LGT-92 low GPS accuracy

We have tested the gps in a field away from tall building and while driving in the street. We faced the low GPS accuracy issue. lgt-92 give up to 100m wrong location. Is anyone have faced the same issue ?

No. What about weather?

Weather was good and i checked number of satellite it was 8. I am using it with US915.
What accuracy are you seeing ?

Also what firmware version are you using ? this what i am getting for a static position of lgt-92

As every gnss fix with the lgt92 is more or less a cold fix (afaik), the accuracy can sometimes be bad due to this energy save mode.
Maybe there is a way that the gnss module could work in hot fix mode to decrease time to fix and increase accuracy with a firmware update.

From that map, some of the position fixes may well be close to 100m out, but the majority look much closer than that. Lots of high trees around too, they wont help.

If the GPS is being run in hot fix mode then running it for longer than the time when it first aquires a fix should improve accuracy at the expense of power consumption, but thats really an issue for the company who are selling the tracker to addess.

If the GPS is run in cold fix mode, with no backup of satellite data, and postition is reported as soon as it gets a fix, then the location variations seem normal to me. Again running the GPS for longer would improve accuracy.

Also difficult to tell from picture but how tall is the building at bottom? If tall you may be in satellite shadow at times and also may suffer from deflection/reflection issues increasing path lengths and peturbing resuts. positioning looks to be quantised so assume your are using a payload format such as LPP?

I have tested lgt-92 in a hot fix with firmware 1.4 but it still gives me more than 100 meters error. I also run it for a longer time but it did not improve the accuracy. In parallel I have another GPS u-blox neo-6m. I am using it with arduino. It is giving me 2 meters accuracy in the same parking lot.
I have checked the firmware how Dragino parsed a NMEA code to get latitude and longitude. I did not see any issue in the parsing equation. Do you guys think there is a bug in the firmware which is changing NMEA code randomly before it parse into latitude/longitude or its a hardware(GPS chip) issue? LGT-92 is using Quectel L70-RL.
@Jeff-UK building is almost 50 feet tall and yes it is LPP format with 11 bytes. u-blox gps is giving me 2 meters accuracy on the same location.


If your claiming that your LGT-92 tracker always has at least a 100m postion error, then perhaps its time to raise it with Dragino, its their product after all.

I have 2 lgt92 since 6month and I didnt see such high and continously offset.
How do you get the Position? Directly over USB/serial or over payload in TTN?
Works your payload decoder correct?

I would like to do the same testing.
How do you get the positions in the map from the lgt-92?

Normally you would print the co-ordinates to the serial monitor before the packet is sent.

Do Dragino provide a test program for their tracker that does this ?

the lgt-92 sends the position data over the usb/serial prot but I dont want to copy paste edit this - I thought there might be a smarter approach availabble.

I used two methods one using serial port and other through TTN cloud, I was receiving the data on the local server from there i am plotting it google maps.
I have one more question, did you guys try to open LGT-92 ? is it has a passive antenna or active ? the one i got has a passive antenna and i am thinking that might be the reason.

It uses a L70 GPS.

The L70 GPS works just fine on a small passive antenna.

What accuracy are you seeing ? Could you please perform an experiment. Take 30 mins reading and plot on maps ?
BTW i also have L70 chip with passive antenna.

Once I dropped the lgt-92 on the street the case was open and the film-antenna was ripped because it is glued at the inner side of the case. But I soldered a piece of wire on the platine and the tx-signal is now stronger as with the original antenna. Looks not that nice but works.IMG_20191019_153226_694_0~2_compress92_resize_5 IMG_20190615_180927_757_0~2_resize_84

1 Like

Typically within 10m, normal for a GPS really.

I own the LGT-92 since a few days. Yesterday and last night I made some experiments concerning reproducibility and accuracy of position. 95% of the reported positions were in a circle with a radius of 12m around the real position. The device had no totally free view to sky.

I did some logging.
One lgt-92 with 30sec intervall and the other with 6min.
The data where recorded by data storage integration.
In my european region the variance of the coords from lets say N48.1223 to N48.1225 causes 20m and E7.5515 to E7.5517 is 15m. And the values do have only 4 decimal.
Within 100 measurements of the 30sec-tracker the position didnt change, there were less then 5 points that had another value.
The other tracker works with 6min intervall (and maybe deepsleep and cold fix) and gives 9 measurements that vary in lat xx.xx86 to xx.xx89 and in long x.xx69 to x.xx77.
This is much more as the 30sec-tracker and maybe due to the cold fix and that the device sends the first fix instead wait for 2-3 positions to get a better result.
So the worse the reception is, the more differs the position when the tracker does a cold fix. Dragino should improve the firmware (or someone else) to keep the gnss-chip in hot fix mode between the intervalls or send the gnss-fix after a few seconds to generate a better position. Maybe its better to keep the chip in hot-mode instead looking in cold fix mode 2min for a position were the chip has 20dB less sensitivity.