"Gary E. Miller" <[email protected]> writes: > Yo Greg! > > On Sun, 09 Nov 2025 20:01:59 -0500 > Greg Troxel <[email protected]> wrote: > >> EPX/EPY/EPC are 10/10/20 m. That's bad, but poor signals indoors and >> seems reoasnable. >> EPS/EPC are 83 km/h and 167 km/h. That's bonkers. > > [...] > > These are from the GPS, unless the GPS does not provide them. The > algorithms are specified in the ICDs, and do not reflect current reality.
The question is then whether the receiver is really outputting values that are so high, or if there is a problem in translation. >> So, it would be good if people could look at their EPS/EPC and see if >> they are sane compared to the EPX/EPY/EPV. > > The equations to derive one from the others are standard. Which means > the receiver can't add any black magic to get them to correlate to > reality. I am saying that the relationship between EPS/C and EPX/Y/V does not appear to match the expected standard conversion. The LLh position error estimates are plausible. Yes, I realize that 50%, 1 sigma, 2 sigma is not specified and unclear but very broad brush that's a factor of 2+, not a factor of 50 or 100. > This should be an FAQ. For some reason people think they mean something. Reasonable people think they mean *something* but they aren't quite sure what. My receiver is currently tracking 6 GPS satellites, and nothing else. Reported SNR 16-36 dB. One expects poor performance. If I eyeball EP X Y V, and eyeball RMS them, that plausibly matches pACC from the XYZ values, this minute around 21m. Whether that's 1 sigma or 2 sigma I don't really care right now. If I look at velocities in XYZ, they are all several cm/s, sometimes tens of cm/s (the receiver is on a desk). The vAcc values bounces around, and I'm going to say my visual average is about 0.8 m/s. Given ~20m position uncertainty, ~1 m/s velocity uncertainty seems entirely reasonable. If I translate 1 m/s to km/h, I get 3.6 km/h. This is walking speed. That matches my over-time experience of velocity errors under very poor fix conditions. But, I'm seeing EPS of 140 km/h. Which is vastly more, and not just driving speed, but speeding ticket speed (except maybe on rural interstates). I would expect vAcc = sqrt ( EPS^2 + EPC^2 ). While it's theoretically possible that ublox intends vAcc to be 50% and EPS to be 6 sigma, that doesn't make any sense. Assuming that as an explanation is unreasonable. It's possible that this particular receiver has buggy firmware, and it's possible something else is going on. Hence, I was suggesting that other people look at their output and see if EPS/EPC are reasonably related to vAcc. If this is just my receiver, it's not a gpsd problem. If it's a pattern, then I think we should look harder. I'll try a few other devices.
