60csx Elevation

[kb9nvh]

Just so you know there's a few of us reading this...thanks for your efforts. I took a longer look at my csx today on the way home.

So, my question is if mine is set to autocalibrate and I set my home altitude at 528ft then why, in a few days is altitude off? Because of error introduced during the autocalibrate mode?

Edited May 26, 2006 by kb9nvh

[-Oz-]

I'm running firmware v2.62 Beta.

Thats your problem. Upgrade to v2.71 beta. You'll have no more problems.

[60csxuser]

Garmin 60CSx altimeter testing, Part 2... with a surprise

 

This time the altimeter auto-calibration was off and the altimeter manually set to an elevation of 28 metres. The altimeter was left in the Variable elevation mode (more on that later). The test was run for slightly more than 12 hours (21:18 to 10:05). This time the GPS was located close up to the window glass. Also, it was connected to the PC with only the NMEA serial cable. In the previous test also the USB cable was connected to power the GPS. I have a hunch that the cable connection might slighly degrade the GPS reception due to additional electrical noise entering the GPS via the cable. Anyway, the reception conditions improved because the GPS-altitude track, as recorded over the NMEA link by OziExplorer, collected only 3157 data points and the "virtual travel distance" was only 13.95 kilometres this time. The track recorded by the unit itself contained 722 data points.

 

 

 

Let’s again start by looking at the ambient pressure plot.

 

The Garmin shows a 1002.7 mb high at 01:42, a 1002.1 mb low at 05:10 and rises through 1002.8 mb at 10:00. The Vaisala plot shows a 1003.8 mb peak at 01:50, a 1003.3 mb low at 05:30 and 1003.8 mb at 10:00. The differences in the pressure changes between both devices are thus around 0.1 mb, which is good enough for practical use. (Note that the absolute pressure values differ by around 1 mb, I assume this is due to the difference in elevation between the two sites).

 

Next then the elevation plot. The following Excel plot again shows the NMEA-logged GPS elevation data in dark blue. The violet colored line is a 10-point rolling average and the blue line is a 50-point rolling average of the GPS elevation data. The yellow line is the Garmin’s elevation plot, in this case non-auto-calibrated.

 

 

The yellow line as well as the Garmin screencapture show an elevation variation of around 10 metres, which corresponds very well to the 1-mb pressure variation. Also the timing of the highs correspond to the lows in the pressure, so this shows the Garmin’s altimeter works as it should.

The GPS elevation data shows slightly less scattering than the previous plot, although it still is jumpier than I’d expect.

 

Finally the Garmin’s barometer plot and today’s surprise. The plot was as jumpy as yesterday’s GPS-auto-calibrated version. And comparing the plots, for example the peak at 07:36 and dip at 08:20 fit reasonably well with the Excel GPS altitude plot, so it seems clear these jumps come from the GPS altitude.

 

 

The remedy came when I changed the setup from ”Variable elevation mode” to ”Fixed elevation mode”. Then the plot changed and showed the same smooth curve as the ambient pressure plot, with the reading differing by 3.1 mb which corresponds close enough with the 28 metres elevation value I used in the manual calibration.

 

 

But this seems like a clear bug in the 2.70 software in the auto-calibration off mode. Apparently the unit stores the GPS auto-calibration values in any case, and then erroneously applies them to the barometer plot, if the unit is set to variable elevation mode. That’s at least how I understand it. I wonder if this is the same ”2.70 Bug” reported in another thread? Any comments?

 

Anyway I’m considering the following further tests. One is to do a long term (at least 48 h) test of the altimeter (without auto-calibration) to check for possible long term drift. The second is to drive a fixed route several times and compare the in-motion elevation accuracy with and without auto-calibration. I have a hunch that stationary testing of the GPS-altitude accuracy is not very useful, at least in my case with a limited view of the satellites. Hopefully that would improve in a drive test.

[kb9nvh]

OK, Trying to understand here. It would seem to me that "autocalibration" mode would be absolutely desired if one was moving as the gps would have no way of knowing if a pressure variation was due to altitude change or ambient pressure change. Sort of like, "lock to roads" is turned on when using follow road mode for routing.

 

So, I would think that when in "variable elevation mode" or not you would always want correction applied to the barometer plot just like you always want to lock on roads when you are following roads (even if you turn off lock on roads).

 

The functionality of this would be nice to have cleared up by Garmin itself.

 

Garmin 60CSx altimeter testing, Part 2... with a surprise

 

This But this seems like a clear bug in the 2.70 software in the auto-calibration off mode. Apparently the unit stores the GPS auto-calibration values in any case, and then erroneously applies them to the barometer plot, if the unit is set to variable elevation mode. That’s at least how I understand it. I wonder if this is the same ”2.70 Bug” reported in another thread? Any comments?

 

Anyway I’m considering the following further tests. One is to do a long term (at least 48 h) test of the altimeter (without auto-calibration) to check for possible long term drift. The second is to drive a fixed route several times and compare the in-motion elevation accuracy with and without auto-calibration. I have a hunch that stationary testing of the GPS-altitude accuracy is not very useful, at least in my case with a limited view of the satellites. Hopefully that would improve in a drive test.

Edited May 27, 2006 by kb9nvh

[60csxuser]

Hi,

I agree it would be excellent to have a detailed description by Garmin (not only for the altimeter by the way).

 

You are probably right that autocalibration may be good when moving. Problem is, it seems it's not so good when stationary. I've thought about if it's possible to devise a way to suppress autocalibration, or at least give a warning, in situations that it does more harm than good.

One possibility might be to utilize the fact that ambient pressure usually changes slowly while GPS altitude varies much more rapidly. For example, even in a quick ambient pressure change (weather stormfront) the corresponding altitude change rate is slow, less than 1.5 fpm (foot per minute) while in my NMEA recordings the GPS altitude change rate is in many cases 500 fpm or even more.

 

So in theory, if the GPS detects it's more or less stationary, it should disregard quick changes in GPS-altitude, while if it's moving, quick changes should be recorded. But to implement this in a reliable automatic feature is probably easier said than done

 

As to the bug suspicion, I presume the main intended use of auto-calibration is to enhance the accuracy of an elevation plot. So in the elevation plot it's OK to use it. But to me, a barometer is a barometer and should not show anything else than pressure variations. At least I can't think of other barometer applications than stationary and 2D (sea navigation).

 

OK, Trying to understand here. It would seem to me that "autocalibration" mode would be absolutely desired if one was moving as the gps would have no way of knowing if a pressure variation was due to altitude change or ambient pressure change. Sort of like, "lock to roads" is turned on when using follow road mode for routing.

 

So, I would think that when in "variable elevation mode" or not you would always want correction applied to the barometer plot just like you always want to lock on roads when you are following roads (even if you turn off lock on roads).

 

The functionality of this would be nice to have cleared up by Garmin itself.

Edited May 27, 2006 by 60csxuser

[TracknQ]

60csxuser, thank you for your hard work. I'll have a question for you at the end of this post.

 

I, too, have been studying the 60CSx altimeter ever since my first two brand new 60CSx units showed +98,422' and -4,921' elevation readings near sea level.

 

TOTAL ASCENT INACCURACY

 

I reported on total ascent inaccuracies in the Version 2.70 Bug thread. I got total ascent readings that climbed continually when my 60CSx was stationary. I got total ascent increases when I was going downhill. I got very different total ascents when taking walks over the same routes. I've turned my total ascent data field off. It's unreliable and therefore useless to me.

 

ALTIMETER INACCURACY

 

My 60CSx's shows greater elevation variability when auto-cal is enabled than when auto-cal is off. For example, the following profile shows a 29' elevation range with auto-cal off and a 111' elevation range with auto-cal on. My 60CSx was recalibrated to 75' and then left leaning against a window for about 6 hours.

 

 

My 60CSx gives higher and more inaccurate elevation readings when auto-cal is on than when it's off. For example, my 60CSx showed 13' with auto-cal off and 63' with auto-cal on at a survey-controlled 13' bench mark. (I live in Hawaii, where atmospheric pressure is usually pretty stable.)

 

Also, my 60CSx's elevation readings consistently creep upward or downward over time/distance. The following profile was from a 6.55-mile walk. I recalibrated the altimeter to 14' at the start and began and ended the track log at the same exact point. The elevation crept 80.4' upward.

 

 

The following table shows elevation 'creep' measured in various recent walks. Always, my 60CSx's altimeter was recalibrated to 14' at the start, and the track log was begun and stopped at the same location outside my home.

 

 

The elevation creep was always more when auto-cal was enabled than when it was disabled.

 

I have no scientific training. To better understand my 60CSx, I've studied geodesy, how GPS receivers and altimeters work, and weather forecasting. I'll put links to some of the articles that I've read at the end of this post. I've learned much.

 

I now understand that there are many reasons why elevation readings can be twice as inaccurate as EPE (estimated position error) inaccuracies.

 

GPS receivers triangulate signals from satellites to locate X, Y and Z (latitude, longitude and elevation) coordinates on an ellipsoid model of Earth. GPS receivers measure ellipsoid height but not orthometric height or geoid height. The geoid is an equipotential model of the Earth's surface based on 1G gravity. Geoid height is the height of the geoid above or below the ellipsoid. Orthometric height is the elevation of a topographic location above mean sea level. Orthometric height is ellipsoid height minus geoid height. My 60CSx's 2.70 firmware contains a very long algorithm. The algorithm includes an obsolete table that approximates geoid heights. Some of my 60CSx's altitude inaccuracies are due to height biases in the models and the obsolete table.

 

Other factors make my 60CSx's elevation readings inaccurate, including poor satellite geometry, slowing of satellite signals in the ionosphere and troposphere, satellite signals bouncing from objects on Earth (multipath), clock drift, orbital (ephemeris) errors, and electronic noise from sensors in my 60CSx.

 

I believe that the 60CSx is more prone to the inaccuracies I discussed above because Garmin wed its existing hardware/software with SiRF without adequate pre-market testing and without appropriate pre-market revamping of the firmware algorithm.

 

Here are links to some the articles that I read:

 

The GPS Observer: Geoid models

GPS Altitude Readout > How Accurate?

How reliable are GPS heights?

Sources of Error in GPS

D.L. Wilson's GPS Accuracy Web Page

Converting GPS Height into NAVD88 Elevation with the GEOID96 Geoid Height Model

A Discussion of Various Measures of Altitude

GPS Errors and Estimating Your Receiver's Accuracy

Modeling of GPS Vertical Errors

The NGS GEOID Page

Barometers for Weather Watching

Applying the Barometer to Weather Watching

Conversion Units

Conversions

 

Also look at Wikipedia's articles about GPS, geodesy, altimeter, atmospheric pressure, etc.

 

60csxuser, in Post #99, you said: "In the 60CSx, this resetting is according to an internet posting done automatically by using the GPS altitude as an reference and making adjustments every 15 minutes." Can you give me the URL of the Internet posting to which you referred?

 

I don't think GPS altitude has anything to do with barometer plots. Today, I turned my 60CSx's GPS receiver off (Satellite Page > Menu > Use With GPS Off) and leaned the 60CSx against a window. My 60CSx continued to give me barometer plots when no GPS altitude was being obtained.

 

I'd appreciate receiving comments and suggestions.

 

TracknQ

Edited May 28, 2006 by TracknQ

[kb9nvh]

Ideally, with autocal off I would hope that your begin/end altitude inaccuracy is due to barometric change at the start/end loaction. Next time you should try and find barometirc info to ad to the table (from an independent source). I think the autocalibrate is problematic for Garmin. since the GPS alititude is so inaccurate it would seem there'n no good way for garmin to adjust for real altitude change. The best way would be for elevation info to be included in the basemap (topo). but even that would be prone to error due to mapping inaccuracies and much more so where elevation changes rapidly.

 

Next time to a recalibrate at your end location and see what your ambient pressure is comparied to the start of your walk. I dont think the GPS will ever give a perfect elevation plot over long periods of time.

 

I wonder if Garmin uses position AND alitutude data to decide to recalibrate the barometer? Like, use altitude data but ONLY if the position has changed and you have at least a 5 minute average of altitude in one position. Of course, that would not work for an activity that doesn't stop moving (hiking, bikeing, skiing)..

 

**********************************************************

ALTIMETER INACCURACY

 

My 60CSx's shows greater elevation variability when auto-cal is enabled than when auto-cal is off. For example, the following profile shows a 29' elevation range with auto-cal off and a 111' elevation range with auto-cal on. My 60CSx was recalibrated to 75' and then left leaning against a window for about 6 hours.

 

My 60CSx gives higher and more inaccurate elevation readings when auto-cal is on than when it's off. For example, my 60CSx showed 13' with auto-cal off and 63' with auto-cal on at a survey-controlled 13' bench mark. (I live in Hawaii, where atmospheric pressure is usually pretty stable.)

Edited May 28, 2006 by kb9nvh

[60csxuser]

60csxuser, in Post #99, you said: "In the 60CSx, this resetting is according to an internet posting done automatically by using the GPS altitude as an reference and making adjustments every 15 minutes." Can you give me the URL of the Internet posting to which you referred?

 

Hi TracknQ,

 

This is the link to the posting: http://regex.info/blog/2006-04-16/179 .

 

As to your very interesting "creep" table. Google was kind to me and found this site which gives an on-line barometer plot from Honolulu Airport. As far as I can decode it, the pressure variations are not very strong (within 5 mb) but fairly quick. For example, the pressure rises 1,5 mb in 2 hours from 04:00 to 06:00 UTC on the 27th. Assuming your walk was 2 hours, this pressure rise corresponds to a change of 41 feet in the indicated elevation. Your "auto-cal off" results show lower elevation "creep" than that, so I'd be inclined to explain the "creep" with the pressure variation. You might want to compare your future results with the air pressure variations plotted at that site and see if your barometer plot matches theirs, in which case you may use it to correct your elevation plots.

 

As to the "GPS auto-cal on" results, one reason could be if the GPS map datum parameters happen to have a local error which puts the GPS-defined "sea level" below the actual sea level. I'm not familiar enough with the auto-calibration workings, but one guess is that it strives to "pull" the device's elevation reading towards the one indicated by the GPS altitude. And if the GPS map parameters give wrong data about the local sea level elevation, this may result in the "creep" you see. You might want to give the GPS time to calibrate itself for one hour or so before starting your walk and see if it makes any difference. This link may be of use.

[TracknQ]

kbnnvh and 60csxuser, thank you both for your help. You've taught me to look at ambient and barometric pressures more closely. I will start doing that as soon as I figure out how to do so.

 

60csxuser, thanks for the link to Jeffrey Friedl's Blog. His first and second posts about his 60CS's altimeter were informative. I share his frustration. Also, thanks for the link to the online barometer plot at the Honolulu Airport. That link got me searching for other online barometer plots. The Citizen Weather Observer Program offers free online barometer plots from many stations around the world. Finally, thanks for linking me to the article about the 76S barometer. Before my walks, do I now have to wait outside for an hour, with my 60CSx powered on, while the auto-calibrator 'warms up' and 'settles down?' If, as the article suggests, the barometer is calibrated by the GPS receiver, why did I get a barometer plot with the GPS receiver disabled yesterday?

 

I haven't digested the following data yet. I'd sure appreciate hearing from you or anyone else if any helpful connections are seen.

 

Here's the barometric pressure plot from the PHNL weather station that's 8.6 miles from my home:

 

Here are two screen shots that I saved at 16:15 HST yesterday, after my 60CSx-in-window-ACon-&-off test and a 6.73 mile walk with ACoff.

 

Here's a .pdf file containing track log data from yesterday's walk.

 

The following screen shot shows the profile from yesterday's walk.

 

 

Notice that there was only a 2-foot differential between my start and end elevations.

 

I recorded the following two screen shots this morning. They show 8-mile and 32-hour elevation plots that were recorded by my 60CSx. (I guess that the -19' elevation suggests a pressure increase. I took the shots in my 8th-floor apartment. Auto-cal was still off.)

 

Finally, here are 24- and 48-hour ambient and barometric pressure plots that I recorded at about 09:30 HST this morning:

 

Thanks again.

 

TracknQ

[TracknQ]

I forgot to thank 60csxuser for his very intelligent explanation for my 60CSx's elevation 'creep.'

 

There's an active volcano a few hundred miles from my home. Magma raises and lowers the Earth's crust in Hawaii. The National Geodetic Survey updates its geoid model every 3-4 years, to account for the gravimetric changes that occur here. Unfortunately, Garmin has not added data from recent NGS geoid models to the algorithm in its 2.70 firmware. That alone creates a height bias when auto-cal is on.

 

Moreover, Garmin has not sufficiently revamped the algorithm in its firmware to properly wed the new hypersensitive SiRF chipset with its already-sensitive barosensor. My 60CSx's SiRF chipset and barosensor each add electronic noise. The noise causes readings to fluctuate. The fluctuations cause creeping increases over time, like the increase in total ascent.

 

In addition to noise, ambient pressure constantly changes.

 

Due to noise and pressure changes, I haven't been able to discern a consistent height bias that's caused by the obsolete geoid model in the firmware.

 

It would be nice if I could rely on my 60CSx's elevation readings without guesswork. From my experience in Hawaii so far, that's not been possible, because there have been no constants.

 

I'll try letting my auto-calibrator 'warm up' and 'settle down' before I head out on future walks. Since I get multipath and weak SV signals in my apartment, I'm not sure that pre-walk warm-ups will help that much. Also, if my above hypotheses are correct, I doubt that a warm-up will help. I'll let you know.

 

I'll be in Central Oregon soon. I look forward to testing whether my 60CSx's auto-calibrator will give me more accurate elevations there. (Unfortunately, Central Oregon also has active magma movement.)

 

Jeffrey Friedl may be correct. The altimeter may be virtually useless.

 

I'll start comparing my 60CSx's pressure plots with the PHNL barometer and let you know if I see any consistent patterns.

 

Thanks again 60csxuser and kb9nvh!

 

TracknQ

[kb9nvh]

All in all, I would still rather have the altimeter than not. I find it pretty consistent driving back and forth to work. Bike rides will be nice to have a somewhat accurate profile of my trip.

 

I also like the compass mode for geocaching since it always knows what direction I'm pointed even if not moving...

 

I'm glad I paid for the "s" version even with the flukes and problems.

[TracknQ]

The last four screen shots in my Post #109 show a 1016 mb ambient pressure and a 1021 mb barometric pressure. These readings occurred simultaneously at 75' above sea level. I thought ambient pressure closely matched barometric pressure near sea level.

 

Can anyone explain the 5 mb difference? If the 60CSx's 'auto-calibration' function is somehow involved in creating barometric pressure, could the 5 mb difference help explain why my altimeter gives elevation readings that 'creep?'

 

TracknQ

[kb9nvh]

Well, this is confusing and Garmin is using ambient to mean something different from barametric however, ambient means pretty much what your experiencing right now

 

def: (Prevailing environmental conditions, as opposed to measurement in a laboratory or waste stream. 2) Refers to the overall conditions surrounding a place or thing.)

 

One of them must be adjusted for something...Maybe I'll go read the manual and see what it says.

 

 

The last four screen shots in my Post #109 show a 1016 mb ambient pressure and a 1021 mb barometric pressure. These readings occurred simultaneously at 75' above sea level. I thought ambient pressure closely matched barometric pressure near sea level.

 

Can anyone explain the 5 mb difference? If the 60CSx's 'auto-calibration' function is somehow involved in creating barometric pressure, could the 5 mb difference help explain why my altimeter gives elevation readings that 'creep?'

 

TracknQ

[60csxuser]

Hi again,

 

after some serious headscratching I'll try to comment TracknQ's post.

 

Remember that a traditional barometer measures the ambient pressure, but is calibrated according to the local elevation (1 mb/27 feet) to indicate sea level pressure. So in a stationary situation one expects that the difference between ambient pressure and barometer reading stays constant, and = 2.8 mb for TracknQ's 75-ft elevation (when properly manually calibrated).

 

I make the basic assumption that the ambient pressure reading of the 60CSx comes straight from the sensor (with some basic dampening) and is not influenced by calibrations, GPS altitude etc. So it can be highly trusted, except for local pressure disturbances such as opened windows.

 

In the first two screenshots the pressure difference is 6 mb, so obviously something is wrong. Remember my "barometer bug" suspicion? The barometer plot shows the same jumpiness as mine, a hint that it is affected by the GPS-altitude (the bug). I guess the unit was in "variable elevation" mode? The 6 mb pressure difference implies an elevation of 162 feet. This jibes with our previous assumption that the GPS device's map datum parameters have a local error that places the sea level lower than it actual level, increasing the elevation the unit "sees".

 

The track log's 2-foot differential may indicate that the pressure was steady during TracknQ's walk. No problem here.

 

In the next two screenshots the ambient pressure change during the night was 1016-1012.6 = 3.4 mb corresponding to a -92 ft elevation difference. Lo and behold, 75 ft + 19 ft equals 94 ft, which is close enough! No problem here either.

 

The final four screenshots show an ambient-barometer pressure difference of 5.2 mb, again different from the 2.8 mb that corresponds to 75 ft elevation, and a hint that the GPS-altitude is again messing up the readings (the "barometer bug").

 

In order to check this I made a small experiment with my 60CSx. I manually calibrated it to my elevation, 25 m (82 ft) with auto-calibration off and in the Variable elevation mode. The barometer indicated 996.8 and the ambient was 993.8, perfectly according to the 27 ft/mb formula. Within 30 minutes the barometer reading had changed to 995.7 while the ambient stayed constant. (Such a barometer fall of 2 mb/hour would in the real world belong to a hefty stormy weather). In fact this was caused by the "barometer bug". Interestingly, the indicated elevation stayed constant at 25 meters. Setting the unit in Fixed elevation mode instantly returned the barometer reading to the original 996.8, implying that in this mode we don't have the bug.

 

Now we can try to understand the logics of the 60CSx. For some reason unknown to me, in the Variable elevation mode the barometer reading is affected by GPS-altitude even if "auto-calibration" is off (the "barometer bug"). The visible effect of this is that GPS-altitude fluctuations change the indicated barometer pressure while keeping the indicated elevation constant, in effect "moving the sea-level reference". Whereas changes in the ambient pressure due to weather change the indicated elevation while keeping the "sea-level reference" constant.

 

Complex, yes! Hopefully this explanation has not caused an even higher level of confusion....

 

The practical conclusion so far: Trust the barometer reading only when in the Fixed elevation mode. Actually this makes great sense, as any barometer applications I can think of are either stationary or move only horizontally (such as a ship on the seas).

 

edit:typo

Edited May 29, 2006 by 60csxuser

[endura]

i did not realised this Max elevation showing ---- until my friend informed me. I tried to reset ON/OFF several time but still showing ---

 

i did this to reset the max elevation and now it work fine.

 

on altimeter page, menu, reset, tick max elevation, apply.

[-Oz-]

i did this to reset the max elevation and now it work fine.

 

on altimeter page, menu, reset, tick max elevation, apply.

Download the latest firmware via webupdater and that will fix the problem permanently. Get 2.70 or 2.71 (beta).

[+theslowskys]

I see you are trying to figure out the altitude changes.

 

I am not so worried about it since right now I just want to get the directional movement to slow down.

It seems to walk around a bunch when I and stopped.

 

I am new to this GPSr stuff. Only had it 3 days. I bought a 60csx hoping the up to 3 meter accuracy would mean I would get close to that.

 

When I geocache and look for a cache I get within 20 to 70 feet of it. THat is about all it can do with 5 satellites or more locked in.

 

If a stay in one spot it will say the cache is moving from 20 to 120 feet away and back.

 

Even in a very open field when I get 8 or so locked in, It will still walk away on me up to 100 feet or so.

 

If I leave it sitting in my house it will only have 3 satellites and it jumps from 20 feet away to 210 feet away. usually in a somewhat north to south path.

 

Also I have to reset my altitude to 729 every day for my house. It jumps around.

 

On the main satellites screen I just now noticed that it says LOCATION +- 39, 43, 89 feet. # readings within 10 seconds. If I leave it still on the table it seems to settle in at 29 to 44 feet + or -.

 

Is this normal?

I was hopping to get within 10 or 20 feet not an 80 foot circle.

 

The constant "walking" of this is driving me nuts.

 

I am running the 2.70 version of the software.

 

Also the pointer to the waypoint on the COMPAS page just jumps all over when you get within 20 feet or so, I guess since it walks so much.

 

I thought the 2.70 was suppose to cut the walking down?

 

Any help anyone?

AM I just expecting too much from it?