Unprompted Power Cycle

I am using the Ardusub system on a ROV for the MATE Competition with 6 thrusters in the standard vectored orientation, supplying power through a 50ft tether with 8AWG silicone wire. The electronics inside the enclosure are similar to those on the BlueROV2 but without a Fathom tether interface board. We were having trouble with the power unexpectedly cycling at our last pool test after about an hour and a half of continuous time in the water. After some testing we found out that it also cycles when switched between depth hold and manual mode very quickly. It doesn’t just cycle the power to the Pixhawk or Pi, but cycles the power to the entire system.

The power supply enters the control box and is distributed via a double bus board, with one of the connections coming off of it being the wires to the robot. There are separate systems connected to the same bus bar that cycle at the same time, so I’m guessing there is a power surge through the entire tether. However, only 12V is used and the current for each thruster is limited through PWM limits. Each thruster should only be able to draw 2A at a time because of these limits. There is a 25A fast-blow fuse before the power source, and this is still perfectly intact, so it isn’t surging like crazy. Does anyone know how to fix this or should I just drive carefully and pull the robot out of the water occasionally?

UPDATE: Did some voltage tests and found that the power supply box we were using that is said to max out at 13.8V was jumping all the way to 22V when we switched modes. Tried doing the same tests with a 12V battery and could not replicate the power cycling. We are using the Powerwerx 30A Box, and have had it for about 3 years now and I’m currently trying to find out what caused this problem.

Hi @Makenna,

Probably, the current is being limited by the power supply, making its output voltage low enough (and lasting long enough) to cycle the whole system.

To understand it better, my suggestion is to hook up some kind of data logging of the voltage (and current, if possible) with an independent power source, like a small battery, and then analyze the data with simple plots of voltage and current over time. If you can reproduce it easily, an oscilloscope monitoring the power supply’s voltage and current might be enough.

The fuse, even if it’s classified as ‘fast’, might not be fast enough to blow before the power supply limits its current, which is fine given that the power supply current limiting seems fast enough to protect against any damage.

The voltage surge you are seeing in the power supply must be a control issue, it’s current/voltage control loops are trying to give the required current at the required voltage but are failing to do the job, it might be too dynamic/fast for that control.

You can, as you suggested, try to avoid such power surges by managing your control, but it might not be fast or effective enough. But there’s another alternative:

You can add more capacitance to the input of the ESCs, which will make the current peaks be drawn from the capacitors instead of from the main power supply through the cable, possibly reducing the peak effects seen by the power supply. You can also add capacitors right in the power supply’s output terminals, some manufacturers explicitly give you hints about that, so you can try contacting them if not mentioned in their manuals, also because adding too much capacitance can make their control unstable.

When adding capacitance, keep in mind that too much capacitance will require an initial current peak to charge the capacitors, leading to more sparks when you connect/disconnect the cables, and if high enough capacitance, could end up requiring an in-rush protection circuit to limit this initial current peak. I’d be trying to play with 100 to 1000 microfarads (electrolytic capacitor) on each ESC’s input, and maybe 10000 to 100000 microfarads on the power supply. And of course, be careful with the polarity, voltage, and temperature ratings of those capacitors.

I hope you find it useful, thanks!

1 Like