Anablog

It looks like adding an air traffic control signal to the latest GPS satellite has degraded the GPS accuracy by an order of magnitude. You folks all know these post-production signal integrity disasters. Fortunately for us most of us, our delivered products are not in geosynchronous orbit. From reading the article, it is hard to decide whether the new Boeing or this new Lockheed satellite is a bigger piece of junk. I don’t think the rocket scientists have gotten any worse, but I bet the bureaucrats have joined those at the DMV and Post Office in the national festival of incompetence that seems so prevalent in our public-sector projects. Finding out that you have fundamental signal integrity problems in production is why there are so many companies that are helping you evaluate what is going on in simulation and testing. Ansoft, Comsol, CST, Sonnet and dozens of others make field solvers and signal integrity software. This is why you should all attend the DesignCon conference here in Silicon Valley. I saw a great presentation from NEC about their power integrity products. That may be what is going on in the satellite. The L5 and GPS signals interact in the power systems as well as in the circuit boards and antennas. Mentor Graphics has been helping us with signal integrity for a decade; their Hyperlynx product works with layout packages like Orcad and Altium as well as Mentor’s PADS and big-iron layout products like Mentor’s and others.

I wonder if some of these aerospace and satellite projects are not failing because we as a society feel that they can do things on the cheap. Sure, these high-performance software packages cost a lot. If you get Hyperlinx with the full lossy transmission lines and everything else it can set you back almost $50 grand. But they have simpler packages that cost way less. And they can do power integrity, like NEC. Let me dig through my notes---- yeah, NEC told me that they can sell EMIStream for $30k, with a floating license for $38k, the expert version is $45k/57k for a fixed/floating license and their power integrity tool called PIStream is $20k additional. Remember, in this economy everything is negotiable. I met some really nice folks from NEC Japan at DesignCon but the local rep for the software is an outfit called TechDream. They may only rep the west coast but I am sure they can hook you up with the right NEC signal integrity software rep. The tech contact is Yoshi Fukawa and the sales guy is Eriko Yamato. NEC has a 90% market penetration in Japan with these tools so there may be some good stuff there.

This is not to slight all the other outfits doing signal and power integrity. Agilent, AWR (Microwave Office), I think SiSoft can help as well as dozens of others. The point is that it may sound like a lot of money for software but the great thing is that all these tools pushes the risk to the beginning of the project. If the simulation shows you have a problem 8 months ahead of introduction you have time and talent and money you can throw at the problem. If it is two weeks before production and your product fails FCC or CE testing, or like this satellite, just doesn’t work, well, that is misery no manager or engineer needs. And if your product launch is really a launch, well, $50 grand is pocket change to insure that you don’t have signal integrity problems in space..

[Update] Thanks to commenter SATguy for posting this link in his 6/24/2009 comment. The government explanation has a lot of bureaucratic doublespeak but the operative paragraph seems to be:

....the real problem was determined to be the way the L5 payload is integrated through the auxiliary payload port that causes reflected energy from the L1 and L2 signals to be reflected back into the broadcast antenna. This reflected signal causes a phase shift between the L1 and L2 signals, which affects how they are being formed and transmitted. The physical manifestation of the problem for users is that the resultant phase-center bias makes the satellite appear to be up to 150 meters closer to the Earth than is actually the case. This of course disrupts the ephemeris calculations of the satellite and makes the clock appear to also be in error.

Now I will withhold my apology to the rocket scientists as requested by SATguy. See, it is not the interference of the L5 signal that is causing the problem, if I understand it right, it is reflections of the GPS signal off the structure of the L5 module or antenna that makes the satellite look like it is 150 meters off. This is still a signal integrity problem and it still reflects badly on the rocket scientists. Did no one run a simulation of the GPS system with this L5 hardware added to the bird? And the "cure" is to just fudge the data so that it makes up for the satellite being 150 meters wrong. That is like correcting for a golf slice by facing the clubhouse when you swing. Every engineer knows it is system integration that is the hardest challenge and where these disasters happen. We have all drilled a hole or moved a component and had something go wrong. I would expect NASA or AFSPC or whoever to know that. At least they can fix it in the future. Also see this great article from the same site about how the whole GPS system is a risk.