Replies to This Discussion

Permalink Reply by Tom Yochum on July 27, 2010 at 2:57pm

How much was it moving about? It is expected to move around +-10 meters.

Did the error distribution drop when you started walking?

• ▶ Reply

ModeratorPermalink Reply by Ken on July 27, 2010 at 3:43pm

1. Yes, it will show movement about a few meters. It is a fun project to collect to collect the data for a day or so then plot the points on a 3D graph.

2. I'm not sure what, if any, WAAS satellites (or their European equivalents if there are any) the EM-406 may be able to use. So your accuracy may be "un-augmented" giving 2x or 3x the error I get over here.

Probably not affecting you, but a general note to those users seeing less accuracy this summer than they had before, 1 of the 2 WAAS satellites that serve North America went out of service this April. I've not heard about a replacement / repair yet.

• ▶ Reply

Permalink Reply by simonl on July 27, 2010 at 4:12pm

Thanks for the swift replies.

@Tom: Loads of movement - in the order of 100m or so. When I was walking it was pretty accurate though.

@ Ken: Good point about WAAS. I have no idea. I think the lack of accuracy is greater than 2-3x though.

I'm off on a car journey tomorrow, so I'll take it with me and see how it fairs in a different part of the country.

Thanks for you r help.

Cheers,
Simon

• ▶ Reply

ModeratorPermalink Reply by Ken on July 27, 2010 at 4:37pm

I went to look up the specs for the EM-406A and it is supposed to receive EGNOS, the European "WAAS" and MSAS, the Asian "WAAS." However, one document implied that you had to send a command to enable EGNOS or MSAS. If I have time I will look into that, I surely do not recall a SiRF command that enabled it but then, it isn't something I'd be looking for.

I checked on the April loss of the WAAS satellite to get an update. Only SW Alaska should be without access to the remaining WAAS satellite. But there is some kind of switchover that the satellites or the ground stations do 5x per year that shuts off WAAS for 5 minutes. Normally not an issue since a second bird was normally available. Not so any more. So, aside from Alaska, North American users should have WAAS except for 5 minute outages 5x over a year. Replacement satellite to be launched December. Online after checkout.

A 100 meter error is not normal. That sort of thing happens when you've got really bad satellite geometry. Is your horizon obstructed (to microwave signals) up to 20 or 30 or more degrees above the horizontal? What does the GPS say for HDOP? (Look in the $GPGGA message for the 6th field, after number of satellites, before the altitude.)

• ▶ Reply

Permalink Reply by simonl on July 28, 2010 at 10:43am

@Ken: Thanks for your further thoughts. I think you hit the nail on the head regarding satellites in view - my testing last night was with the GPS just outside my window, so much of the sky would've been obscured by my house. Today's test didn't show the +/-100m inaccuracy; only about +/-10m whilst stationary.

So do others experience reported "movement" of +/-10m whilst stationary? And how does that play out with autopilots for 'copters?

Thanks again.

Cheers,
Simon

• ▶ Reply

Permalink Reply by Tom Yochum on July 28, 2010 at 11:10am

Yes, +- 10 meters is typical assuming non-SBAS (WAAS, EGNOS, etc).

I can't comment with authority on the helicopter autopilot topic since I have never created one, but I can guess that one would use other means to stabilize a helicopter in hover. An IMU and optical flow come to mind. Some of the big "real-world" helos use doppler radar measurements.

• ▶ Reply

Permalink Reply by Martin Seven on July 28, 2010 at 11:34am

We combat GPS inaccuracy using IMUs and filters that combine the GPS data (absolute = not drifting, but noisy) with IMU data (relative = drifting, not noisy) to get the position solution.

Additionally you can use simple filtering like rolling averages to smooth out the jaggies in the GPS signal.

• ▶ Reply

Permalink Reply by simonl on July 28, 2010 at 2:57pm

Thanks Tom / Martin, your confirmation of this being "normal" is both what I'd hoped and disappointing - the former because it makes me think it's not my GPS; the latter because I was hoping the GPS would show a stationary reading even if it was out by +/-10m.

Thanks again to all. Now onward to figuring out the math :-(

Cheers,
Simon

• ▶ Reply