Home        Store        Learn        Blog

Odd Heading-Hold / Depth-Hold behavior with Ardusub 4.0.1

Hi All:

Last week I tested a newly-built BROV2 (6-thruster configuration), and noted an odd problem. The thrusters all worked perfectly fine in manual mode, but if the vehicle was in either stabilized mode (HH or DH/HH) then the gain on the vertical channel (climb/dive) was suddenly greatly reduced. I still had control of the vehicle, since it was very close to neutrally-buoyant. But even with full stick, the vertical thrusters were only using about half an amp of current. I tested this at 25% throttle gain and 50% throttle gain, and the results were similar. There were no issues at all in the horizontal plane with the 4 vectored thrusters.

Software configuration was QGC 4.0.5, Ardusub 4.0.1, Companion 0.0.20.

One thing that might be relevant here is that the vehicle was set up for deep diving, with a Bar100 (1000m) depth sensor. I didn’t make any changes to Ardusub to support the 1000m depth sensor- it appeared to give correct depths out-of-the-box. Maybe there’s a config change needed in the software somewhere?

This was the first vehicle I’ve driven with Ardusub 4, but I don’t remember this ever being an issue with Ardusub 3.5. Without reverting to Ardusub 3.5 while on the water it’s tough for me to know whether this is related to the software build or the specific depth sensor.


Walt Holm

Hello Walt, can you please upload a parameter file from your vehicle, and when you get a chance, a short telemetry log to demonstrate the problem.

You may check the JS_THR_GAIN parameter which seems to be involved, the default value is 1.0.

Hi Jacob:

Attached is the parameter file. Unfortunately the log file from the day in question is huge. I’m going to try to recreate the issue in my bathtub, hopefully that will work.


Holm_DeepROV_Params_20200709.params (25.9 KB)

Hi Jacob:

So I got an interesting result when doing the bathtub test. It looks as if one of the vertical thrusters has an issue, maybe in the ESC, maybe in the wiring, but for whatever reason it barely works at all. This certainly wasn’t an issue when I dry-tested the ROV upon initial assembly, and figured out the correct thruster directions.

So, when you engage HH or DH/HH, does it automatically try to level the ROV? If so, that might explain some of the behavior- but it wouldn’t explain why the ROV worked normally in manual mode. Sigh, looks like I’ve got to crack open the main tube and fix something there before I get you better test data.

I’ve attached the tlog from the bathtub session, just in case you can see anything in it.2020-07-09 12-26-34.tlog (1008.7 KB)

Hi Walt - yes the vehicle tries to straighten itself out to be perfectly level in stabilize and depth hold modes. In manual mode, it doesn’t perform any sensor-based corrections, it just passes the pilot’s joystick movements to the motors.

Were you at sea when you experienced the issues in your first post?

When you do encounter the problem in some deeper water, please send a dataflash log. I’m sorry I should have asked for that instead of telemetry.

Hi Jacob:

Thanks for the reply. Yes, I was at sea when I encountered the stabilization issues. That was before there were any thruster issues, as the vehicle worked perfectly fine in manual mode.

I took a look at the dataflash logs on the vehicle’s Pixhawk, and, similar to the Telemetry Logs, the files are pretty huge. There was also the issue that, after some logs that were taken in February (hmm. That was before this ROV even existed. I must have been using the Pixhawk for something else then.), all of the logs are marked as “date unknown”. Is that normal in ROV applications, as the Pixhawk doesn’t have a GPS connection?

I’ve set aside some time tomorrow to crack open the 4" tube and see what’s going on with the bad thruster. Assuming I get that resolved, I’ll try to do another run in the bathtub and get you a concise dataflash log.

Alternatively, I could get you some logs that are ~50 MB in size that would show what happened in the field. I just don’t know how practical that option is.

Walt Holm

Hi Jacob:

I cracked open the ROV today, it turns out there was an open winding in the thruster of the bad channel. Easy to diagnose, but will take a bit longer to fix. Every T200 I have is in a project right now, so I’ve got a new one on order. And then I’ve got to install/pot the custom penetrator design that I’m using on this ROV.

So it will probably be a couple of weeks before I can do a bathtub test and hopefully get you some telemetry that re-creates the original issue I saw.

Walt Holm

Hi Jacob:

So I got a new thruster in the ROV, and last week I managed to do a lake test with it. I was not able to reproduce the problem. So the issue was probably a creeping failure of a phase wire in the one vertical thruster, coupled with the leveling function of the stabilized mode.

I will keep an eye on this, and repost if I see it again. But I think you can consider it closed for now.

Walt Holm

1 Like