Venus GPS logger battery power weirdness

I’ve had good results with the Venus GPS logger w/ SMA connector while running on external power (e.g., FT233RL 3.3v output connected to serial connector of GPS module). In this configuration we get valid log files saved to flash as expected. The problem occurs when we run the device off of a 1s LiPo (1100mAh SF pack). The recorded GPS position jumps around in 0.5km increments (e.g., in a ‘grid’) around the actual position. The problem seems to be unrelated to satellite reception because the exact same setup works when not running on battery power.

My initial guess was that this was a regulator ground noise issue, but I can’t believe that the FT232 output is any cleaner than the onboard regulator (MIC5205-3.3?) (haven’t checked on the scope yet.)

One weird thing I noticed was the (loaded) voltage I measured on the GPS board while running on USB power (e.g., the working configuration) was 3.15 volts. Under battery power, I measured exactly 3.3V from the regulator.

Any ideas?

very strange

the ft232 3.3V is rated at 50mA while the regulator is rated at 150mA.

do you see any difference in signal levels when you connect it to the GPS view software that comes with the gps

More debugging revealed no difference in performance between the voltage sources. The strange ‘grid’ results came from decompressing the downloaded log files using the ‘converters-> decompress’ command in the skytraq GPS viewer instead of the ‘log decompress function’. Of course, I only found this out after adding better bypass capacitors on the regulator, so I can’t say yet whether they had any effect at all… Basically only found out after making hardware fixes, THEN watching the user go through his workflow. Now I know to do that first!

I did see about 5mV p-p noise on the regulated voltage while running off the FT232. While running on the onboard MIC5205-3.3, I saw about 7mV of noise. Both of these are pretty bad - I have designed other boards using the same regulator and see less than 2mV of noise. Nonetheless it did not seem to negatively affect GPS solution fix, but might cause a problem for onboard ceramic antennas (the power noise will be injected into the LNA unless absorbed by the inductor). My designs have always used a 470pF cap on the bypass pin and 2.2uF tantalum on VOUT (standard low-noise configuration from MIC5205 datasheet) and haven’t had any problems yet.