Jollydirtyoldwoman
|
Thanks for registering 208.127.11.96! I've moved our discussion from that IP's talk page to here. - Davandron | Talk 04:54, 16 March 2007 (UTC)
I want to thank you for your addition to the GPS article, having someone familiar with the hardware is important. However, right now your addition is not sourced, and stating you are the designer is not sufficient (else it would be considered original research. It would be appreciated if you could find this ability of the hardware explained in a presentation or scholarly article. Thanks, and I look forward to more information from you! - Davandron | Talk 18:02, 14 March 2007 (UTC)
======
editThat might be a tad difficult. I believe that most of the subject is quite firmly classified - or at least I have not been debriefed on it. The comment I made is simply one that a person with a high school background and a need to correct an oscilator could figure out with a stick while scrawling diagrams in the dirt. {^_-} The only perhaps missing piece of data would be the fact that the synthesizer would have to have steps small enough to make this pracitcal. But that becomes obvious to someone who has performed synthesizer design with rigorous noise analysis. The step size must be exceptionally small so that it can "hide" in the noise.
I can remember one of the meetings about it that was great fun for me. (I am often rather unrepentant and irreverant towards entrenched authority. {^_-}) Anyway we were discussing making step sizes even smaller. I was getting annoyed - the tradeoff between parts, reliability, noise, and step size simply didn't allow me to go to a smaller step size without blowing some other parameter. So finally I simply broached the subject of differential GPS, which another colleague had mentioned to me recently. It was fun watching that poor Air Force captain I was briefing try NOT to listen to his project go up in theoretical smoke. It was rather funny. (I also do NOT have a soft spot in my heart for Colonel Parkinson.... Some of his dealings with the group at Rockwell International I worked with ended up being what *I* believe to be unethical - he had as consultants at our design reviews for the test receivers people from a company that became our competitor for the third test receiver design. They produced untraceable trash, by the way.)
However, if you think carefully about what a frequency synthesizer with roughly 2 micro Hz steps at 10.23 MHz can do by way of correcting the frequency of a Cs or Rb frequency standard you could see this secondary utility for the denial of accuracy facility on GPS.
I believe the prior means of adjustment for the Rb standards involved tweaking the magnetic fields in the Rb standards. The Cs standards didn't have such a tweak to my knowledge. The FSDU, Frequency Synthesizer and Distribution Unit is a more stable and finer grained adjustment.
One of the chief requirements on it was that the synthesizer did not degrade the GPS signal in such a manner that it was detectable outside. This small step size and a fairly exotic direct frequency synthesizer inside a feedback loop involving a a couple digital dividers and a couple frequency mixers achieved this end, as long as enough shielding was used to keep the oscillator blow by down below about -100 dB relative to the signal. (The goal I set the actual designer was -120dB, which is in practical terms impossible. The absolute minimum that could be tolerated was between -100dB and -80dB depending on the frequency excursion made.)
Annecdotally I was also involved in RF design both the original pre-launch test receivers, their immediate successors, and in software design for the Navigation Data Unit's prelaunch and acceptance test. Those were fun days.
My basic experience with GPS is with the Satellite segment. Although later on at Magnavox I did some (very) small pieces of work with their various GPS receivers. I had basically moved from RF design to software design at Magnavox.
{^_^} username now jollydirtyoldwoman - aka jdow for short but that was already taken. Joanne — Preceding unsigned comment added by Jollydirtyoldwoman (talk • contribs) 00:05, 16 March 2007
- Wow, thats a lot! 8^)
- As I said, I believe you are being truthful and while what you said makes perfect sense, the wikipedia guidelines tell us to consolidate information published by reputable sources, and to avoid original research. The inclusion of the very specific number will raise a lot of eyebrows. That said, lets keep it and try to source it (instead of deletion until its sourced).
- Do you know if that capacity (for clock offset correction) has been utilized in the design? I would encourage you to do some searching of journals and even the web to see if its been mentioned. As far as I've looked, I never saw anything about how the SA was applied.
- In fact, I figured it was a highly tolerant activity; you're adding highly randomized jitter to the clock! 8^) But then again, I was thinking analog noise being mixed in. Thats some fun new knowledge. Hadn't thought about someone sniffing the SA signal directly; makes me wonder how many high gain setups are pointed at all of our various satellites.
- Having your kind of knowledge could go a long way to improving the article. At the least, you can flag things that are probably untrue when they enter the article (even if you can't say why you know that to be the case). At the better end, you can use insider terms / bragging friends to locate journals and articles which contain published information. Its amazing what is out there already; Tom Clancy made a living out of putting two and two together.
- Thanks again Joanne! - Davandron | Talk 04:54, 16 March 2007 (UTC)
- (BTW, placing four tildas (~~~~) after your post, inserts a signature with the date and a link to your name. And colons indent text, to help isolate different editors comments. Let me know if you have any wikipedia questions!)
Regarding the clock correction - that was noted at the time the design documents were promulgated. I can quite unequivocally say it has been used. It was an intended side use. The Air Force may have done some stupid things at the time. But you can bet this was not missed. I mentioned the idea and the person who heard me talking declared this was intended, particularly because the existing correction method was too coarse and imprecise.
The added noise was structured to look very much like natural oscillator noise. I was fed some nice noise plots (Allan Variance and "just plain old spectrums") to work from and told "don't degrade the clock." Being a purist I observed (politely for me), "There's no way that can be done." (I deleted the explitive that would be expected of someone told to "get on this pogo-stick and jump over the Empire State Building over yonder." {^_-}) And really, NO degradation at all is not feasible. But a degradation of less than 1dB is quite possible if the synthesizer is kept clean. I had left Rockwell by the time it was being built. It required very expensive construction. "Lots of grounds," was my sentence fragment mantra. The fellow who designed the actual circuit and made it work put in "lots of grounds in a hogged out aluminum chassis." Production tried to remove some of the ground bonding from the ceramic PC card - printed circuit card in those days - and never got it working that way. 100dB let alone the target 120dB of input to output rejection when the two are essentially on the same frequency is insanely difficult to achieve. (Add emphasis to insanely. It's appropriate.)
With that background the injected noise is shaped to look just like normal oscillator noise and hide underneath it. I doubt the full range of the synthesizer (over 1Hz) has been used as that is a rather extreme and easily detectable error. The general idea was to prevent the commies from sending their equivalent of a cruise missile to the US and have it intentionally land in "Jimmy Carter's toity." (He being President at the time.)
In light of article accuracy issues I didn't really note anything grossly inaccurate. And I noted the significant improvements added since hte block II stuff I worked on. The 1381 GHz frequency is not known. I helped the designer of that puppy fix an modulator bug he had. He fell victim to a phenomenon I've fallen victem to myself. You learn a new technique and just HAVE to try it out. So of course, you try it out on the first likely project to come around. Smart companies take the (usually more or less junior and sometimes even fairly senior) aside, give him a little money, and order him to build it and characterize it end to end. Jim had not had that chance. So he had a modulator that allowed the modulation signal, a really nice squarewave sort of thing with sharp edges, leak out the RF ports. I showed him how to at least partially balance that out so he'd meet spec.
There were a lot of fun challenges building that satellite. I worked for a group at Rockwell who did both black and red world projects. And I must say it was the over all smartest group I ever worked for. Towards the end of life of that department we tended to pick up projects for GPS that other companies had failed with miserably. We had a "reputation" in that regard. I'm not the only zany that was doing things like calibrating the test receiver's GPS modulator to 1/4 the error tolerance of the satellite in half an hour from setup to teardown while ITT was spending a whole day to calibrate theirs. This sort of stuff was noticed. So our group became one heck of a pickup group. (As a result of work there I know, for a fact, that you can with late 70's technology build a very slow data rate satcom terminal that would make the average slim line cell phone look lumpy big. {^_-} The key is slow data rate and a crazed modulator. Sometimes ratty signals don't really matter.)
Most of my GPS went away when I left Rockwell (for a massive pay raise and a miserable job) in 1982. I learned later that the department self-destructed about two years later. It has been coming for awhile. Sam Costanza, who started it, passed away from pancreatic cancer, I believe it was, and the infighting to take over this wonderful department proceeded to destroy it. Before GPS I did some fast frequency hopping work exploring its utility for military communications. As a result of what I learned there the current slow hopping SINCGARS nonsense toys are not safe anywhere near me. Fast hoppers have the speed of light on their side for repeater jamming. Slow hoppers are HORRIDLY vulnerable to relatively close in beach ball jammers that are mostly battery with a tiny bit of electronics. Fast hoppers force the beach ball to be very much closer.)
I'll leave you with one of my favorite observations, oscillators don't really oscillate. If you have an RF background you might figure that one out. We'll see. I do like to approach things from a skew angle. I "think" but do not know that a stray comment I made to the fellow at Seal Beach my group was working for on some GPS stuff was worrying one day about "upsets" in the nav data unit from stray gamma rays or the like. I made an off hand off the wall comment, "So reboot the silly thing periodically." I learned through a side door path this became the accepted technique. It IS a viable alternative for a ROM based computer.