Forums > Windsurfing   Gps and Speed talk

DIY GPS Logger

Reply
Created by decrepit 2 months ago, 26 Jul 2018
decrepit
WA, 8706 posts
4 Oct 2018 9:15PM
Thumbs Up

Yes, that's what it looks like, the time is between when it first turns on with no fix, and when it does get a fix and the time becomes correct.

decrepit
WA, 8706 posts
4 Oct 2018 9:23PM
Thumbs Up

Looks like version 3 with a BN880 and logomatic V2 is a goer!
No missing points, so I think that confirms the initial fix was the problem before.

Here's the results from about a 90min balcony test at 10hz NAV-PVT and a 2s interval NAV-SAT

54147 points read: checksum: 0, corrupted: 0, missing: 0

I'm still using the old 2200maH power bank with it, so there's no room to fit it in an enclosure. A new 1000maH lipo should arrive early next week so I'll be able to take it for a sail, and compare with version 2 and the GW52.

boardsurfr
673 posts
4 Oct 2018 9:54PM
Thumbs Up

Select to expand quote
decrepit said..
A new 1000maH lipo should arrive early next week so I'll be able to take it for a sail, and compare with version 2 and the GW52.


When you test it, try to have the two prototypes right next to each other. I did some tests with one unit on an arm and the other on the head, and the ranges did not overlap for longer runs. Differences were only 0.04 knots or less for 500 m, though. Turns out that this corresponds to just about a millimeter difference per data point, which could very easily be from differences in actual movements (perpendicular to the direction of motion, so it adds a bit to speed over ground).

decrepit
WA, 8706 posts
5 Oct 2018 9:33AM
Thumbs Up

That's what I haven't tried yet, a logger and a GW52 fit in the same bag on my head, but I don't think two loggers will. Must go and check now.

decrepit
WA, 8706 posts
5 Oct 2018 2:00PM
Thumbs Up

So there's no room in my new pouch for both loggers, but the carrier I'm making to fit on my helmet fits both loggers and the GW52. I just need a bigger pouch.



The mark 3 BN880 will go in the middle one once the new battery arrives.

seanhogan
2928 posts
5 Oct 2018 2:05PM
Thumbs Up

don't fall at speed !!! that thing will snap your neck!!!!

decrepit
WA, 8706 posts
5 Oct 2018 2:08PM
Thumbs Up

That's always a worry with any helmet, maybe I'll fix it so it comes off easy and floats.

decrepit
WA, 8706 posts
5 Oct 2018 6:50PM
Thumbs Up

Just tried NAV-PVT and NAV-SAT, both at 10hz,
didn't work.

40190 points read: checksum: 927, corrupted: 636, missing: 692.
So I won't bother with 18hz.
Next is NAV-SAT at 1 hz

Well that's odd, some of those points are inside the area that's reached a fix, but I can't see them in GPSResults
Instead there's points with missing acceleration data.



So is this due to missing data? or some other cause? I thought that GPSResults calculated acceleration from changes in speed?

So pushing to the limit now, 18hz both PVT and SAT, what's the bets? more missing acceleration points?

Don't think I'll be doing this whatever the outcome, 18hz with pvt and sat is a huge file, 25mb for an hour.
here's what the parser says.
12286 points read: checksum: 42, corrupted: 12, missing: 195.

I'll go and see what it looks like in GPSResults.
That's another thing, 18hz is 55.555555555555 ms in both u-center and the parser I don't get a decimal point choice.
the above is with the parser set to a period of 56, the below is a period of 55
12286 points read: checksum: 42, corrupted: 12, missing: 12145

And this is with the parser on 57

12286 points read: checksum: 42, corrupted: 12, missing: 195
very interesting, why pick 18hz if that can't be set in U-center?

the 195 missing would be right, not sure if I want to see what happens with NAV-SAT removed? I think 10hz with NAV-Sat at 1 sec, and go for a longer time.

decrepit
WA, 8706 posts
6 Oct 2018 10:00AM
Thumbs Up

So changed my mind and did an 18hz NAV-PVT only, not a very long test, but just wanted to see what happened.
this is the result from the parser.
000001: 0019 07 092
000002: 0059 07 092
010012.264 059ms
010030.184 112ms

The first three are before fix, the last one is immediately after, it jumps from 4 to 11 sats.

Peter is also having acceleration induced dropouts at 10hz, in his case due to GPSResults filters. I'll see what happens to mine if I turn filters off.
Back later.

decrepit
WA, 8706 posts
6 Oct 2018 1:09PM
Thumbs Up

So checked the 10hz PVT and 10hz SAT file with filters off, and it made no difference, there's still points with missing acceleration data. However Daffy has observed that the file is only an hour long and GPSResults gives an hour result with filters on. So I'm not sure what is going on with that file.

Earlier I set V3 back to 10hz with NAV-Sat at 1 second, then took it, along with the GW52 and v2 for a bike ride, but due to a seniors moment v3 had no card in it!
So it's on the balcony again seeing how that setting works. If it looks good I'll try another bike ride later.
Well the file looks good, BUTTTTTTTTTTTT it's only 14minutes long, the %%$# battery went flat. I don't want to fully charge it, because then it won't stay on, just long enough to give me an hour's bike ride would be good, but how long is that?

sailquik
VIC, 4091 posts
6 Oct 2018 6:10PM
Thumbs Up

Select to expand quote
decrepit said..
So checked the 10hz PVT and 10hz SAT file with filters off, and it made no difference, there's still points with missing acceleration data. However Daffy has observed that the file is only an hour long and GPSResults gives an hour result with filters on. So I'm not sure what is going on with that file.




It looks like GPS-Results does not mind if part of the sentence is missing that has the Acc data in it as far as a valid run goes. It seems to accept it if the main speed data is there and the point is not actually missed.
I'd say thats a trucated sentence.

decrepit
WA, 8706 posts
6 Oct 2018 4:10PM
Thumbs Up

Select to expand quote
sailquik said.. >>>>>> I'd say thats a trucated sentence.


That's my best guess as well.

boardsurfr
673 posts
6 Oct 2018 9:14PM
Thumbs Up

Yes, this is a corrupted/truncated sentence. GPSResults will definitely not include this region in 2 sec, 10 sec, and nautical miles. One missing point in a nautical mile, and it will be discarded. However, it seems GPSResults tolerates missing points in 1 hour runs. It has to, since crashes often lead to missing points, and most hours be mere mortals contain crashes (or points that are filtered for other reasons).

The programs differ in how they treat missing or filtered points. GPSAction Replay Pro does not care about missing points at all, it always calculates results. IIRC, ka72.com will not calculate 2 sec and 10 sec results if a point is missing/filtered, but will calculate longer runs like nautical miles. I'm not sure were exactly the switchover is for ka72.com.

decrepit
WA, 8706 posts
7 Oct 2018 3:06PM
Thumbs Up

I've just found one of the downsides with the logomatic. It draws heaps of current, it may be possible to reduce this by playing with the firmware, but I'm not up to that yet.

So at 10hz with 1s NAV-SAT info it draws around 160ma that's only 6:15hrs with the intended 1000ma/hr battery.
If I drop the NAV-Sat info the current drops to about 135ma, that should give another hour. Strangely, changing to 5hz doesn't make much difference, about 133ma.

boardsurfr
673 posts
8 Oct 2018 2:39AM
Thumbs Up

Select to expand quote
decrepit said..
I've just found one of the downsides with the logomatic. It draws heaps of current, it may be possible to reduce this by playing with the firmware, but I'm not up to that yet.

So at 10hz with 1s NAV-SAT info it draws around 160ma that's only 6:15hrs with the intended 1000ma/hr battery.
If I drop the NAV-Sat info the current drops to about 135ma, that should give another hour. Strangely, changing to 5hz doesn't make much difference, about 133ma.


That's a lot of current! I found one customer review on the Sparkfun site that confirm high current use. They say at low logging rates, a 2000 mA battery lasts 1.5 days - so that's about 55 mA. Compared to 6 mA for the Openlog, that's still a lot.
For the Adafruit Feather loggers, I've seen numbers of 10 and 20 mA. Have not tested it yet, but this would be ok, since the GPS module also needs something around 50 mA.

decrepit
WA, 8706 posts
10 Oct 2018 2:04PM
Thumbs Up

Finally got V3 in an enclosure with it's new battery, I'm sure I ordered 1000ma/hr but the one that arrived is 1100ma/hr. I'm not complaining, it fits in nicely.
So I'm doing a discharge test to see how long it holds up with NAV-PVT @ 10hz and NAV-SAT at 1second.
I couldn't resist a quick peek at the data, so after 4hrs I get these errors.

84105 points read: checksum: 21, corrupted: 17, missing: 23

000001: 0009 07 092
012204.112 112ms
000002: 003a 07 092
025418.700 200ms
000003: 0065 07 092
030240.700 200ms
000004: 005d 07 092
030243.700 200ms
000005: 005d 07 092
031150.700 200ms
2008300.600 713400100ms
000006: 0065 07 092
031410.700 200ms
000007: 0065 07 092
031645.700 200ms
000008: 0065 07 092
031841.700 200ms
000009: 0065 07 092
032031.700 200ms
000010: 0065 07 092
032319.700 200ms
032509.700 700ms
1300404.600 460680100ms
000011: 0065 07 092
032751.700 200ms
000012: 0065 07 092
033109.700 200ms
000013: 005d 07 092
033246.700 200ms
000014: 0065 07 092
033402.600 200ms
033411.700 200ms
033557.700 200ms
000015: 0065 07 092
033731.600 200ms
000016: 0065 07 092
033746.700 200ms
033923.600 600ms
000017: 0065 07 092
033929.700 200ms

All the "0000" times are before it's got a fix, but the cluster in the 033s are real.
So unless I can get into the firmware and improve things, looks like I'll have to drop NAV-SAT.

I have to buy another card to replace the one I pinched from my music player phone, so I'll see if I can try a faster one.

OK just ordered this.
SanDisk Extreme 16GB SD SDHC Memory Card U3 UHS-I 90MB/s Ultra HD VIDEO 4K 16G
I think that's twice the speed of the one in there at the moment, so we'll see if it makes any difference.

decrepit
WA, 8706 posts
10 Oct 2018 4:34PM
Thumbs Up

That's thrown a spanner in the works, 2nd session didn't record, a bad connection in the 3.3v rail to the GPS. So now I don't know how long it's been on, and if said bad connection is responsible for the missing and corrupted data. So back to square 1 with the tests.
I can't find the actual bad joint, all connections look fine through my X3 specs, so I've just bypassed the section.

boardsurfr
673 posts
10 Oct 2018 9:19PM
Thumbs Up

Could be the bad connection, but the misses I've seen in other 10 Hz data look very similar - mostly single point misses.

It is possible that the limitation is on the GPS chip, not the SD card or logger. I read somewhere that the chip would be able to take 40-50 measurements per second, but it's spec'd only to 10-18 Hz. As Raymond has pointed out, it does not even reach this speed when running from Flash.

When running at the maximum specified speed, I'm sure there are some limits on what data can be sent. I can't imagine that the chip would be able to send all UBX and/or NMEA sentences. The question is where exactly the limit is. Perhaps using Bluetooth and GPSLogit BT that's now available on a decent phone might provide some insights ... although the only definitive answer would be if things work. If there's still data loss, it might be due to the Bluetooth connection or the phone. The same thing is true for using another logger like the Pi or the Adafruit Feather Logger M0. Maybe I''l take some toys with me on vacation, just in case the weather is bad.

decrepit
WA, 8706 posts
10 Oct 2018 9:25PM
Thumbs Up

But it looks OK in U-center, all though I haven't saved a file there to have a close look.
It does look like18hz NAV-PVT and NAV-SAT is coming in.

boardsurfr
673 posts
11 Oct 2018 1:48AM
Thumbs Up

U-center is definitely a useful tool that might help here. It is difficult to see missing points in u-center when it's just one point missing every few thousand points. Recording from u-center to a file may work, although this adds a couple of new possible bottlenecks: u-center's file recording, and the connection between the GPS module and u-center (in particular any USB-to-serial cables).

You've made me curious now. I just put my Surface connected to the Drotek NEO-M8 outside (since that can be connected directly via USB), recording NAV-PVT and SAT at 10 Hz. Will let it run for a couple of hours (if Windooze lets me) and see if something's amiss.

boardsurfr
673 posts
Thursday , 11 Oct 2018 7:20AM
Thumbs Up

So I ran the test for a few hours. Logging both NAV-PVT and NAV-SAT worked ok for about 1 minute, then the first points were missed. This got worse over time. After an hour, most points were 200 ms apart, with just occasional short 100 ms periods. After a few hours, it converted to 5 Hz almost completely - I found one period were all points were recorded at 5 Hz for about 5 minutes, without a single point at 100 ms. It would appear that this is a limitation of the GPS chip. Not that the logger can't also loose points .. probably does it just to make things confusing .

decrepit
WA, 8706 posts
Thursday , 11 Oct 2018 10:16AM
Thumbs Up

Confusing seems to be the operative word!!!!!
I've done something to my unit now, seems in the process of fixing the 3.3v rail I've stuffed the RX and TX connection to the BN880. Off to have another play now.

decrepit
WA, 8706 posts
Thursday , 11 Oct 2018 1:32PM
Thumbs Up

Problem was in the BN880 between the breakout holes and the socket. Seems mounting it by the pegs isn't a good idea, I think mechanical strain has cracked the RX track. It's working OK now using the plug and ribbon cable. I've just got to put it back together, without doing more damage!
I'll get ride of SAT-NAV, and see if I can do a long session with only 10hz NAV-PVT

boardsurfr
673 posts
Thursday , 11 Oct 2018 9:59PM
Thumbs Up

Select to expand quote
decrepit said..
Problem was in the BN880 between the breakout holes and the socket. Seems mounting it by the pegs isn't a good idea, I think mechanical strain has cracked the RX track. It's working OK now using the plug and ribbon cable. I've just got to put it back together, without doing more damage!
I'll get ride of SAT-NAV, and see if I can do a long session with only 10hz NAV-PVT


I always use the cable. Makes it easy to switch GPS modules, hook them up to u-center, etc.

It will be interesting to see if you can get several hours of 10 Hz NAV-PVT without any lost points.

decrepit
WA, 8706 posts
Friday , 12 Oct 2018 3:47PM
Thumbs Up

Finally got it back together, completely reconfigured. I'm much happier with this layout.
Managed 1/2 hr at 10hz PVT before the battery went flat. just had 1 corrupted and 1 missing at the start, before fix.
I'm charging it now. and there's wind forecast for tomorrow, so I may get to test it live.

sailquik
VIC, 4091 posts
Friday , 12 Oct 2018 10:43PM
Thumbs Up

Select to expand quote
decrepit said..
Finally got it back together, completely reconfigured. I'm much happier with this layout.
Managed 1/2 hr at 10hz PVT before the battery went flat. just had 1 corrupted and 1 missing at the start, before fix.
I'm charging it now. and there's wind forecast for tomorrow, so I may get to test it live.



I recorded my car trip Sale to Sandy Point today. Roughly 2 hrs at speeds up to 50 knots.
10Hz, GPS-Logit. Manfreds BT GPS receiver. PVT and DOP @10Hz /SAT @1Hz

31 points missed in a tight cluster within about 5 seconds in the middle of the trip. No others. I cant see any reason why they dropped except the sudden appearance, and then disappearance, of a single Beidou satellite in the list on U-Centre which coincided with a jump from 11 to 12 sats used. Otherwise just the usual GPS and QZSS sats.

Hmmm. ??

I also did a stationary test with GPS-Logit set to 18Hz (GPS only) and had only two missing points in over 1 hour (and I think right at the very start). That is much better than I had experienced before and indicates that using only 1 GNSS may be an advantage. I will do some moving tests @18Hz tomorrow. Unfortunately, I can't configure Logit to do only one GNSS @5hz or 10Hz to test that.

decrepit
WA, 8706 posts
Friday , 12 Oct 2018 7:57PM
Thumbs Up

Maybe the theory about updating the chip with sat data has something going for it.

sailquik
VIC, 4091 posts
Friday , 12 Oct 2018 11:00PM
Thumbs Up

Select to expand quote
decrepit said..
Maybe the theory about updating the chip with sat data has something going for it.



Maybe, but I had the GPS going in the backyard doing stationary tests for a few hours before the trip. It should have been well and truly 'updated'.

I am now suspectig it may be just too many sats that is the issue??

boardsurfr
673 posts
Friday , 12 Oct 2018 10:04PM
Thumbs Up

I have come to the conclusion that the Openlog ismost likely fast enough to record 10 Hz NAV-SAT data, and that missing data points are probably often caused by the u-blox chip. I would say that chances that the chip is at fault are very high in cases where there are many dropped points close to each other.

For the first speed, the chip was setup to send a lot of different UBX records that have a fixed size (unlike e.g. NAV-SAT, where the size depends on the number of satellites visible or tracked). The chip put out ~ 538 bytes of data 10x per second, and did not track any satellites. The amount of data is comparable to sending NAV-PVT and NAV-SAT at 10 Hz, but since the chip had no reception, there was no concern about computing speed. The results was 24 corrupt records in a 42 MB file. So the Openlog almost was able to keep up, but not quite. I also checked the TX buffers and IO status in u-center when sending data at this rate, and saw no signs of problems. The TX buffer usage always stayed below 25%, so it's not a buffer problem on the GPS module site.

I then ran a 10 1/2 h test with just NAV-PVT at 10 Hz. That file was "only" 38 MB. Unfortunately, the chip started tracking satellites after a few minutes. There are two incidences of missing points:4 points in a row missing after ~ 4 hours,and 3 points after ~8 hours. There were no corrupt sentences, so the misses definitely appear to be caused by the GPS module.

Based mostly on Julien's explanations and Raymond's and our observations, here's what I think is going on:
-when recording at 10 Hz from 2 or 3 GNSS networks, the u-blox 8 is running "all out". It comes to a valid solutionalmostall the time, but about once every 1000 to 100,000 calculations, something goes wrong, and no valid solution is achieved within the ~100 ms the chip has. This leads to missing points - sometimes one point, sometimes a few points. This seems to happen more when about 15-20 satellites areused for the solution, but can also happen with fewer satellites, even without Galileo and Beidou.

In 5 Hz data, such problems are either absent, or much less frequent. We have seen several examples from different units and locations where the chip reverted to 5 Hz rates for a period of time when 10 Hz calculations failed. Possible reasons are that the calculations take just a bit longer than 100 ms, or that the chip can "start over" after the first calculation failed (but does not quite have to start from scratch, letting it finish the second time around).

Note that this "missing points" behavior is not limited to u-blox 8 chips. I have seen instances in 5 Hz Locosys data where points were missing in the middle of a run. However, these seem to be rare, and they are easily missed unless you specifically look for them. Canmores can miss points on a more regular basis.

decrepit
WA, 8706 posts
Saturday , 13 Oct 2018 8:57AM
Thumbs Up

Select to expand quote
boardsurfr said.. >>>>> Canmores can miss points on a more regular basis.


All that makes good sense Peter, but for this last sentence, I'd use "do" rather than "can"



Subscribe
Reply

Forums > Windsurfing   Gps and Speed talk


"DIY GPS Logger" started by decrepit