ACO Experimenters Guide
If you have any questions that are not covered in the Table of Contents below, please feel free to email or call the appropriate person on the Contact page.
Customer Interface Information
- Where is the ALOHA Cabled Observatory deployed?
- Should I get into a group to propose a sensor system to connect ?
- Who will be responsible for system operation?
I Have Questions About Power.
- How much power can I get ?
- My sensor has a large turn-on electrical transient. Is that OK
- What happens if I draw more current than allotted?
- What about High-Voltage Connections?
I Have Data Communication Questions.
- How much data can I send over the system to my desk?
- I want to send large amounts of data in bursts. Is that OK?
- What happens if I use more bandwidth than allotted?
- Can I send commands to my sensor to change its configuration?
- Can I send commands to change software?
- I'd rather not connect to the system with a hard wire. Can I connect with an acoustic data link?
- How secure is my network connection and data link?
I have questions about system integration.
- What is the data interface ?
- What are experiment configuration options?
- How far away from the observatory can my experiment be placed?
- How do I prepare my instrument for installation?
I Have Timing Questions
The ACO (ALOHA Cabled Observatory) is located at 22° 44.324'N, 158° 0.372'W, approximately 1.2 km SSW of Station ALOHA (22° 45'N, 158° 00'W). See aloha.manoa.hawaii.edu for information about other activities at Station ALOHA.
Combining compatible experiments into a sensor network that uses only one observatory connector is likely to be the most cost effective way to get your data from the ocean floor.
System operation responsibilities are handled by the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST). They have the ability to manage the day-to-day operation of the power and communication links of the cable system at the Makaha Cable Station.
Power from the cable is substantial but limited (total of about 800 W available to users). This power has to be shared by the entire community, and some experiments have a legitimate need for more power by the very nature of the experiment. We are not therefore allocating a fixed amount of power per experiment, but are providing "reasonable guidelines". In the interests of the larger community, we are asking each experimenter to make a conscientious effort to conserve power.
A "reasonable limit" below which there is probably no major concern is perhaps 25 Watts. With this kind of conservation, we can probably supply up to 100 Watts or even more on a case-by-case basis to power-hungry systems. At present we have very little feedback on power requirements, so let us know what you have in mind.
Each user node is powered through a remotely programmable 8-step circuit breaker. The current limit for each user will be set to allow a reasonable start-up transient. This is necessary in order to safely manage the power system for the other users.
If your equipment intrinsically requires a large surge current, we may have to temporarily increase your breaker setting for turn-on. The limit will then be set back to a safe limit for normal operation. If the transient exceeds the limitations of the circuit breaker, then the user will be required to buffer current transients with a battery or other method.
A programmable circuit breaker will trip causing the 48 volts supplied to the user interface to be shut off.
The Observatory also provides two higher voltage and power (400 Volts at 200 Watts) user ports for experiments or sub-multiplexer nodes that are located a considerable distance from the Observatory site. This higher voltage reduces cable power loss by a factor of 50.
Data Communication Questions
The present ACO infrastructure supports a single 100-base T Ethernet link to the shore side Makaha Cable Station. From the Makaha cable station to the University of Hawaii, the bandwidth is 3Mbit/sec link. Individual users will be able to access only a fraction of the 2 X T1 link. Future expansion could increase the available bandwidth. Care must be taken with high data rates, in that high-speed Internet connection costs can be prohibitively expensive
An intense burst of data on one channel may cause latency problems for another time-critical experimenter. In order to provide the best possible service to all of our experimenters, we may limit the bandwidth available to an individual port. Your data will still get through, but it may take a bit longer. If you have an experiment that requires burst data, please contact the operators.
individual users is controlled by using a bandwidth manager at the
Yes, in the same way that you do it while testing in your shop.
Software changes are dependent on the users programs.
Experimenters desiring acoustic modem connections to the observatory will be supported, but the acoustic modem will be part of the experiment- not part of the current ACO infrastructure.
It should be understood that acoustic modems interfere with each other, so only one such link can be operational at a time. The other alternative is to have a shared acoustic link. In this scheme, the ACO link would poll the individual experiments.
The security of your network connection to SOEST is controlled by you and your ISP (Internet Service Provider). There are several firewalls between the external interface to SOEST and the ACO seafloor network. The seafloor subnetwork is isolated from the Internet. Maintaining security of your login information is vital.
System Integration Questions
The Observatory is basically a 100 Mbs LAN. Your experiment can plug in either to a 10Base-T Ethernet port or to a serial port. In principle, if the connection works in your lab on your local network, then it should work on the ocean floor. Ethernet protocol is preferred, but low-rate systems on long cables may be best accomplished using RS-232 or RS-422 serial. Experiments on long cables (>100m) to the observatory that require high data rates must be heavily tested. Ethernet has cable-length issues, and the necessity to pass the signals through connectors may degrade signal.
(The ACO wiring diagram (image 1) shows the recommended jumper assembly for connecting a user to an observatory node. The above are uncontrolled copies, for up-to-date information contact Bruce Howe or Ethan Roth.)
There are several options to connect to the ACO:
1. Simple, reliable sensors that can be attached to the observatory structure on deck. These sensors can be installed when the observatory is brought to the surface for repair or modification. The advantage of this method is that an expensive wet-mate connector is not required. The disadvantages are that your sensor will be brought to the surface whenever the observatory is recovered, and that it cannot be deployed away from the observatory structure.
2. Sensors installed permanently near the observatory. These sensors will need an emplacement system compatible with the ROV being used for installation. These sensors will require an ODI Nautilus wet-mateable electrical connector to connect the sensor to the ACO on the ocean floor using an ROV. The costs of the ODI wet-mate connector and cables will be borne by the experimenter.
3) Sensors installed at distances greater than 150 m from the observatory. Free falling of these sensors is possible at these distances, but may have to be moved around the seafloor by an ROV. They will require an ODI Nautilus wet-mateable electrical connector to connect the sensor to the ACO on the ocean floor using an ROV. The costs of the ODI wet-mate connector and cables will be borne by the experimenter
4) Sensor networks. Groups of experimenters are encouraged to pool their interests and proposals for sensor networks which can use a separate module to share the power and bandwidth supplied by the observatory.
In all cases, sensors
must be tested at the SOEST Observatory Test Bed prior to installation.
This insures that the sensor is compatible with the observatory and other
There are two factors which limit the distance to the experiment. One is the voltage drop in the cable. A low power experiment on a heavy conductor can be placed farther away than a high-power system or a system with a light conductor. We normally supply power at 48 VDC, but two 400 VDC user ports will be available initially. The second factor is the ability to communicate via Ethernet over the underwater cable.
The underwater-mateable connectors cause impedance discontinuities and reflections on the cable. That is why we say that your experiment must be tested on the submersible umbilical before shipping it for deployment.
RS-422 serial can communicate over longer distances, depending on Baud rate. Again, you should test your sensor with the length of cable you hope to use. With the lower frequency of the RS422 signal, the connectors probably have a negligible effect.
1. Check with the ACO operators to determine likely power and data allocations, and other possible options and restrictions for your experiment. Since these factors could be cost drivers, it is highly advisable to determine these parameters during the proposal preparation stage.
2. Check with the ACO operators to determine location, connector, pin assignments, installation cruise opportunities.
3. Test your equipment to your satisfaction using your own Internet connections.
4. Notify ACO operators that you are ready for a test at the ACO test bed. If your experiment is relatively simple, it should be possible to have observatory personnel install it at the test bed and assign a URL for you to check your data and commands from your office - just as though it was installed on the ocean floor.
5. Coordinate with the ACO operators and users group concerning installation procedures and options.
PTP timing signals synced to a GPS receiver in the ACO network is available to users along with an IRIG-B 1 pulse/sec line. Ethernet experiments can use NTP (Network Time Protocol) for a time base.
While this is still an open question, if your sensor system can read the IRIG time code supplied by the Observatory, and convert it to a time stamp in your data, then time should be accurate to better than 1ms. If you do not use this feature, you should still be able to correct your data timing to better than 1 s.