Position Recorders, Accuracy, and Badge altitude gains

On Tuesday, May 29, 2012 11:52:09 PM UTC-7, uros wrote:
On 29 maj, 19:03, Darryl Ramm wrote:
On Tuesday, May 29, 2012 9:16:26 AM UTC-7, jjbird wrote:
On Tuesday, May 29, 2012 10:52:09 AM UTC-4, Grider Pirate wrote:
On May 29, 6:45*am, Tony wrote:
On Monday, May 28, 2012 1:59:19 PM UTC-5, uros wrote:
I have checked both flights and for the first flight it seems to me
that the flight recorder was turned on too late. Every GPS needs some
time to find correct position and when the GPS is moving this time is
even longer. The correct procedure is to turn on the GPS few minutes
before the flight so that GPS could correctly find first position and
will start recording correctly. These jumps could be easily detected
and the recording would be cut off.

Sure I know that you should turn it on on the ground but I forgot.. *And with turning it on in flight of course it will take a little time to "find itself". *However why on earth doesn't it wait until it has found itself to start recording fixes? *And even then, why was it recording fixes over 2000 ft too high if it had found itself?

The bottom line is probably "it shouldn't". *That it isn't reporting a
non-valid 3D fix is troublesome...
My GPS knowledge is quite dated, but doesn't it take something like 12
minutes to get an ephemeris (or was it almanac?) download? Fifteen+
years ago, we would not launch a unit until we confirmed a 3D fix, and
that OFTEN took 12 minutes.

The almanac takes 12.5 min to transmit, but each satellite will transmit its own full ephemeris and other info every 90 seconds. Most new receivers have flash memory that saves the almanac though so they can usually get a fix in under 30 seconds (as quick as a second or two if they were powered down for only a few minutes). The other big change that allows modern gps get rapid fixes is the number of channels they have - in steady operation having 50 channels does no good, but when booting up multiple channels can be trying to get a fix on a satellite which makes the time to first fix a lot faster.

I'm not sure what gps engine the CE uses (or how it is set up internally), but it would have to check more than one message from the gps to get a full picture of the quality of a fix, the standard NMEA messages have a data valid/invalid flag, but don't appear to distinguish there what type of fix is being delivered, it takes a separate flag in another message to determine that.

That said, I can't quite wrap my head around whether the 2d/3d issue could be causing what Steve saw, a 2d fix should pin the altitude to the geoid (which ought to be pretty close to ground level - it's about 475m at sunflower). If for some reason it didn't have a very good view of the sky overhead then it could be just an issue of only using satellites near the horizon (in which case the gps altitude would be pretty poor quality).

The chipsets used in these devices are more than simple GPS receivers, an for example what altitude a 2D fix provides is often very much up to the chipset. For example the SiFR III chipset used in the FR100 provides settings that will generate 2D fixes with extrapolated or software provided altitude data, varies wether the chipset provides 2D data when 3D is not available, etc. etc. (e.g. look up (altitude hold, altitude source, degraded mode, in the SiRF documentation). I'm not at all familiar with the chipset in the newer FR300. But we've got reports here of strange altitude behavior with both an FR100 and an FR300. I have **no idea** what is going on, but if I was looking at this seriously I would want to know all the chipset firmware settings. Basically I don't see any way a NAC or the IGC could/should ever approve a position recorders without knowing the full chipset firmware settings (and verify them and have a way to verify in future updates/revisions), and possibly for the purposes of transparency/disclosure it make sense to publish those setting in the approval documents in future. Maybe Uros might want to shorten this conversation and just make those firmware settings public.

Again I want to point out I have no idea what is going on here, just these *chipsets really need to be understood by the approving bodies. My understanding is for position recorders that is now really the NACs not the IGC itself. I am not sure if the NACs to have the technical resources/skill level to tackle this type of thing--it would be great to be wrong here.

Darryl

I do not hide that I am using the consumer electronics device. The
firmware is implemented by the device manufacturer. They only
implemented few changes so that I can implement some of the flight
recorder features. The chipset inside FR300 is SKYTRAQ and which is
listed on my page http://www.flywithce.com/recorder300.html.

All device manufacturers are using standard chipsets (or even more –
standard modules). I have not heard that anyone would implement
changes to position calculation. So even if you would spend few
hundred EUR more there is no guarantee that GPS altitude would have
perfect altitude (bonus there is that you have pressure altitude).

And consumer electronics device is the main reason why the device can
cost 89 EUR instead 500 EUR like any other fully certified (and
purpose built) flight recorder. I believe that this is great device
which is very simple to use and very affordable.

Best regards

Uros
www.flywithce.com

Uros

Thanks, I'm not ever suggesting that you were not disclosing the chipsets being used or that these were not based on consumer devices, or that is necessarily a problem. All that I was aiming at is that with these advanced modern GPS chipsets is the important of needing to understand the firmqare settings controlling many of the chipset behaviors. Telling us the chipset used just lets people know potentially the many different chipset settings that could be relevant to practical use and approval as a position recorder.

I'm hoping you know those settings used in the devices from whoever you purchase them. I think its is important that the approving agencies are provided with that data as well, and that they understand it. And I think it would help with transparency if that data was in future made public as a part of any approval process. With many GPS chipsets telling us the model chipset used without that extra data is next to useless.

Thanks


Darryl

On Wed, 30 May 2012 07:03:43 -0700, Darryl Ramm wrote:

I'm hoping you know those settings used in the devices from whoever you
purchase them. I think its is important that the approving agencies are
provided with that data as well, and that they understand it. And I
think it would help with transparency if that data was in future made
public as a part of any approval process. With many GPS chipsets telling
us the model chipset used without that extra data is next to useless.

Daryl,

In my quite limited experience the only GPS I know for certain smoothed
the trace was the Garmin GPS II+ and I know why as well: it only stores a
few points (1024 or 2048 IIRC) and so to hold all of a long trail it has
to condense runs of fixes into a straight line with just the ends
retained. Of course this would have made it utterly useless as a COTS
logger. However, AFAIK it has never messed about with the fixes it
outputs. At least I've never seen any sign of this in the traces dumped
from my EW Model D logger, which is invariably connected to one of my
Garmin GPS II+ units.

On a slightly different topic: GPS altitude. I've always known that all
GPS altitudes are relative to the WG-84 geoid but have never known how
precisely that corresponds sea level, so I finally did some research and
it turns out that its within +/- 1 metre of AMSL.

That is less than the error in a standard GPS receiver's height
measurement. I don't believe I've ever seen an EPE of less than 3 metres.
Just now one of my GPS II+ units said EPE=5m when I took it outside. So,
my guess is that for almost all our purposes its reasonable to take a
valid GPS height as being equivalent to altitude AMSL provided an error
of +/- 20-25 feet is acceptable for the task in hand.


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |

On Wednesday, May 30, 2012 3:52:41 PM UTC-7, Martin Gregorie wrote:
On Wed, 30 May 2012 07:03:43 -0700, Darryl Ramm wrote:

I'm hoping you know those settings used in the devices from whoever you
purchase them. I think its is important that the approving agencies are
provided with that data as well, and that they understand it. And I
think it would help with transparency if that data was in future made
public as a part of any approval process. With many GPS chipsets telling
us the model chipset used without that extra data is next to useless.

Daryl,

In my quite limited experience the only GPS I know for certain smoothed
the trace was the Garmin GPS II+ and I know why as well: it only stores a
few points (1024 or 2048 IIRC) and so to hold all of a long trail it has
to condense runs of fixes into a straight line with just the ends
retained. Of course this would have made it utterly useless as a COTS
logger. However, AFAIK it has never messed about with the fixes it
outputs. At least I've never seen any sign of this in the traces dumped
from my EW Model D logger, which is invariably connected to one of my
Garmin GPS II+ units.

On a slightly different topic: GPS altitude. I've always known that all
GPS altitudes are relative to the WG-84 geoid but have never known how
precisely that corresponds sea level, so I finally did some research and
it turns out that its within +/- 1 metre of AMSL.

That is less than the error in a standard GPS receiver's height
measurement. I don't believe I've ever seen an EPE of less than 3 metres.
Just now one of my GPS II+ units said EPE=5m when I took it outside. So,
my guess is that for almost all our purposes its reasonable to take a
valid GPS height as being equivalent to altitude AMSL provided an error
of +/- 20-25 feet is acceptable for the task in hand.


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |

Martin that is all nice, but for particular devices here people are observing altitude errors of 1,000' or so (presumably caused by 2D fixes being marked incorrectly as valid 3D fixes). To me that's the issue, not whether GPS altitude in principle is usable. It is, but as I think this is showing the devil is in the details of how devices work/are usable in detail in the field (and consequently what the specifications they are required to meet are and how they are approved for use). Hopefully there are some quick product fixes possible here and some look at improving the specs and/or approval process.

Darryl

On Wed, 30 May 2012 16:13:59 -0700, Darryl Ramm wrote:

Martin that is all nice, but for particular devices here people are
observing altitude errors of 1,000' or so (presumably caused by 2D fixes
being marked incorrectly as valid 3D fixes).

Point. And, a related one: how come there are all these GPS devices out
there (Garmin, I'm looking at you) whose display real-estate is so
precious that they can't devote any of it to flagging up invalid fixes.
At least neither LK8000 nor XCSoar suffers from this problem.

I've never seen anything like the sort of errors you mention, though if
my EW model D is left running on the ground its trace often hops up and
down a few feet. Do we know if any of those mega-deviations spike down to
exactly zero ft?


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |

On Wednesday, May 30, 2012 3:52:41 PM UTC-7, Martin Gregorie wrote:
[snip]
On a slightly different topic: GPS altitude. I've always known that all
GPS altitudes are relative to the WG-84 geoid but have never known how
precisely that corresponds sea level, so I finally did some research and
it turns out that its within +/- 1 metre of AMSL.

Actually I'm not sure where you get +/- 1m difference between the WSG-84 geoid and AMSL. It's potentially larger than that (but still that does not mean GPS altitude is inherently not usable). e.g. see this article http://www..esri.com/news/arcuser/0703/geoid1of3.html

Darryl

On Wed, 30 May 2012 16:20:28 -0700, Darryl Ramm wrote:

On Wednesday, May 30, 2012 3:52:41 PM UTC-7, Martin Gregorie wrote:
[snip]
On a slightly different topic: GPS altitude. I've always known that all
GPS altitudes are relative to the WG-84 geoid but have never known how
precisely that corresponds sea level, so I finally did some research
and it turns out that its within +/- 1 metre of AMSL.

Actually I'm not sure where you get +/- 1m difference between the WSG-84
geoid and AMSL. It's potentially larger than that (but still that does
not mean GPS altitude is inherently not usable). e.g. see this article
http://www.esri.com/news/arcuser/0703/geoid1of3.html

He "The new World Geodetic System was called WGS 84. It is currently
the reference system being used by the Global Positioning System. It is
geocentric and globally consistent within ±1 m. Current geodetic
realizations of the geocentric reference system family International
Terrestrial Reference System (ITRS) maintained by the IERS are
geocentric, and internally consistent, at the few-cm level, while still
being metre-level consistent with WGS 84." - from:

https://en.wikipedia.org/wiki/World_Geodetic_System

yes, I know its only Wikipedia, but IME it seems to be generally OK for
technical details, but its backed up by this source:

http://kartoweb.itc.nl/geometrics/Re...faces/body.htm

IOW the 'best fit' geometric elipsoid, which can deviate from AMSL by up
to 105m is adjusted for gravimetric factors (the reference data set seems
to be ITRF96) and the result is the WGS84 geoid. According to the second
reference: "The World Geodetic System of 1984 (WGS84) datum has been
refined on several occasions and is now aligned with the ITRF to within a
few centimetres worldwide." which appears to be referring to the
deviation from AMSL though this is nowhere stated explicitly but appears
to be the meaning in the given context.


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |