Well... this took longer to test than planned. My first test on a M-class sorbitol motor Cato'd right off the pad and destroyed all electronics. I rebuilt another board and was back in two weeks with a commercial CTI L motor and that case burned through one second off the pad, destroying the rocket, but thankfully the board was repairable.
So, last weekend I finally got a good test. I flew the new GPS to 11K feet on a commercial K motor in a 3" rocket. In the previous tests I had as many as 24 satellites fixed on the pad, but on this flight I only had 14 locked. The AvBay for this test was smaller, shorter, and I had added a camera and more wiring butted right up against the helical antenna blocking a lot of reception.
Overall, the GPS performed much better than the M9 or M8, but it still dropped out. At T+ 4 seconds the satellites dropped below 3 and it lost lock (going about 790mph -- just a hair faster than Mach 1). It dropped for 2.5 seconds and then regained, but then dropped again for another 3.5 seconds. At twelve seconds (about half way to apogee) it had a good lock and stayed locked the duration of the flight. The M9 would have dropped early and then regained after apogee (usually). Given the sat data and the known velocity, I don't think it dropped due to cocom user limits, instead likely doppler issues.
There were a few interesting anomalies in the data. On launch the reported altitude went negative -- even with 7-10 sats, until it regained lock at twelve seconds and started reporting proper altitude. I saw the same thing in the cato launches, albeit for only 2-3 samples. I have the GPS enabled for 2D mode, since Lat/Lon are far more important for locating the rocket vs. waiting for a confident 3D lock.
Also, in the GNS NMEA data there is a "pos Mode" field. I am using six different constellations, so there are six indicators (A = valid, N = no fix). It seems I've got a valid fix with GPS and Galileo, but none of the others. Also, it doesn't look like Ublox is doing much "satellite fusion". In other words, if you had twelve sats, 2 from each constellation, you would have any valid fixes. You need at least three from any constellation. I need to research to see if that is true, but it isn't clear how they are fusing data from all the systems.
Again, overall this is a huge improvement. I'll keep flying it and keep collecting data. Below is my raw log (launch through apogee) for this flight...
