Page 1 of 1

Testing a Digital Compass

PostPosted: Nov 4, 2013 9:43 am
by caver.adam
I'm getting ready to test a digital compass and am wondering whether anyone on here has any experience with testing equipment. I seem to remember another publication that did caving electronics, but can't remember their name or if they had a message board.

My plan is to break the testing into 3 phases on a figure 8 course with 30 stations.
1) Digital compass and inclinometer pointed directly from target to target.
2) Suunto Tandem used per standard survey teqnique (with backshots)
3) Digital compass used while level, and inclination while pointing directly from target to target.

For my testing I am planning on using multiple groups.
A) 3 highly experienced surveyors each run the course.
B) 3 cavers with at least 3 days of survey but less than 30 days each run the course.

Using information gathered from these 18 runs I should be able to determine how well the digital compass performs compared to the Suunto while pointed directly at targets or while maintained level for compass. I should also be able to perform a Gage R&R to determine the repeatability of the measurements for different operators.

Any thoughts?

Re: Testing a Digital Compass

PostPosted: Nov 4, 2013 10:16 am
by trogman
Sounds like a good plan, especially doing a Gage R & R to analyze the results. Are you using MiniTab for this? I wonder how good the results would be using standard non-digital survey equipment.

What brand and model compass are you using for this? Most digital compasses I've looked at in the past have had very poor accuracy, on the order of +/- 5 degrees. The ideal would be to have one with a built-in laser, but to my understanding that's a tall order because of issues with the batteries and electronics interfering with the compass. Maybe that's why the accuracy ratings on most digital compasses are so poor.

Trogman :helmet:

Re: Testing a Digital Compass

PostPosted: Nov 4, 2013 10:22 am
by Scott McCrea
Sounds like a good opportunity to get feedback on the usability of the instrument, also. Accuracy is important, but if it is not easy to use, easy to read, easy to sight at awkward stations, durable, water resistant... it's not useful.

Re: Testing a Digital Compass

PostPosted: Nov 4, 2013 3:23 pm
by Extremeophile
What your proposing sounds like design of experiment (DOE). As long as you think through the variables, levels, and keep the experiment balanced, then you'll be able to analyze the components of variation. I've done similar testing with a DUSI digital compass and clino in an effort to better understand the electronic parameters that best minimized measurement variation. I used a Taguchi approach, which utilizes an inner and outer array. The inner array contains experimental parameters that can be controlled (e.g. type of instrument, reading technique, instrument calibration method). The outer array contains "noise" variables that typically are not controlled (e.g. direction and vertical orientation of measurement, person taking measurement, ambient conditions). By analyzing the data this way you can determine the best way to calibrate and use the equipment regardless of who is using it and what measurement they are taking. If you're designing or developing a digital survey instrument then additional inner array variables may include the type of magnetometer, the type and location of batteries, etc. It may be necessary to have a "true" value for your survey course so that you can evaluate both accuracy and precision. An approach such as Gage R&R only looks at precision. Having a course with fewer than 30 stations might be advisable.

Re: Testing a Digital Compass

PostPosted: Nov 4, 2013 7:40 pm
by Bob Thrun
The very first digital compass on the market was intended for boaters. I have one that I bought from another caver after he tried it. It has rudimentary sights and a single magnetic sensor. It reads the direction of the magnetic field in the plane of the sensor. It does not measure the direction of the magnetic field in a horizontal plane. When the compass is tilted side-to-side or front-to-back, the reading varies wildly. Boaters are able to level the compass with the horizon.

Brunton has a digital compass with a different sensor and similarly poor sights. I looked at an early one. The maker of the magnetic sensor gave more information on the compass than Brunton did. The sensor was supposedly gimballed to keep it level, but the readings still varied when I tilted the compass. I called Brunton to suggest that they put the digital module in a pocket transit with its circular level and good sights. Brunton had already marketed two pocket transits with digital readout for an analog compass. I was told that the digital sensor would not be as accurate as the analog pocket transit.

Re: Testing a Digital Compass

PostPosted: Nov 6, 2013 12:04 pm
by caver.adam
I'm actually using an LSM303DLHC (http://www.pololu.com/product/2124) that I'm programming with an Arduino Micro. My current status is that I'm building a protoboard version and am almost through wiring all the jumpers. Then I need to finish the calibration routines for the accelerometer and compass. But I'm trying to get ahead of things on planning out my testing because I can plan out my testing even when I can't physically work on the hardware.

Re: Testing a Digital Compass

PostPosted: Nov 6, 2013 12:08 pm
by caver.adam
Oh, and using an onsoldered breadboard connected to a CPU via a USB cable with an old regulator that had been overvolted....I was seeing jitter of about 1 degree (+/- 0.5) on my readings. The hard soldered one should fare better. I still don't know anything about temperature drift on this chip either.

Re: Testing a Digital Compass

PostPosted: Nov 27, 2013 3:34 pm
by caver.adam
For anyone interested in the project, the current hardware is listed here: http://projectsampson.weebly.com/rev-c- ... table.html . I'm afraid to go out and get a tripod mount on thanksgiving weekend, but will need one to do some testing. So far seems to be giving pretty good results.

Re: Testing a Digital Compass

PostPosted: Nov 27, 2013 4:57 pm
by LWB
I'm not at all familiar with that company that makes that combo 3 axis magnetometer / accelerometer. I've tested several cave survey compass/clinometers using the Honeywell and the PNI magnetometers. The last go round with Honeywell magnetometers (a prototype) seemed to indicate a temperature sensitivity that was causing the calibration to deteriorate with temperature variation. The PNI magnetometers never had this trouble. That is why the new version of the Disto X, went back to the PNI magnetometers.

I tested by using a modified non-magnetic Brunton tripod with a custom fixture(s) to hold the instrument (tested Disto X, DUSI and Shetland Attack Pony). I have a course set up in a small cave with markers on walls, floor and ceiling for targets, and I compared to Theodolite data shot on the same course. I use some of the markers for calibration also.

I had a survey course set up (about 15 station loop), but it was much harder come to conclusions from that data. I could get good closure from an instrument that couldn't produce consistent agreement in foresights and backsights.

Here is photo of tripod with the first prototype of the second generation Disto X on it.



Image

Re: Testing a Digital Compass

PostPosted: Nov 27, 2013 11:14 pm
by caver.adam
That's excellent information! I had seen that you were using the PNI, and was curious how the ST would perform. I'm also curious from the standpoint that the ST had an Arduino library on it. I was just noticing how hard it is to hold a compass steady...

Re: Testing a Digital Compass

PostPosted: Nov 28, 2013 6:22 am
by LWB
Not me using the PNI, but Beat Heeb who makes the Disto X. I only test and use them, not build them.

For testing you definitely need a tripod.

Re: Testing a Digital Compass

PostPosted: Nov 28, 2013 12:08 pm
by caver.adam
Well, the PNI RM3000-f appears to have a noise of +/- 0.25 degrees or so. I can't seem to find the resolution on that particular sensor either. But it appears to cost around $40 for the suite.

The ST LSM303DLHC costs $18 on an evaluation board and has a +/- 0.80 degree noise and close to a 0.25 degree resolution (in Kentucky). So my limiting factors are actually noise, and drift.

Re: Testing a Digital Compass

PostPosted: Nov 28, 2013 1:38 pm
by LWB
I'm calibrating 4 or 5 Disto Xs today. If your device will fit on my tripod you are welcome to use it and the test course tomorrow. If you aren't too far from Hamilton Valley...

Re: Testing a Digital Compass

PostPosted: Dec 1, 2013 9:26 pm
by caver.adam
Unfortunately, I think I'm pretty far. We are working on setting up a test course for December. Was checking an old map compared to actual this weekend and didn't get on cavechat.