Thread: About 37xx converted to 34xx satellites reception accuracy

Originally Posted by disneysw

I believe the issue may be temperature calibration........ On my unit it takes a few days, but the accuracy and lock on speed if the unit is left powered on.

Originally Posted by angelozip

disneysw, very interesting your information, if you are sure that it is true. I honestly never knew that the garmin gps had temperature sensors inside......

Nor was I aware of that, and it is very interesting. However, it doesn't really explain the accuracy in my 37>34 remaining the same (apparently). I've done multiple conversions/reconversions and the lock time and accuracy seems the same always. Since the [Only registered and activated users can see links. ] (about 2 weeks ago), before i ran the side-by-side with the 3590 i had used it as 3490 for only about an hour previously to that test and it was getting 3-8m and fast lock straight away.... Logically it seems to me that the different experiences with 37>34 has to be elsewhere. We're all using the same FW, so what's that leave .. hardware differences? I'll get another 3790 by the weekend, have a play around with it and compare to my current device.

disneysw, very interesting your information, if you are sure that it is true. I honestly never knew that the garmin gps had temperature sensors inside. Anyway is information to be considered. Soon when I installed the 3490 firmware on my 3790 I noticed a great inaccuracy, much below any other model I tested, but this does months, in a period in which virtually no used it ( I preferred the 1390 for ease of visualization of traffic). There is little more than a month I started to using it daily and noticed that the accuracy is greatly improved, reaching sometimes values ​​very acceptable of 5-6 meters in more favorable conditions. Anyway, the comparative test I did recently with a 2595, both side by side, the 3790 modified lost heavily in terms of precision, since 2595 showed an average accuracy of 3 meters and 3790 mod of 9-11 meters in the open, leading me to conclude that their efficiency is actually lower. The question is how the 3790 original would behave , has I not yet had the time or guts to return to the original firmware to check. I even considered what could be any, or some, files originating from 3790 that were hampering precise tuning and even differences between the drivers, since both have different chipsets. In the next days I will do more comparative tests with users who have 3790 in my country, maybe can arrive at some conclusion.

Originally Posted by angelozip

disneysw, very interesting your information, if you are sure that it is true. I honestly never knew that the garmin gps had temperature sensors inside..

I found this:

we (Garmin Engineering) use a relatively cheap oscillator, and
a lot of software finesse. Oscillator compensation data is
stored in the unit as a table based upon temperature. When the
unit locks on, it calculates the unit's oscillator error and
enters a correction factor into this table based upon current
internal temperature. In this fashion the unit is CONSTANTLY
"learning" and fine tuning itself. We burn these units in when
new to calculate and store calibration constants across the
entire rated temperature range of the product.

source:

Code:

Please Login or Register to see the links

Originally Posted by voyager56

I think a car "jumping" also depends on accuracy. This means that car "jumping" occurs when accuracy is poor.

Car jumping as I know happens on real 34xx too , interesting is once I loaded CN South Africa in my Nuvi and car wasnt jumping at all, may be dasilvarsa can make a test and tell us if it's true or if it has been just episodic in my 37xx/34xx.
Btw I agree with [Only registered and activated users can see links. ] and [Only registered and activated users can see links. ]

disneysw, really very interesting this information, I put two gps screen information about Hadware, a GPS was coverted 37xx to 34xx, 34xx another is original, here we have an indication of three temperature, see the results:
34xxc Temp1 36 ° C Temp 2 Temp 45 ° C 3 43 ° C converted
34xxo Temp1 32 ° C Temp 2 Temp 39 ° C 3 43 ° C original

I believe that the error in this table is accurate temperature

No Jumping with CN Southern Africa.
I think it's because it's a small Map only 80.6 MB

I think it's conclusive that the jumping is a factor of the lock to road feature and not a result of satellite (in)accuracy. My 3490 doesn't do it at all with any map large or small. Giomen didn't respond to my question following his post (the first link by Caty):

Originally Posted by Neil

@giomen
Are you saying that the 'lock on road' (snap to road) function in 34xx is less efficient or less sensitive than other Nuvis?

I then forgot about it, but i can't recall anyone reporting this problem when their 34xx is in 'off road' mode which disables the magnetic lock on road feature. It's of no real consequence whether the lock function is 'less efficient' or 'less sensitive' as i asked him, but if users reporting jumping in normal routing (simulated or real) don't experience it in off road mode then it's definitely not related to the accuracy issue in 34xx.

Got hold of a 3790 to test with the 3490 convert. Did a 40min drive with them side by side. Cloud was full overcast. Stationary, the stabilised 2D accuracy was close (only generally a metre discrepancy) but moving was remarkably different. Couldn't do any camera shots while on the move as i was by myself, but i noticed that the readings of the 34 were up to double that of the 37 at times. e.g. 37 at 3m / 34 at 6m and 37 at 4m / 34 at 8m were the worst comparative 'at the same time' results. The overall individual best/worst readings that i saw: 37 was 3/6m, and 34 was 4/10m. Of even greater discrepancy was the elevation, at times while moving there was up to 30m difference, but less when stationary. As i was in familiar territory, i have to say that the 37's elevation reading were certainly the more accurate of the two. Compare the satellite locks in the 2 pics. They are not the same, maybe this is the answer as to accuracy problems. It seems to me on these results that 34 is much worse maintaining both 2D and 3D accuracy especially when moving. Just shot my own 'urban myth' hypothesis to hell! Two pic's are at the start and finish of the trip: