Poor GNSS accuracy? [Bug 8]

[Gadwin]

I made a test by wrapping 67 in aluminum foil, waiting for it to give a signal about the loss of satellites and walked away a dozen or so meters. I removed the silver for the poor from 67 and proceeded perpendicular to the earlier route. 67 was walking parallel with me, a dozen or so meters away.
66sr the silver wrap test passed excellently did not fall for it.

I did a few more tests and each time the result is the same.
When 67 does not see satellites, when the constellation returns, it does not refresh its position in real time and starts its journey from the last position.

I'm curious, have you also noticed this problem? Am I the only one seeing this?

I did today an extended hiking to very small mountains and I drove via tramways and subways. I tracked everything and compared with my older tracks, which I have several of those. It is my usual weekend trip. I noticed indeed, that those 3m at the GNSS satellites page is not just a visual error. I was also at my usual spot on top of a mountain and the deviation of the altitude was 3m. That is not severe, but neverless the 66sr was always spot on and I was several times there. Also it is a really small mountain. Let be honest, if the 66sr never existed, we couldn't compare to it. But it did exist and we know the 67 should and hopefully could perform better.

But beside all the harsh critic, I did not experience what you did with the faulty refresh. When my subway came overground the 67 immediately picked up the signal. I did not look at the device during the time, it was in the top pocket of my backpack, but I can see it clearly on the track, that it did pick up the signal immediately after caming out from the underground and it was on the "right spot". I don't know if it would show somethingelse if I looked at the device at the moment, but the track does not show your experience.

It could be a bug in the refreshment of the map. I shall look directly at the device next time. Did you also record a track during your test?

[Nail]

It could be a bug in the refreshment of the map. I shall look directly at the device next time. Did you also record a track during your test?

Yes, I have a recorded test.
Is your subway journey also recorded?
If so, until when did GPS place points underground? Yesterday I locked myself in the toilet, it would seem, without access to the GPS signal. Around the windowless wall and reinforced concrete ceiling, a 67 picked up the signal. It happens that when I drive a car in a tunnel, the GPS does not lose the signal.
The best way to cut off the signal turned out to be aluminum foil.

[Gadwin]

When the subway was underground it was not recorded, it stopped recording when signal was lost. That happens underground immediately . When the subway was overground it started recording again, when signal was locked again.

When the tramway/subway drove under a bridge quickly for a short time, it didn't lost signal and recorded normally. But when the tramway/subway stood under the bridge for some minutes, then my cursor began to dance and it was still recording, so there was like a spiderweb on the track. That happens when the signal is poor, but not completely lost.

That is also the case with 66sr. So about picking and losing signal and how the cursor behaves, when signal is poor, it behaves normal, if you compare it with the 66sr.

[Gadwin]

But just to be clear, the GNSS accuracy was always around 3m with the 67 today, where normally the 66sr would have stable 1.8m. It was an extended area where I was several times with the 66sr and never experienced a loss of the GNSS accuracy, even when it was heavy raining, dense forest or clear sky at top of the mountain. It is a really diverse area.

The 67 only showed for a brief period at best 1.8m, but that was really for some seconds, then it went up again to 2.6m and 3m.

[JungleJim]

I did a more extensive test this weekend, a hike of around 10km through forested area. Not really a dense forest, with few leaves on the trees yet.

What I noticed was that my 67 gets a fix quite fast after turning it on (with current CPE). I have seen no drifting like Nail described, but this was outside and with clear visibility of the sky.

During the walk the best accuracy was between 2 and 3m. Never saw less than 2m. I kept the 67 in my coat pocket, and every time I took it out, accuracy would be at 4m and then improve to around 2.6m or a bit less. Maybe antenna position? The track log does not show any large deviations, even when standing still.

I don't have my 66sr anymore so I cannot do a direct comparison by recording the same hike with both devices and then compare track logs. That would be the best way to view any differences in accuracy, I guess? But from memory the 66sr showed <2m accuracy more often.

So for me it seems with the 67 the accuracy is still good (for my use cases), but 66sr was somewhat better. So hopefully Garmin will improve GPS accuracy with future firmware updates to make it on-par with (or even better than) the 66sr.

[Gadwin]

I had contact today with US Garmin and they sent me a link

Spoiler

https://support.garmin.com/en-US/?faq=Z ... textPage=1

, where it is stated, that their outdoor handheld accuracy is 3m (10ft) for around of 95% the time. But one support employee said to me, that he also thinks, that the 67 accuracy should be at least like the 66sr has. Therefore he opened a ticket and sent it to the engineer team. I really hope the engineer team will change it, because otherwise it is really a downgrade. I guess they made the 66sr better than they wanted, huh?

[Nail]

What I noticed was that my 67 gets a fix quite fast after turning it on (with current CPE). I have seen no drifting like Nail described, but this was outside and with clear visibility of the sky.

That's right, this problem occurs indoors, the vehicle drifts far away and stays there, then when you go outside it starts from last position and it takes too long. You will enter the shelter for a meal, and after leaving, track 67 will follow another ridge instead of behind you. I see a problem here.

Today I calibrated the compass in GPSMAP 67 and 66sr. 67 has more correct indications. I don't know if it has any effect, but it seems to be more accurate after this treatment. I also did a test, I carried both side by side so it should be the same. In my opinion, 67 was more accurate, although it left less points in than 66sr. I haven't looked at the accuracy of the satellites, no matter what they show, the trace will be important. You can also see that despite the shift, the traces go evenly. The area was quite open, low buildings, light clouds, clouds high.

Spoiler

67vs66sr7.jpg

[Gadwin]

So it seems, when my satellite page indicates around 2.1m accuracy, then I don't see almost no difference with the 66sr on the track. But there is clearly a difference, when the satellite page shows an accuracy at 3m.
The biggest anomaly you can see in the picture below. That is showing me in a grocery store. The middle pictures shows me when I was finished with shopping and turned my 66sr on. The very right picture shows my whole route in the grocery shop with the 66sr. Of course it is not perfect, but at least it shows almost the real position and for sure how I went through the shop.

But the very left picture with the 67 it is total off. It should look like the very right picture, I always walk the same route in the shop, but even the entrance and exit are not correct.

I did it very often with the 66sr and it almost looks the same. I just did once with the 67, so I have to see it for several times more, but I never experienced this behavior with the 66sr.

I mean, ok, it is in a building, but still, the 66sr can do it, so at least the 67 should also can do it.

Spoiler

accuracy.png

[mimichris]

I don't think it's a good comparison in a room where the GNSS signal is very bad, the best would be to make a comparison in a dense forest, with beech and fir trees for example because it's a situation that the we meet during our hikes.

[Nail]

I did it very often with the 66sr and it almost looks the same. I just did once with the 67, so I have to see it for several times more, but I never experienced this behavior with the 66sr.

This is probably the problem I'm talking about. It is not important what happens inside the building, but what happens immediately after leaving. I don't know where the entrance/exit is? Looks like 67 needed more time to position itself

Spoiler

67vs66srxyz.jpg

The problem is when we have a device running with poor satellite coverage. After that, he can't come back quickly.