Fix then float, same environment


Yesterday I was doing some testing with a rail based vehicle. We did 3 test runs. 2 of them (1 and 3) came back almost entirely “fix” with good ratios when processing in Emlid Studio. The second one came back as float. The receiver was indicating “fix” the whole time. Ratio is around 1.0 mostly.

What could be causing that?

I will send the files to in a Google Drive link.

Thank you

If you can share your raw-files with the community you’ll probably get a slightly faster and broader response and analysis.

@wizprod good idea, here they are:

The 1838 rover file is the one that is float. The 1958 is fix. Same area, same trajectory, within an hour of each other.


I get the same behavior, though only 8% fixed in EzSurv on the fix-obs.

The files incl the base have a 5 min on/off oscillation I have never seen before in obs-files. Do you have some equipment that turns on for 5 mins, then turns off for 5 min?

Check the SNR’s here, for all 3 obs:

This is interesting. We do power the RS2 through the expansion cable, using a generator that also powers the electric powertrain. I will look into perhaps powering it via USB instead.

I will check with my colleagues what could be causing this.

Just saw the base is also affected. Hmm. Pretty stumped about this.

Do you have some pictures of the setup ?

So both the rover and base get this signal drop so strange? Something must be competing in close proximity. Also the repeating time interval suggests its automated.

Any chance your near a big dirty RF emitter like a Weather Radar station or near a Defence Base?

Nothing that I am aware of, checking with the engineers on the project.

I still don’t understand why 2/3 runs were “fix”, 1 was float, if all 3 had the interference throughout

The interference is strange also, so the lower range I think is normal. When the signal spikes its looks like it is mostly multi path error.

The interference also lines up with utc time.

Right now I can only suggest aliens, but the bigger question is what has synchronized 5min intervals.

1 Like

Ha. At this point aliens does seem like the most plausible cause.

Lines up? Not as I see it, if we are talking about hitting minute-marks ?

Have to disagree on that. Below is a screenshot of an RS2 placed in fair location, similar obstime, same view as above, L2 on GPS. Notice the snr goes to 40-41 dbhz , which is the max the RS2 seems to be delivering.

So I would argue that when the SNR drops, that’s where problem start, and when it goes up again, that is where the problem temporarily ends.
The multipath is however much lower than I have in ANY observation from an RS2 ever. That sort of multipath average is usually only seen on choke ring antennas. At the same time, the SNR is lower, so it really doesn’t make sense.

Likely because the one that fixed somewhat had a longer observation time. 10 minutes is on the low side for postprocessing, especially if the obs quality isn’t awesome.

Oh I see. This was testing. In reality, we will be doing stretches of over 100 minutes.

I just see this repeating square wave that lines up with utc.

5on / 5off

Ah yes, on the base! Right.

So then even wierder that 2 rover logs are not aligned to the same pattern.
I guess that means that whatever it is, it is not airborne, but something coming into the receiver via a wire, or at least low enough rf power to not spread to both units.

Hi guys!

Francis, I’ve processed the files you shared. With the 1838 raw data log, I also got just 10% of Fix with the Forward filter type. But when I changed the filter type to Combined, I achieved a 100% Fix.

I think the difference occurs as the 1838 raw data log is slightly worse than 1958. You can see that the 1958.21O file has signals with better SNR (more green lines) than 1838.21O. The Forward filter type is usually more sensitive for such differences.

However, PPK allows playing with the settings such as filter type or SNR/elevation masks values. So sometimes it allows improving the results significantly, as in your case.

At the same time, the L2 SNR changing every 5 minutes is indeed weird. So we’d like to investigate it. I’ve contacted you via email as further troubleshooting requires sensitive information from you.


Hi FPelletier

with EZSurv you can look at the raw observations (for each signal components). That being said, we clearly see that the problem is on L2 carrier phase (signal recorded is really poor). However, signal for L1 carrier is excellent.

With EZSurv you can also ignore L2 when processing (Edit/Process parameters/ check the box “for dual frequency data, use only L1 frequency data”. This will lead you to 100% fixed results for the rover file that last +30 minutes.
The other rover file (10 min) is too short to allow fixed results.

This topic was automatically closed 100 days after the last reply. New replies are no longer allowed.