NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
From: Paul Dolkas
Date: 2017 Jul 24, 12:55 -0700
Don-
If temp drift is a problem, then one solution is to include a temp sensor and then have the software compensate based on a look up table of gain vs. temp. The digital altimeters they put in watches have this problem, and that’s how they got around it. I always wondered why these watches included a thermometer (why do I need to know how warm my wrist is?) until somebody explained that the sensor was included to improve the accuracy of the altimeter.
Paul Dolkas
From: NavList@fer3.com [mailto:NavList@fer3.com] On Behalf Of Don Seltzer
Sent: Friday, July 21, 2017 8:54 AM
To: paul@dolkas.net
Subject: [NavList] Re: New ways to create false horizons?
On Thu, Jul 20, 2017 at 6:13 PM, Pete Solon Palmer <NoReply_Palmer@fer3.com> wrote:
Maybe there is an electronic AH solution. The sensors (mag, gyro, accel, gravity) in my smart phone seem to be less accurate than a ball bearing and a toy compass, so I don't see a solution involving phones soon.
Ever since I took Frank's course I have been toying around with this idea. Some years ago I developed some automated test stations for measuring performance of multi-axis MEMS sensors such as those used in phones and other devices today.
A two axis MEMS accelerometer is very plausible for determining the gravitational vertical. The three main error sources are bias offset, scale factor error, and temperature drift.
Bias offset is easy to deal with, and is conceptually similar to sextant index error.
Scale factor error is a problem for measuring a wide range of angles. When simply trying to determine a very small range around 'zero' angle it is of minor concern.
The most challenging problem is temperature sensitivity of these devices. Changing temperatures will cause all the other parameters to shift.
Analog Devices is a leading manufacturer of MEMS multi-axis accelerometers, some costing less than $20. I once did a rough error analysis of some of their devices and it seemed that it might be possible to achieve a resolution of about 1 or 2 arc minutes around zero.
Don Seltzer
