Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Andy Green wrote: Somebody in the thread at some point said: | I'm using OM2007.2 updated to today morning (Thursday July 31), and | TangoGPS. | Also have the map cache directory on the card. Normally works fine, but | when I move indoors, | gps shows 10 satellite fixes and gives my speed as 482 Km/h (really | it did). | | So either I broke the land speed record for a guy on foot ... or I think | I will have to have that | capacitor fix. I dunno how a 10pF capacitor on SD_CLK is going to fix the GPS chip acting erratic when it genuinely loses the satellites as you move indoors. FWIW you'll need to wait for tomorrow's kernels to get yesterday's two fixes in there. - -Andy Hi Andy, I'm the guy who said that. Actually I based my observation on the fact that i saw an entry of 10/x/y as the endmost entry on the TangoGPS status line. Isn't that supposed to mean that it still has fixes from 10 satellites. So I thought it must be a SNR issue and the capacitor fix is supposed to help on that ... right? This was all guesswork on my part. If somebody could help with how to make observations on GPS performance, I'd be glad to do testing and provide input. The kernel update is going on as I type this. So will test again today if possible. And don't worry ... will be doing the hardware fix only after OM finally, after all software fixes, decides it will still help / otherwise not. -- Vibhav ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Hi there! I'm not Andy obviously, but I think I still can answer that one ;) Hi Andy, I'm the guy who said that. Actually I based my observation on the fact that i saw an entry of 10/x/y as the endmost entry on the TangoGPS status line. Isn't that supposed to mean that it still has fixes from 10 satellites. So I thought it must be a SNR issue and the capacitor fix is supposed to help on that ... right? Actually, the first number in 10/x/y in TangoGPS' status line represents the number of visible satellites, not the number of fixes you got. So if it says 10/0/x, that means you can see 10 satellites, but don't have a fix (because, for example, the signal is too weak). So it's quite normal you didn't get a gps trace if the second number was 0. HTH, Konstantin ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Somebody in the thread at some point said: | Actually I based my observation on the fact that i saw an entry of | 10/x/y as the endmost entry on the TangoGPS status line. | Isn't that supposed to mean that it still has fixes from 10 satellites. Our GPS chip is quite capable to report garbage as fact, for a while anyway. I also saw things like 12 satellites each with individual SNR reported, when I inserted or removed internal antenna while powered (something you aren't meant to do BTW). After some seconds these disappeared from the reporting back to the virtual reality they came from. So madness under extreme circumstances is normal for it I believe. | So I thought it must be a SNR issue and the capacitor fix | is supposed to help on that ... right? At the time the cap workaround was conceived, we ran SD_CLK all the time even with the card idle and did not test for what it does during card transactions. Since then, the way we treat the card has gotten seriously more optimized: ~ - SD_CLK is stopped unless we are making an access ~ - Voltage to shipping SD Card is reduced to 2.7V from 3.3V ~ - Edges on all SD signals have reduced slew rate ~ - SD_CLK rate is reduced if GPS unit is powered The reason I avoid recommending the cap is there are 6 signals going to the card, not just SD_CLK, and during transfer 5 of them have a lot of rapid transitions. The reasoning for cherrypicking SD_CLK is no longer entirely valid. | This was all guesswork on my part. If somebody could help with how to | make observations on GPS performance, I'd be glad to | do testing and provide input. I'm afraid I don't know the userspace apps. I literally cat the serial port and read the NMEA. Maybe someone else can advise, this Tango GPS thing seems to be widely admired. | The kernel update is going on as I type this. So will test again today | if possible. And don't worry ... will be doing the hardware | fix only after OM finally, after all software fixes, decides it will | still help / otherwise not. Great, more info the better we can understand where we are at. - -Andy -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iEYEARECAAYFAkiSrNcACgkQOjLpvpq7dMpomACdGwqta90evIBOQdaKfiJT8jt4 FV0AoI2NVwAiSoxG2uEKCSEc+Dmj+wWN =w8aN -END PGP SIGNATURE- ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Vinc Duran schrieb: But... what are our chances GTA03 will have a 3D gyro? Could the accelerometers be substituted for a 3d gyro? No, you need both accelerometer *and* gyro measurements for dead reckoning. Andreas ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
On Friday 01 August 2008, Vinc Duran wrote: On Thu, Jul 31, 2008 at 5:28 PM, Andreas Bogk [EMAIL PROTECTED] wrote: snip I'm not a mathematician, but I think I have sufficiently understood Kalman filters and dead reckoning. A little known trick is to use ordinary integration with the strap-down equation given the accelerometer and gyro input, and have the Kalman filter model the error terms instead of the physical model state space. This way, you only have to update the Kalman filter every time you get a GPS fix, not every time you get a measurement from the other sensors. But... what are our chances GTA03 will have a 3D gyro? Andreas Andreas, Could the accelerometers be substituted for a 3d gyro? We only need to replace 1 of the 2 3D accels with a 3D gyro. I've not looked at the size or cost of solid state gyros so I don't know how practical that might be. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Al Johnson schrieb: We only need to replace 1 of the 2 3D accels with a 3D gyro. I've not looked at the size or cost of solid state gyros so I don't know how practical that might be. Well, if this: http://www.invensense.com/products/idg_300.html had three axes instead of two, it would be suitable... it's designed for small size and low cost. All other gyros I've found so far are too big for a mobile phone application. Andreas ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Al Johnson [EMAIL PROTECTED] writes: We only need to replace 1 of the 2 3D accels with a 3D gyro. I've not looked at the size or cost of solid state gyros so I don't know how practical that might be. That's not a bad idea... I haven't looked into what accelerometers we have in the FR, but there are plenty of MEMS gyros available; maybe one will have the same or similar pin configuration. I know the gyros and accelerometers made by Analog Devices are very close to pin compatible. Last time I checked (about a year ago), there weren't any 3D MEMS gyros on the market. There was one company offering what was basically 2 2D gyros glued to each other. I think there's some kind of manufacturing hurdle that has to be overcome to make them 3D, so they may not be available in a small size or integrated unit. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Do folks know wher to find the SOP for the cap fix? Tony, if you havent already, go ahead and post the SOP for the cap fix. It's easy enough to do And undo. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Jay Vaughan Sent: Wednesday, July 30, 2008 1:14 AM To: List for Openmoko community discussion Subject: Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue I haven't had any problems with TTFF since the software update was done. In fact my impression from all the emails on the subject was that the software fix was enough. Try putting tangogps in a configuration where its updating the tiles frequently - zoom way in, so that you have a closeup of your map, and then walk around a bit. In between tile updates, GPS will go way off the map. Thats the problem: its a workaround, but not a complete, effective fix. Looks like we have to have the capacitor-fix in place to make this work smoothly without further glitches .. ; -- Jay Vaughan ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Hmm. Jay... Amaybe when SD I/O occurs one could use the accelerometers to dead reckon? They are noisy and drift but for short periods MIGHT be able to keep wild errors in mapping from occuring. Just a thought. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Jay Vaughan Sent: Wednesday, July 30, 2008 1:24 AM To: List for Openmoko community discussion Subject: Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue (We have external LNA) It means it can track with a 16dBm reduction in GPS signal after it made the satellite acquisitions. So if you understood it will get a fix OK with the software workaround, and we saw many people report this, you should expect the tracking phase to work better than the acquisition phase since the signal requirement is relaxed by 16dBm. My experience has proven that this is not the case; yes the fix happens sooner, but if SD I/O occurs during tracking phase, wild errors are still introduced into the GPS signal sufficient to drive us off the map. Just sayin', not trying to stir the pot. I've been testing this GPS fix every day since it arrived, by tracking my progress to work, and if I zoom too closely (meaning there is more SD I/O on map refreshes) the GPS goes wild. That'll be interesting. But in the real world many thing perturb GPS tracking and sophisticated software filters are anyway used to try to see through excursions, I guess this will be the case for typical SD Card usage pattern here. My ride to and from work is through a wide open space with no obstructions, lots of blue sky, and for sure .. SD I/O still massively impacts GPS *accuracy*. Not just in the acquisition phase, but also tracking .. ; -- Jay Vaughan ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Somebody in the thread at some point said: | Do folks know wher to find the SOP for the cap fix? | | Tony, if you havent already, go ahead and post the SOP for the cap fix. It's | easy enough to do | And undo. Did you ever do it, steve? - -Andy -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iEYEARECAAYFAkiRzVQACgkQOjLpvpq7dMpGLwCeO+mNJpn0Fk/bMnaoSN3n6o5a 1bEAniOPccvfrk8WZ+/lIUjVlxUE5pZd =xFkQ -END PGP SIGNATURE- ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
steve schrieb: Amaybe when SD I/O occurs one could use the accelerometers to dead reckon? Dead reckoning without gyros is pretty hard. In theory, the two accelerometers are not mounted in the same place and thus could be used to derive rotation, but in practice they're pretty close to each other, and given the noise of measurement one is unlikely to derive a good rotation from them. Andreas ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Somebody in the thread at some point said: | steve schrieb: | Amaybe when SD I/O occurs one could use the accelerometers to dead reckon? | | Dead reckoning without gyros is pretty hard. In theory, the two | accelerometers are not mounted in the same place and thus could be used | to derive rotation, but in practice they're pretty close to each other, | and given the noise of measurement one is unlikely to derive a good | rotation from them. I don't think we need to worry about it, but in fact we wouldn't need to dead-reckon. All we would need it for would be to put a ceiling on claims from noisy GPS about what accelerations we could have experienced. If the GPS sample says we jumped 100m in 1s, but the accels didn't really see anything, we could ignore or meddle with the GPS sample. Whether that is any better than just filtering between previous and later samples I doubt. - -Andy -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iEYEARECAAYFAkiRz5EACgkQOjLpvpq7dMpNLQCfcPdC9ZdSelPGiP+D9TVapDI0 94AAnjkmnCWge5RxXewJ40gchhA1n+Ne =jCLG -END PGP SIGNATURE- ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Andy Green wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Somebody in the thread at some point said: | steve schrieb: | Amaybe when SD I/O occurs one could use the accelerometers to dead reckon? | | Dead reckoning without gyros is pretty hard. In theory, the two | accelerometers are not mounted in the same place and thus could be used | to derive rotation, but in practice they're pretty close to each other, | and given the noise of measurement one is unlikely to derive a good | rotation from them. I don't think we need to worry about it, but in fact we wouldn't need to dead-reckon. All we would need it for would be to put a ceiling on claims from noisy GPS about what accelerations we could have experienced. If the GPS sample says we jumped 100m in 1s, but the accels didn't really see anything, we could ignore or meddle with the GPS sample. Inertial navigation systems for the masses! Anyone here who wants to write a kalman filter? Besides the point that a small pocket-IMU would be great. To GPS jumts that big exist? As fas as i understand GPS it should have little jitter. -- Drucken Sie diese Mail bitte nur auf Recyclingpapier aus. Please print this mail only on recycled paper. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
This one? http://wiki.openmoko.org/wiki/Image:Gta02_gps_10pf_rework_sop.pdf On Thursday 31 July 2008, steve wrote: Do folks know wher to find the SOP for the cap fix? Tony, if you havent already, go ahead and post the SOP for the cap fix. It's easy enough to do And undo. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Tilman Baumann wrote: Andy Green wrote: Inertial navigation systems for the masses! Anyone here who wants to write a kalman filter? Besides the point that a small pocket-IMU would be great. To GPS jumts that big exist? As fas as i understand GPS it should have little jitter. Hi, I'm using OM2007.2 updated to today morning (Thursday July 31), and TangoGPS. Also have the map cache directory on the card. Normally works fine, but when I move indoors, gps shows 10 satellite fixes and gives my speed as 482 Km/h (really it did). So either I broke the land speed record for a guy on foot ... or I think I will have to have that capacitor fix. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Besides the point that a small pocket-IMU would be great. To GPS jumts that big exist? As fas as i understand GPS it should have little jitter. Trimble had a few GPS models out with dead reckoning. Not pocket sized. I think they are all off the market now. I got one off Ebay a few weeks ago but have not had time to set it up yet. To me it's more interesting reading the manuals and learning how it works since I don't use my car more than once a month. Trimble was addressing loss of signal in urban canyons not (ahem) antenna placement issues. See http://www.trimble.com/tms/placer450.aspx?dtID=overview They use a small gyroscope. There are much smaller gyros available now but my intuition from reading their materials is that it will work in a car if you are going fast enough (they want you to go 22 mph) to generate steady predictable g forces but if you are going at a walking pace it would never work. If you are walking around and want a reasonable position the Kalman filter would probably be easier to deal with. Is there a mathematician in the house? Brian ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Somebody in the thread at some point said: | I'm using OM2007.2 updated to today morning (Thursday July 31), and | TangoGPS. | Also have the map cache directory on the card. Normally works fine, but | when I move indoors, | gps shows 10 satellite fixes and gives my speed as 482 Km/h (really | it did). | | So either I broke the land speed record for a guy on foot ... or I think | I will have to have that | capacitor fix. I dunno how a 10pF capacitor on SD_CLK is going to fix the GPS chip acting erratic when it genuinely loses the satellites as you move indoors. FWIW you'll need to wait for tomorrow's kernels to get yesterday's two fixes in there. - -Andy -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iEYEARECAAYFAkiSDf0ACgkQOjLpvpq7dMq0iQCdEkhemui8JzNoBt3NeyOGD55P /nkAni7SJUfnNdS+cOzZL2l9PmTQ+X8X =TUti -END PGP SIGNATURE- ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
Brian Wilson schrieb: See http://www.trimble.com/tms/placer450.aspx?dtID=overview They use a small gyroscope. [...] If you are walking around and want a reasonable position the Kalman filter would probably be easier to deal with. Is there a mathematician in the house? I'm not a mathematician, but I think I have sufficiently understood Kalman filters and dead reckoning. A little known trick is to use ordinary integration with the strap-down equation given the accelerometer and gyro input, and have the Kalman filter model the error terms instead of the physical model state space. This way, you only have to update the Kalman filter every time you get a GPS fix, not every time you get a measurement from the other sensors. But... what are our chances GTA03 will have a 3D gyro? Andreas ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Warranty after fix for GTA02 GPS rework for SD card interferenceissue
On Thu, Jul 31, 2008 at 5:28 PM, Andreas Bogk [EMAIL PROTECTED] wrote: snip I'm not a mathematician, but I think I have sufficiently understood Kalman filters and dead reckoning. A little known trick is to use ordinary integration with the strap-down equation given the accelerometer and gyro input, and have the Kalman filter model the error terms instead of the physical model state space. This way, you only have to update the Kalman filter every time you get a GPS fix, not every time you get a measurement from the other sensors. But... what are our chances GTA03 will have a 3D gyro? Andreas Andreas, Could the accelerometers be substituted for a 3d gyro? ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
