Satellite
to Data Logger Revisited
The
week of March 12th, 2001, consultant Dan Withers and I revisited
my attempts (Diary #24) to get reliable connection results
involving the employment of a Tachyon Internet Satellite
base station, Campbell Data loggers and their NL1000 Interface,
Wireless data radios, and the PC208W 3.2 'Ethernet' capable
software.
Dan
at work in my Lab
We
also experimented with a lower cost (than the Campbell NL100)
Ethernet to Serial adapter based on Lantronix Cobox-Micro
embedded device adapter.
I regard the efforts as a failure. And have concluded that
it will take changes in either the Campbell Data logger
software - at the loggers or at the computer - before a
Satellite-Wireless can be successful on a sustained and
reliable basis.
Repeat
and Progressive Tests
We
started to re-explore this avenue - which could have significant
value for research projects in areas so far from terrestrial
links to the Internet that only satellite communications
is feasible - by building up one added feature/device at
a time.
In
all the trials below we used 12 volts of power to the Data
Loggers and NL100 or other devices, from a Marine Battery,
as it would be in the field.
Trial
#1
- We used Dan Wither's NL100 and the CRX10 which he has
been experimenting with in the Seattle area. We connected
it up by a 10base T cable to a Hub, off of which is Windows
95 machine running PC208W 3.2 software. The connection worked
as it should, and there were no drops in the 'connected'
state over a full 10-15 minutes. This is as long as a Data
Logger needs to be connected for routine downloads of collected
data.
Trial
#2 -
We then reconfigured the NL100 in my Lab and connected it
first to the CR10X, and then to the CR23X that stays in
the lab. Again the short distance, Ethernet wire connection
between the software and the data loggers, worked as advertised.
Trial
#3 -
For this trial, instead of an in-lab wire-to-wire connection,
we accessed the IP Number in the NL100 - which is the way
one reaches the attached Data Logger - via the Tachyon Satellite
connection. Using another Windows 95 computer on the premises,
we went out over the 192.160.122.0 network to Red5 Net service
connected, in San Diego, via Tachyon's service on SatMex5,
down to the 1 meter dish on my Lab premises, then to the
Indoor Tachyon unit, out its Ethernet port to the same hub
and same ethernet wire we used in the previous trials. In
Diary #24 we had used this same arrangement with more success
that we experienced this time. That is, while it readily
Connected to the data logger, it only ran for about 2 minutes
before dropping the connection, then reconnecting 5-15 seconds
later. The latency over the KU band satellite link was no
worse than we had observed in previous tests - from 800-1200ms
- when we had been able to hold the connection up for upwards
of 5 to 10 minutes. We varied the 'delays' in the PC208W
software from 200 to 2000ms with, however, similar results.
Repeated
efforts to sustain a connection, done at different times,
to account for Internetwork congestion, generated the same
results. A research would have been able to 'work through'
such drops, though they would be annoying.
Lantronix CoBox Micro Adapter
http://www.lantronix.com/products/embedded/coboxmicro/
Lantronix
makes a number of OEM devices that can be configured for
numerous embedded process control tasks. The CoBox Micro
device is one of the simplest and lowest cost ways to convert
an Ethernet (10mbps) connection to a serial, such as RS232,
connection. Costing in the range of $175 in individual quantities,
it terminates in a TTL connection. Thus it requires a second
board to convert the TTL into 9 Pin RS232. Dan Withers fabricated
that board so we could experiment and see if it could usefully
substitute - at perhaps a $200 'plug and play' cost, for
the $500 NL100, which has many other features than just
the Ethernet-to-Serial conversion.
The
CoBox with added RS232 Board
While
the NL100 requires a 12volt DC feed to it, which can come
off the same 12v battery that powers the data logger in
the field, the CoBox Micro requires 5 volts. Withers cleverly
made a connection to pin 1 and 2 on the Campbell Data Logger
CIO Port (which is not standard RS232), that supplied the
requisite power.
We
found throughout these trials that the CoBox could substitute
most of the time for the NL100. This is significant because
the NL100 actually was not made to be deployed - at a $500
add on cost to a $1,000 CR10X cost - at every data logger
site in the field. It appears to have been designed to work
inside a Research Lab to permit Computer Workstations running
PC208W 3.2 software over a local area LAN network, to reach
a serial 'gateway' to the data loggers - whether they are
reached hard serial wire or wireless.
If
the CoBox Micro can be packaged and deployed at the data
logger end of either ethernet wires, or Ethernet Wireless,
at $200 each, a fair savings can be realized.
The
only lingering question is how low a temperature the CoBox
Micro can reliably handle. Its spec sheets seem to say it
can only go to 0. Which would not work in the Arctic.
More Trials
Trial
#4 -
In this trial we substituted the Lantronix CoBox Micro device,
with the 9 Pin RS232 Board attached, for the NL100.
CoBox
integrated into network
This
worked properly and with the same reliability as the NL100,
both in the wired, and via Satellite, configuration. No
better or worse to this point.
Trial
#5
- We then added Wireless Data Radios between the Satellite
Ground station, and the NL100. In this case we used an Aironet
BR500 Access Point radio - which would be a proper choice
for a point to multi-point radio at a field Satellite Internet
Ground Station - and an single mac-address Aironet external
radio at the data logger location. When accessing all of
this from a remote location Internet attached computer:
Internet
<--> Satellite <--> Aironets <--> NL100 <-->CR23X
it
worked, but more intermittently than when there were no
radios in the loop.
Trial #6 - When we
substituted the CoBox Micro for the NL100, in the above
model, including the wireless link, it failed to connect
at all. Several attempts were unsuccessful. The only differences
were in that between the NL100 Interface and the CoBox Micro
device. It is not clear what the problem was, but it appears
that there is simply less error correction capability in
the $200 CoBox than the $500 NL100, which was designed specifically
to work with Campbell Data Loggers.
Trial
#7
- We then dropped the Satellite, with its 800-1200ms latency
trait, and connected a microcomputer running PC208W 3.2
software, to the Aironet Access Point and accessed the CR23X
data logger via wireless only. We used both the NL100 and
the CoBox Micro. Both worked equally well, and just as reliably
as a wired ethernet connection.
Results
These
trials showed:
a. Reliable ethernet
speed Wireless links between Data Loggers equipped with
either NL100 or CoBox Micro Interface devices, and a distant
Computer running PC208W 3.2 software can reliably operate.
With Aironet radios and proper antennas that could be up
to a range of 10 miles or more.
b.
Reliable links using a high-latency Satellite Internet link
between the remote researcher and the data loggers cannot
be sustained with off-the-shelf hardware and software.
c.
A lower cost CoBox Micro ethernet<->serial device can be
used in many, but not all, circumstances, as a substitute
for an NL100.
