New Sub Designs

A few hours of hacking away at AutoCAD Inventor led to this design for the robosub chassis. Notice the rendering and exotic tundra. I do not think that the sub will be effective on land, but it had much more dramatic shadows in this location.

I have generated some drawing file for the above components and we will attempt to cut them as soon as possible. The next step is to make sure that the motor controllers are working.

Corrosion test

After the thrust profile test was completed, I did not have enough time to dismantle the experimental set up, so I left it in our lab for a day until I could return to it. Upon extraction, I was much surprised to find the level of corrosion present on the DC-DC converter and 2 of the 3 shaft couplers on the motors.

 Corrosion on the DC-DC converter

Corrosion on the shaft coupler

The rate of corrosion is really the interesting part of this test. The parts were likely submerged for 36 hours and look as if they are starting to grow barnacles. In order to prevent this, I believe we should apply a coat of acrylic paint to surfaces which will be continuously exposed to the water.

If it works for ships, it should work for us.

Thruster with new power supplies

We exchanged the original power supply: (see photo)

 

For two, also old but better power supplies:

 The difference lies in the current ratings. The Cenco power supply is only rated to provide 3 amps at 50 volts, whereas each Lab-Volt can handle up to 10 amps. The reason the two are wired in series is because the Lab-Volts can only provide up to 36 Volts. Neither power supply can provide for all of our needs in the thruster profile, so two in series is the way to go.
 

Like last time, the thrusters were immersed in water to simulate the load the will experience while in operation. Unlike last time, the DC-DC converter unit was also submerged for the duration of the test.

Instead of varying the voltage on this trial, I merely set the Lab-Volts to generate a potential of 48 Volts, and then connected the motors.

Here is a summary of the results:

Interesting differences compared to the previous experiment include that the sag voltage at the source is the same for each configuration (40V) and the current changes only marginally from two motors to three. If the resistance in the cable is approximately 5.5 Ohms, as was previously calculated, the power lost to the cable is 30, 38 and 43 Watts respectively.

Thruster Profiles

A question arose about the efficiency of our DC-DC converter, and this also brought to our attention that we had not completed a thrust review on our new motors. This post serves to correct this omission and to characterize the efficiency of our power scheme.

First, the motors that we use are called Johnson Pump Motors, and are traditionally used as replacement motors for bilge pumps. You can see a picture of one on their website:
http://www.johnson-pump.com/JPMarine/products/bilge/sparemotor.html

We would like to understand the power profile of these bilge pump motors so we can make a power budget for the rest of the craft.

First, all of our motors were submerged in a bucket of water, in order to properly simulate the load they will experience while in use.

Next, a variable power supply was connected to a single motor, and the voltage was adjusted in 1 volt increments from 1 to 12 Volts.
The results can be found in this table:

 At full power, one motor can absorb 60 Watts from the power supply. At this point, I am concerned about our DC-DC converter. The power on the input side is only rated to go to 120 Watts, and assuming 100% efficiency, the brick can only provide enough power to run 2 motors.


Now to characterize the efficiency of the DC-DC converter.

 The unregulated power supply was set to 48 Volts DC, this voltage was sent through a 108 ft ethernet cable, and the motor was connected across the 12 V converter.

Immediately the supply voltage sagged to 38 Volts, this is understandable as the resistance of the ethernet cable is proportional to length, and some power is lost in the cable. Cat5 cable has 24 awg solid core wire in it, and its resistance per foot is .0256 (according to powerstream), this corresponds to a resistance of 5.53 Ohms (forward and back). The current in the cable is 2 amps, thus the voltage drop across the cable is approximately 11.05 Volts. This puts the predicted voltage at 36.9, which is close to actual.

Next, the supply voltage was varied and the current and voltage across each motor was recorded.

In the one motor configuration, the follow data was collected.

Unfortunately, I used the readings from the supply to provide the current and voltage readings, so the results are not compensated for the losses in the cable. Additionally, the supply voltage sagged to the recorded number AT the supply, so the real voltage across the DC-DC converter is likely (5.53 Ohms*Asource) less than the reported voltage. If you account for these losses the approximate efficiency is 78% from power into the brick to power into the motors.

When a second motor was added to the test, the following data resulted.

The efficiency is mostly similar, but the voltage across each motor has dropped from 11.8 to 9.7 and the current has dropped to 3.83. Additionally, the power budget per motor drops from 62.5 Watts to 37 Watts.

On the third motor test, the following data was collected.

Again, the voltage across the terminals drops significantly with the increased load, and the power budget per motor drops to 22.2 watts ( note that the third table has a data point 2 volts above the previous data)

It appears that the driving factor for this power limitation is the brick itself. The output has a maximum current of 10 amps, which is insufficient to run 3 motors simultaneously.

Another concern is running the rest of the electronics on the same system. If the voltage drops below 9 volts, many of our cameras will cease to function, meaning that if we drive full blast we will lose the ability to see.

I recommend that we take further data using a separate Ethernet cable to transfer power. This would result in a 4 fold decrease in the cable resistance, and would cut our cable losses to 2.75 Volts which may help alleviate some of these problems.