Living Observatory Sensor Network

The Central Stream Cam at Tidmarsh, which normally streams 24/7 live video that you can view by clicking here, stopped responding on November 20th. The last video it recorded didn't provide any clues, as it just shows ducks swimming by. Our Southern Marsh cam also failed this fall, after working flawlessly for several years. Yesterday, I made the trip down to Tidmarsh to investigate, with Glorianna in the field with me to help and document.

When we arrived at the camera, the problem was immediately apparent. The cable to the camera had been completely severed, as had the cable connecting the water temperature sensors to the sensor node. The connector had also been chewed off of the box that used to be connected to a pH probe. The cuts were remarkably sharp for rodent damage, but too crude to be made with wire cutters. Three cables were cut on the camera pole, but the nearby base station was untouched.

The camera (at the top of the pole) still seemed to be in good condition, so we set about replacing the cable. This cable runs about two meters to the base station right behind it, so it was easiest to replace the entire cable rather than trying to effect a splice.

The base station (behind the camera) powers the equipment from two solar panels and a battery, and provides the communication uplink (the large white dish) back to our head end in the barn. The two small black antennas above the dish link to the low power wireless sensor nodes, including the one below the camera.

The cable repair brought the camera back online, which you can once again see here.

Southern Marsh Cam

We also went to investigate the failure of the Southern Marsh Cam, which is still offline. This camera is at the end of a 100+ meter cable that runs way out into the marsh. This cable has been chewed by critters in the past; in 2018 and 2019 we made several splices to repair damage. Remarkably, it kept working from then up until August of this year.

Troubleshooting a very long cable that runs through the marsh can be challenging. It is impractical to visually inspect it for damage, as the majority of the cable runs underwater and through the reeds. (We try to keep as much of the cable as possible underwater and hidden, where it is safest from rodents and vandals). And even a small nick can be enough to damage one of the eight wires inside, so the damage can be hard to spot unless you know where it is.

To locate the fault, we use a technique called Time Domain Reflectometry. The TDR meter sends a pulse of electricity down the cable, which travels at the speed of light. (Actually, the speed of an electromagnetic field inside of a cable is a bit slower than the speed of light and depends on the geometry and materials of the cable; in this case it's about two-thirds the speed of light, which is called the velocity factor of the cable). When the pulse meets a discontinuity in the cable, some of the pulse is reflected back. The TDR meter plots the reflected signal, and by measuring how long it took to arrive, can determine how far down the cable the reflection came from. The polarity of the reflected pulse also tells us something about the nature of the damage. A negative reflection indicates a short circuit (i.e. two wires connected together) while a positive reflection indicates an open circuit (a cut/broken wire).

Since the camera was still connected at the other end of the cable, it affects the reflection as well (assuming the signal makes it all the way to the camera). The Ethernet cable has four separate twisted pairs of wire, each of which is a transmission line that can be tested separately with the TDR meter. This camera sends both data and power over the ethernet cable. Since it is a 100 Mbit/s ethernet device, it only uses the orange and green pairs for data. The brown and blue pairs are used for power. (There are other power-over-ethernet [PoE] schemes that can share pairs for both power and data, but these cameras use a simple "passive" PoE scheme).

The blue and brown pairs both look the same on the TDR meter:

This shows a short circuit (negative reflected pulse) at a little over 600 feet. In this particular passive PoE scheme, both wires in the brown pair are used for the negative supply and the two blue pair wires are used for the positive. They are doubled up so the cable can carry more current. Thus, the two blue pair wires are shorted to each other inside the camera, as are the two brown pair wires. So far, we're seeing exactly what we'd expect from a camera connected approximately where we'd expect to see it.

The orange pair looks different:

It doesn't show much of a reflection at all. Since the orange pair is used for data, not for power, it is connected to an impedance-matched transformer inside of the camera. Reflections can interfere with communication, so high speed transmission lines used for data are impedance matched to minimize the reflection. So again, this is exactly what we'd expect to see for a good connection. Note that there is a little bit of a squiggle at the 613ft mark: this is the slight discontinuity in impedance where the signal goes through the connectors at the end of the cable and into the camera. (As a side note, this is also why there's a little bump at the beginning of all of these plots: me crudely pushing the cut ends of an ethernet cable into the SMA jack on the meter because I forgot to bring the right adapter with me is not a very good impedance match).

Since the green pair is also used for data and not power, we'd expect to see the same thing as we do for the orange pair. But the TDR meter tells a different story:

Here we have a positive reflected pulse at 492 feet. This is about 100 feet closer than everything we've seen so far, and the positive pulse indicates an open circuit, which suggests that at least one wire in the green pairs is cut.

So without even getting my feet wet, I've been able to identify the likely fault. Somewhere around 100 feet away from the camera, a critter has likely nibbled on the cable just enough to break through the green pair. On this trip, I didn't have the right supplies on hand to actually wade out into the marsh and splice the cable. But I now know what to bring when I return shortly and where to start looking for the fault. Conveniently, the ethernet cable we use has the length printed every three feet, so anywhere we find the cable we can read the length marking and compare it to the printed length at either end of the cable to know exactly where we are. Weather permitting, I will be back soon (with help) to make the repair.

Bonus Photos

While I was making repairs, I also shot a few photos for fun.

The pond near the Central Stream Cam.

Ripples in the sand and the light filtering through the surface of the stream.

Lichen growing on a decomposing stump.

Logs.

Moss.

Lichen growing on a log.

Dead wood in the stream.

These were shot on Kodak Tmax 100 (exposed at ISO 200) with a Nikon FG and Micro-Nikkor 55mm f/2.8. Processed in Ilfotec DD-X. My developer is probably nearing the end of its shelf life and this isn't a film stock/developer/push processing combination I've tried before and the negatives came out a little thin, but I still got a few decent images.