PDOP/HDOP/VDOP and Nav Accuracy.

Irish M5 · #1 11-10-2008, 09:21 AM

Nav is sometimes off by 50ft to 100ft i.e. the turn comes before the cursor reaches the turn.

I have read that poor PDOP/HDOP numbers can cause this. Here are some photos of mine on Interstate 70 driving east and west between Dayton, Ohio and Richmond, In.

Anything look out of place?

Thanks...

Sponsored Links

KiwiJochen · #2 11-10-2008, 02:26 PM

All looks perfect. Signal levels are good, antenna is good, PDOP value of <4 is great. I've seen problems when PDOP has been around 12 - 15.

PDOP measurements which are less than 4 gives the best accuracy (under 1 meter). Between 4 and 8 gives acceptable accuracy. Greater than 8 gives poor accuracy.

Remember: accuracy is a combination of
1. GPS calculated location - which can be affected by atmospherics, and is very much afffected by current constellation. It is also affected by surrounding environment (GPS reception in mountain valleys is MUCH harder and is MUCH more inaccurate than GPS reception on a flat plain)
2. Vehicle location (as measured by wheel sensors & gyro)
3. Map accuracy (the digitised map, are the streets REALLY mapped in the right place?)

All 3 of these can be out, the GPS is constantly trying to determine which position is "more" accurate than the others.

For example, atmospherics can cause the following positional inaccuracy:
Ionosphere 0-30 meters
Troposphere 0-30 meters
Measurement Noise 0-10 meters
Ephemeris Data 1-5 meters
Clock Drift 0-1.5 meters
Multipath 0-1 meter

I've seen all of this happen. With experience I know that rainy misty days are bad :-)

As for the map, how do you know the map is right...??

PS: for MAXIMUM nav accuracy, ensure it is using the instantaeneous wheel speed signals, not the averaged speedo signal.

Irish M5 · #3 11-10-2008, 02:39 PM

Jochen, many thanks for such a complete and detailed reply.

BTW, how do you tell which wheel speed measurement signals are being used? I have a 2003 M5.

Thanks...

Quote:

Originally Posted by KiwiJochen

All looks perfect. Signal levels are good, antenna is good, PDOP value of <4 is great. I've seen problems when PDOP has been around 12 - 15.

PDOP measurements which are less than 4 gives the best accuracy (under 1 meter). Between 4 and 8 gives acceptable accuracy. Greater than 8 gives poor accuracy.

Remember: accuracy is a combination of
1. GPS calculated location - which can be affected by atmospherics, and is very much afffected by current constellation. It is also affected by surrounding environment (GPS reception in mountain valleys is MUCH harder and is MUCH more inaccurate than GPS reception on a flat plain)
2. Vehicle location (as measured by wheel sensors & gyro)
3. Map accuracy (the digitised map, are the streets REALLY mapped in the right place?)

All 3 of these can be out, the GPS is constantly trying to determine which position is "more" accurate than the others.

For example, atmospherics can cause the following positional inaccuracy:
Ionosphere 0-30 meters
Troposphere 0-30 meters
Measurement Noise 0-10 meters
Ephemeris Data 1-5 meters
Clock Drift 0-1.5 meters
Multipath 0-1 meter

I've seen all of this happen. With experience I know that rainy misty days are bad :-)

As for the map, how do you know the map is right...??

PS: for MAXIMUM nav accuracy, ensure it is using the instantaeneous wheel speed signals, not the averaged speedo signal.

KiwiJochen · #4 11-10-2008, 03:06 PM

Quote:

Originally Posted by Irish M5

JBTW, how do you tell which wheel speed measurement signals are being used? I have a 2003 M5.

Drive at a constant set speed eg 50MPH or 50KMPH, and report the wheel speed pulses

That'll tell me if you are using avg or direct pulses

If the M5 has factory nav, it'll be the instantaeneous pulses.
If retrofit, hard to tell :-)

Dataready · #5 11-10-2008, 04:29 PM

Quote:

Originally Posted by KiwiJochen

All looks perfect. Signal levels are good, antenna is good, PDOP value of <4 is great. I've seen problems when PDOP has been around 12 - 15.

PDOP measurements which are less than 4 gives the best accuracy (under 1 meter). Between 4 and 8 gives acceptable accuracy. Greater than 8 gives poor accuracy.

Remember: accuracy is a combination of
1. GPS calculated location - which can be affected by atmospherics, and is very much afffected by current constellation. It is also affected by surrounding environment (GPS reception in mountain valleys is MUCH harder and is MUCH more inaccurate than GPS reception on a flat plain)
2. Vehicle location (as measured by wheel sensors & gyro)
3. Map accuracy (the digitised map, are the streets REALLY mapped in the right place?)

All 3 of these can be out, the GPS is constantly trying to determine which position is "more" accurate than the others.

For example, atmospherics can cause the following positional inaccuracy:
Ionosphere 0-30 meters
Troposphere 0-30 meters
Measurement Noise 0-10 meters
Ephemeris Data 1-5 meters
Clock Drift 0-1.5 meters
Multipath 0-1 meter

I've seen all of this happen. With experience I know that rainy misty days are bad :-)

As for the map, how do you know the map is right...??

PS: for MAXIMUM nav accuracy, ensure it is using the instantaeneous wheel speed signals, not the averaged speedo signal.

Very nice explanation, but in addition to all of the above, there is the most important factor, the signal accuracy from the satelite which is controlled by the department of defense, owner of the GPS satelite network.

Irish M5 · #6 11-10-2008, 05:13 PM

Factory Nav so direct pulses.

Thanks...

Quote:

Originally Posted by KiwiJochen

Drive at a constant set speed eg 50MPH or 50KMPH, and report the wheel speed pulses

That'll tell me if you are using avg or direct pulses

If the M5 has factory nav, it'll be the instantaeneous pulses.
If retrofit, hard to tell :-)

trapdoor · #7 11-11-2008, 06:29 PM

You really do need to actually check that you are getting wheel sensor values though - on my E46, nav was dealer retrofit, and it had one of the faulty factory-made cable looms with the wheel sensor pin in the wrong receptacle on the plug into the nav computer.

Gave the same symptoms that it would never be accurate when approaching junctions, turn-offs etc. When it was fixed, the nav accuracy was spot on.

Check you get a reading when moving under the test menu (Wheel Sensor) and if not, probably a trip to the dealer

Irish M5 · #8 11-11-2008, 06:38 PM

Checked the wheel sensors and they change with speed ( + and - for forward and reverse). Also checked the gyro. Both are reacting consistent with information from KiwiJochen's posts.

Quote:

Originally Posted by trapdoor

You really do need to actually check that you are getting wheel sensor values though - on my E46, nav was dealer retrofit, and it had one of the faulty factory-made cable looms with the wheel sensor pin in the wrong receptacle on the plug into the nav computer.

Gave the same symptoms that it would never be accurate when approaching junctions, turn-offs etc. When it was fixed, the nav accuracy was spot on.

Check you get a reading when moving under the test menu (Wheel Sensor) and if not, probably a trip to the dealer

jonny_5 · #9 03-27-2009, 03:45 PM

have you figure out why your nav is late on turns ive recently upgraded my os to v22 /63 for the mk3,then a few days later had a custom splash screen made for me,it is a modded os ver 22/63 witch no reads ver /00 witch had to be hacked for the mk3, before i done any of the upgrades my nav was fine,so to my problem my sat nav is now 10 yards late at turnings is there any thing i can do? to recalibrat thanks in advance

KiwiJochen · #10 03-27-2009, 03:55 PM

Maybe it lost its calibration after the software load.

Take it for a calibration drive - 15min on a motorway / highway at a steady speed, plus then some time in suburban blocks, with lots of right-angle turns.

The nav should re-calibrate.

Sponsored Links