need to subject that puppy to a brutal amount of radiation in our testing and indeed test to destruction.
Bob N4HY
Matt Ettus wrote:
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
Seriously, if we have 2 SDX's, aren't they backups for each other?
Matt
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
I think Bob's dead on with regard to the initial concern about radiation damage. Much of that can be mitigated once the results of the radiation testing are known. At this point we don't have that data. There is also the memory of how AO-10's analog systems continued to function for years long after the digital systems had failed.
We would also do well to head Martin's points about design diversity. We had to blow up a good rocket once on the Eastern Range due to the lack of it. I think the problem with proposing the ACP as providing the diverse backup is that it requires a completely new ground station investment.
There's also added risk by increasing the complexity by adding an analog block to make make an analog transponder. That risk has to be weighed against the benefit of gaining diversity in the overall system as well as the potential to extend the mission even after the SDXs have served a long and productive life. Until some serious work is done on the analog block and the interfaces it would requie, we're trying to compare two unknowns.
My suggestion is to continue the work on an analog path until we have sufficient data to make a reasoned and thoughtful engineering analysis. Then follow that path wherever it may lead.
Lee-KU4OS
On Tue, 2006-10-03 at 01:05 -0400, Robert McGwier wrote:
I think the issue is that people are worried about radiation damage to the SDX CPU. If the DSP dies so are we. We really need to subject that puppy to a brutal amount of radiation in our testing and indeed test to destruction.
Bob N4HY
Matt Ettus wrote:
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
Seriously, if we have 2 SDX's, aren't they backups for each other?
Matt
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
We also need to have backup hardware for digital users. Class 1 users have no backup without 2 SDXs. Likewise, there needs to be some redundancy in the ACP so that class 2 and 3 services don't disapear if one component (ADC, FPGA, etc.) fails.
73,
John KD6OZH
----- Original Message ----- From: "Lee McLamb" ku4os@cfl.rr.com To: eagle@amsat.org Sent: Wednesday, October 04, 2006 00:18 UTC Subject: [eagle] Re: The thorny issue of SDX/Analog linear transpnder
I think Bob's dead on with regard to the initial concern about radiation damage. Much of that can be mitigated once the results of the radiation testing are known. At this point we don't have that data. There is also the memory of how AO-10's analog systems continued to function for years long after the digital systems had failed.
We would also do well to head Martin's points about design diversity. We had to blow up a good rocket once on the Eastern Range due to the lack of it. I think the problem with proposing the ACP as providing the diverse backup is that it requires a completely new ground station investment.
There's also added risk by increasing the complexity by adding an analog block to make make an analog transponder. That risk has to be weighed against the benefit of gaining diversity in the overall system as well as the potential to extend the mission even after the SDXs have served a long and productive life. Until some serious work is done on the analog block and the interfaces it would requie, we're trying to compare two unknowns.
My suggestion is to continue the work on an analog path until we have sufficient data to make a reasoned and thoughtful engineering analysis. Then follow that path wherever it may lead.
Lee-KU4OS
On Tue, 2006-10-03 at 01:05 -0400, Robert McGwier wrote:
I think the issue is that people are worried about radiation damage to the SDX CPU. If the DSP dies so are we. We really need to subject that puppy to a brutal amount of radiation in our testing and indeed test to destruction.
Bob N4HY
Matt Ettus wrote:
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
Seriously, if we have 2 SDX's, aren't they backups for each other?
Matt
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Hi Matt
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
The point is that the commitment to have an analog backup had already been made. You may not have been party to the discussions around July 2005, but believe me they were just as vociferous as the recent disinformation floating about regarding S1. It is important to consider seriously what the members want - after all, they are paying for it, and there would be no AMSAT-NA or Eagle without them.
Seriously, if we have 2 SDX's, aren't they backups for each other?
Well, this raises an interesting point. My understanding from discussions over this side of the pond with Martin Sweeting is that redundancy only truly exists if you also have two different designs: having two devices of the same design does not resolve design flaws. In SSTL's case, Martin's view is that the majority of failures are now down to design, with a minority environmental (radiation, thermal, outgassing etc etc).
73, Howard G6LVB
There's a couple of points here :-
1) Howard is giving a presentation on Saturday about the SDX. It's almost inevitable that somebody will ask about an analogue backup, and it's right to ask the question. Howard simply wants to be able to give the official answer.
2) SDX will fly on P3E which is an excellent proving ground for this technology. It will also fly on ESEO, which is due for launch in 2008. ESEO will go into GTO, so it will be another good proving ground, although how much service it gets before Eagle is ready is unknown.
Seems to me like there are two options :-
A) Plan for 2 SDXs + ACP, with no analogue backup. Risky, but by the time Eagle is ready P3E and ESEO will have been in service for some time. This will either give everyone confidence in the technology or give us an opportunity for a re-design if it fails. And of course we test it properly in advance.
B) Plan to fly at least one analogue backup.
Either way, don't miss Howard's presentation - if you haven't seen the SDX working, you'll be in for a real treat!
regards
Grant
Martin is exactly right. Two of the same things is only partial redundancy. In order to be fully redundant the two systems must be independent, built by different teams and have different technology. We can't afford to stumble on this. It seems to me that Mirek has already solved this problem for us.
Lou McFadin W5DID w5did@mac.com
On Oct 3, 2006, at 4:24 AM, Howard Long wrote:
Hi Matt
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
The point is that the commitment to have an analog backup had already been made. You may not have been party to the discussions around July 2005, but believe me they were just as vociferous as the recent disinformation floating about regarding S1. It is important to consider seriously what the members want - after all, they are paying for it, and there would be no AMSAT-NA or Eagle without them.
Seriously, if we have 2 SDX's, aren't they backups for each other?
Well, this raises an interesting point. My understanding from discussions over this side of the pond with Martin Sweeting is that redundancy only truly exists if you also have two different designs: having two devices of the same design does not resolve design flaws. In SSTL's case, Martin's view is that the majority of failures are now down to design, with a minority environmental (radiation, thermal, outgassing etc etc).
73, Howard G6LVB
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Louis McFadin wrote:
Martin is exactly right. Two of the same things is only partial redundancy. In order to be fully redundant the two systems must be independent, built by different teams and have different technology. We can't afford to stumble on this. It seems to me that Mirek has already solved this problem for us.
Lou McFadin
W5DID
w5did@mac.com mailto:w5did@mac.com
I do not agree that Mirek has solved this for us. The L band receiver is not an analog transponder. There needs to be an interface block between the receiver and the follow on transmitter to provide gain control, impedance matching, buffering, etc.
Mirek COULD solve this should he choose to.
Bob
Louis McFadin wrote:
Martin is exactly right. Two of the same things is only partial redundancy. In order to be fully redundant the two systems must be independent, built by different teams and have different technology.
Ok, then the ACP is the redundant backup for the pair of SDX's and vice versa.
Matt
On Tue, 2006-10-03 at 12:25 -0400, Louis McFadin wrote:
Martin is exactly right. Two of the same things is only partial redundancy. In order to be fully redundant the two systems must be independent, built by different teams and have different technology.
I think we should also remind ourselves that redundancy exists at different levels of granularity. For example, we might build more than one unit of Eagle, P3E provides redundancy with Eagle and vice versa, etc.
It's important that we think through the various possible failure modes and "do the right thing" to mitigate each, and maximize the probability that we're left with something useful even if various important parts fail. However, if we insist on a mil-aero level of absolute redundancy at too fine a level of granularity on a single satellite, we run a real risk of specifying something we can't afford to build or fly. Our prior and future success hinge heavily, I think, on our collective ability to be smart about these sorts of tradeoffs.
Howard, if I were the one giving your talk this weekend I'd be inclined to assert that reliability and redundancy are being considered at every level of the Eagle project, but I wouldn't be inclined to commit us to any particular technology path like promising an analog transponder as backup.
73 - Bdale, KB0G
Bdale: Your asssessment of how we should do the risk analysis is EXACTLY the level I seek -- we're not doing the space shuttle replacement. ' In my mind (?) when we have a few more systems and interfaces defined, I'd like to convene a team who's SOLE responsibility would be to do the "what if" on each failure, and somehow rank risk, probability, and consequences. Then they'd suggest mitigations that aren't already there.
Thanks & 73, Jim wb4gcs@amsat.org
Bdale Garbee wrote:
On Tue, 2006-10-03 at 12:25 -0400, Louis McFadin wrote:
Martin is exactly right. Two of the same things is only partial redundancy. In order to be fully redundant the two systems must be independent, built by different teams and have different technology.
I think we should also remind ourselves that redundancy exists at different levels of granularity. For example, we might build more than one unit of Eagle, P3E provides redundancy with Eagle and vice versa, etc.
It's important that we think through the various possible failure modes and "do the right thing" to mitigate each, and maximize the probability that we're left with something useful even if various important parts fail. However, if we insist on a mil-aero level of absolute redundancy at too fine a level of granularity on a single satellite, we run a real risk of specifying something we can't afford to build or fly. Our prior and future success hinge heavily, I think, on our collective ability to be smart about these sorts of tradeoffs.
Howard, if I were the one giving your talk this weekend I'd be inclined to assert that reliability and redundancy are being considered at every level of the Eagle project, but I wouldn't be inclined to commit us to any particular technology path like promising an analog transponder as backup.
73 - Bdale, KB0G
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Hello Matt!
I know that the current documented thinking is to have two SDX's and no analog transponder, however we did say we'd bring up for discussion at a telecon prior to the Symposium. I know that this was brought up in July 2005 where it was clear that the answer to the members was yes, we'd have a backup _analog_ linear transponder.
Why stop there? We should have a backup to the analog transponder in the form of a CW transponder. Then we can back that up with a spark transponder. Finally, we should have some sort of semaphores or smoke signals to back that up.
We tried smoke signals with AO-40. The results weren't pretty...
We've flown analog transpodners for years, and complained about them for at least as many years. If we have an analog backup, we won' t have to write code to simulate the poor performance (AGC capture, etc.), we can let people experience it for themselves :-)
Seriously, if we have 2 SDX's, aren't they backups for each other?
It's a concern about radiation, I think, more than anything else. Personally, I'm happy to leave it out. The SDX concept is no longer just a power point presentation and associated hand waving, it is working demo-ware, and improving with time.
Lyle
participants (11)
-
Bdale Garbee -
Grant Hodgson -
Howard Long -
Howard Long -
Jim Sanford -
John B. Stephensen -
Lee McLamb -
Louis McFadin -
Lyle Johnson -
Matt Ettus -
Robert McGwier