New Caveatron

[David Gianforte]

Caveatron group,

Hello! My name is David and I just finished building a Caveatron here in Montana.
Thanks Joe for all the work you've put into developing and documenting this.
I have a few notes and questions:

I was able to get a genuine UT390B LRF from centralcomputer.com. This is probably old stock and they claim to have less than 10 remaining.

The 2-pin 2.54 mm JST connector is mentioned in the BoM as being used for the battery, but this connector is too large for the Adafruit battery charger. However the Adafruit battery charger comes with the correct female connector.

The BoM does not mention needing male ports for each of the JST connectors on the main circuit board. The "Single row male headers" work fine, but are not as secure.

I know this has been mentioned before but the STL files are in inches while normally they would be in mm. This would have saved me some confusion and manual scaling.

The #2 washers needed to hold down the LRF are not mentioned in the BoM. I didn't have imperial washers available but M2.5 washers worked just fine.

Has anyone had any issues from silicone outgassing damaging the lasers? My friend studying photonics mentioned this as a potential issue, given that the unit is all sealed up with silicone.

And finally and most importantly, I have been having a really challenging time calibrating the magnetic compass correctly. I did a bit of testing both in a big empty field away from metal and powerlines, and at a cave away from metal. If I use the built-in hard/soft calibration procedure, I get very wonky results. In one instance I did the calibration at a cave entrance away from all metal, and then while rotating the Caveatron through 360 degrees in manual mode, the heading only reported numbers between 200 and 360 degrees in various amounts, rather than 0 to 360.
I used the "CalShots" data and the "Caveatron Magnetometer Calibration Calculator" file to generate some manual Hard calibration values. This seemed to work much better, and I was able to get the "Azimuth Error Computation" graph to within one degree of error. Using these custom values the Caveatron heading was much better in manual mode, but was still off by up to 15 degrees in some cases. But rotational room scans result in poor data, and do not align with existing passage scans.
I plan to replace the MinIMU-9 v5 and do additional calibration and testing.

-David

[David Gianforte]

An update to the issue I was having with my magnetometer. I replaced the IMU and recalibrated, getting Hard Iron Calibration values of -1628.83130,-7869.89404,-9325.96582. These values did not work, and I was still seeing very wonky azimuth values. I then manually generated new Hard Iron Calibration values by averaging the values from the CalShots data, getting values of -788,-9340,-8672. This works much better and now the azimuth readings work as expected in Manual mode, Shot mode, in the "Caveatron Magnetometer Calibration Calculator" excel file, and when processing passages. However I was still getting bad data when processing room scans. But then I tried processing the room scans using the automatic calibration values in the imu file and that works!

I am using Caveatron 2.22, and Caveatron Process 2.2.1. While I now have a usable workaround to get good scan data, it doesn't make sense to me why the automatic calibration values do not work for me for anything except room scan processing. Or why Caveatron Process treats magnetometer data differently for passage scans and room scans.

-David

[Thomas Marguet]

Hi david,

Your 200-360 issue rings a bell. It's not 100% sure but I think I had the same issue as I misunderstood the accelerometer/gyro calibration procedure. I was doing left up and right up to the left and to the right of the axis (i.e. 90° off) instead of on the same axis as top up and bottom up but on the left and right side of the caveatron.

 

Thomas

[David Gianforte]

Thomas,

Thanks for taking a look and sending a response.

I believe I did the accelerometer calibration correctly as you describe. I was following the calibration instructions and the most recent version has pictures for the orientations on page 7. (https://github.com/Caveatron/Calibration/blob/main/Caveatron%20Setup%20%26%20Calibration%20Instructions.pdf)

Looking at the data in the accelerometer calibration excel file I filled out, the X, Y, and Z values look correct for each orientation. (Notably for the Left and Right values, the Y value in column B is the largest and smallest respectively.)

Here are my accelerometer calibration values:

-5.22199E-03,-5.47155E-03,-7.78707E-03
6.12588E-05,3.62830E-07,-2.61218E-06
-8.24966E-07,6.10937E-05,7.44043E-07
2.43501E-06,-3.46944E-08,6.11174E-05

I'm not exactly sure how they are used (probably offset values and a matrix transform), but they are close to the example values (within an order of magnitude). So I believe the orientations are correct.

I have attached my most recent accelerometer and magnetometer calibration files.

-David

[caveatron-discuss]

David,

I'm just now getting caught up and looking through old posts. Thank you for the comments - I'll reply to them in separate posts as get a chance.

I have built many units with the Adafruit Li-Ion charger board and never had a problem with the 2-pin JST connectors. I'm not sure what the issue is.

The BOM does not include every single nut and bolt and just mentions them generically at the end that more small parts required which are listed in the assembly documents (such as the LRF washers). As for the male JSTs, I overlooked that since when I ordered the wires listed from the linked source it came with the male JSTs. Since other vendors may not, you are right that that part should be called out separately.

I have not figured out how to export STLs in the correct scaling factor from Fusion 360 and have not spent time trying to find a way to convert them since its trivial to adjust in the Slicer. However, it does appear that I forgot to mention that in the printing instructions so will update them to make it prominent. If anyone knows of a way to adjust the STLs or export them correctly, please let me know.

When I assemble them, I use very little silicone on the interior and its mostly on the outside joints. I haven't had any LIDAR failures yet. The laser diodes are likely well sealed in their own enclosures inside the LIDAR so it is unlikely they would be exposed to any vapors.

I'll have to look at the magnetometer issue you are having and review the files you sent. Both modes do use the same calibration values, but in slightly different ways. In Passage mode, the Caveatron is moving so the magnetometer data is combined with the gyroscope, accelerometer, and LRF data to estimate orientation and velocity. In Room mode, since the Caveatron is only rotating, the magnetometer and accelerometer data is used directly to determine orientation without considering the gyroscope or performing additional processing. Try v1 processing mode and see what happens with the Passage mode scans. That version of processing did not combine the mag and accel data but used them separately and more directly.

Joe

[caveatron-discuss]

David,

I looked through the two files you sent and the acc calibration looks fine. However, the mag cal file has a rather large offset on the raw Y values. These should vary plus and minus from zero as you rotate it to the different angles for the calibration. This implies that some internal or external electromagnetic field may be influencing the data and preventing it from calibrating properly. The hard iron mag cal values are quite high.

Attached is a text file with example raw mag and mag calibration values from two Caveatrons as an example of what might be more typical.

Joe

[David Gianforte]

Joe,

Thanks for taking a look. I wonder if my batteries are causing too much magnetic interference. I'm using two 18650 batteries. Both the Caveatron unit as a whole and an individual 18650 cell cause a magnetic compass to deflect significantly. I did testing out in two different empty fields and also at a cave entrance, so I don't think it's external interference.

I have attached a part of my test survey with one passage scan and one room scan.

I reprocessed the data as you described and got the following results:

Using the unit's automatic mag calibration values of (1488.55042,-7263.24805,-8078.55078) I get:

With V1 processing, passages process correctly.

With V1 processing, room scans process incorrectly.

With V2 processing, passages process incorrectly.

With V2 processing, room scans process correctly.

Using my manual mag calibration values of (2440.00000,-8500.00000,-8000.00000) I get:

With V1 processing, passages process correctly.

With V1 processing, room scans process incorrectly.

With V2 processing, passages process correctly.

With V2 processing, room scans process incorrectly.

-David

[Brett Wiltshire]

Hi David, I feel your pain on the compass issue, I had a failed caveatron rev A build that worked except for just the compass.  Replacing the LSM303D did not fix it, I did not follow the instructions in building it so I never could work out what caused it and the improvements to the design since make it a bit pointless to continue with that build. 

I also had the compass fail on a working DistoX3, (sorry to go a little off topic, please bear with me) same symptoms as you describe, very different compass bearings when pointing in the same direction and just rotating the device.  And compass will not show the full 0-360 range.  I have replaced the compass board on the DistoX and also given another board purchased at the same time to a colleague to install in a brand new disto unit without any input from me.  Both units still give nonsense compass readings. 

I am told that an electronic compass can not be damaged by putting them in a strong magnetic field (by PNI sensor, the manufacturer that make the compass in the DistoX3 board) but other nearby components could become magnetized.  Some screw drivers for example can have quite strong magnets built into them.  I dont know if this is significant.

Ive since tried buying a cheap degaussing / demagnetising machine like that used by watchmakers to repair mechanical wrist watches.  Initial tests seemed to show an improvement.  I still dont have a working device but I have not had time to properly test in a cave yet and there are also factors to test around how the degaussing machine is used.  

This solution does not really make sense to me, the magnetic field of a screw fixed to the case of a compass should be constant and not that much.  But that is what I have resorted to trying out of frustration and I am just throwing this out there in case it helps.

Brett Wiltshire

--
You received this message because you are subscribed to the Google Groups "caveatron-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caveatron-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caveatron-discuss/f667be6a-acc3-42c8-bd03-f9ac01a41eb2n%40googlegroups.com.

[Thomas Marguet]

I know Dave installed the Uni-T LRF, but maybe that could help others as well: the JRT B707C has a strongly magnetized buzzer - i.e. the small black cylinder on top of the pcb. Using an EMF app and my Iphone, I could read over 1500microT. I have unsoldered the tiny buzzer and replaced it with the same 5V buzzer type I purchased for the caveatron. This one doesn’t seems to disturb much the magnetic field. My caveatron is into pieces as I try to test some improvements but will let you know if I see some improvements.

Fuel for thoughts.

Cheers

Thomas

To view this discussion on the web visit https://groups.google.com/d/msgid/caveatron-discuss/6a72ec30-26f2-f1c5-646a-2ed13df819a8%40gmail.com.