Our lab fields numerous custom instruments on research aircraft. Flights last 8 hours or more. These instruments typically log their computer time, which drifts several seconds over an 8 hour flight. They often also log aircraft navigation data, which usually has GPS time included. We would like to improve on the time-keeping, in two related but different ways. If anyone has suggestions for dealing with either issue, they would be appreciated.
1) Accurate time stamps. The programs need to log a time stamp with each data record (typically once a second). As I said, they often log computer time and a GPS time, but the aircraft nav data may not always be available, and the computer time drifts badly.
2) Accurate loop timing. Ideally, we would have a source of 1 pulse-per-second timing that LabVIEW could lock to. We could route this into NI Daq boards for triggering, and probably even synchronize timed-loops in the software to this signal. Instruments that integrate signals over 1 second or some similar time interval in particular need to have an accurate clock either to control the integration time or to measure how long they really integrated.
I'm working on a possible solution based on a Garmin OEM GPS unit. Assuming you had a GPS signal lock, you would get 1Hz time stamps and a very stable 1pps TTL signal - exactly what we need. However, inside the aircraft, it is likely that we couldn't get a GPS lock, and we usually can't go mounting an external antenna. It would be great if these units would continue to give reasonable 1pps signals even after the lock is lost, but so far Garmin hasn't answered my request as to the accuracy of the pulses once they lose lock. They only spec the accuracy while the unit is locked.
Ideally, there would exist a small box (that is inexpensive!) the could lock to either WWV or GPS. Once it obtained a lock for a short period of time, it would use an accurate internal oscillator to continue to put out time stamps over ethernet or serial as well as an accurate TTL pulse at 1 pps. The experimenter could take this box outside the aircraft and let it obtain a lock, then bring it back in and hook it up to the instrument. I suppose an eithernet connection with a mechanism for synchronizing the box's timestamp to an NTP server would be a nice feature too, in case you couldn't get WWV or GPS. In that case, the internal oscillater wouldn't have been sync'ed to anything in particular, so it would have to have reasonable accuracy on its own for the 1pps pulse.
I've looked at the specs for NI boards, and unfortunately, they aren't encouraging for most of our use cases. The standard Daq boards don't have good enough specs on timers to supply a 1pps clock for 8 hours that is any better than the computer clock. There are PXI timing boards that could do it, but they run $3K, which is a bit much. Also, not all our computers can be PXI crates. Some of them could use PCI cards, but most are laptops. A USB, ethernet, or serial solution would be more general.
Any ideas?
Regards,
Dave
-------------------------------------------------------------
David Thomson Original Code Consulting
www.originalcode.com
National Instruments Alliance Program Member
Certified LabVIEW Architect
Certified Embedded Systems Developer
-------------------------------------------------------------
There are 10 kinds of people: those who understand binary, and those who don't.
