Two range walks, and one number we won't publish yet

August 16, 2026 · build log

Trail-camera range claims are usually a number with no method attached. We would rather publish the method and let you argue with it, so this month we surveyed a wooded northeast-Ohio property twice with a handheld probe, pushing a payload the size of a real alert onto the mesh at every stop and logging what happened at both ends of every link. All of it is now on a numbers page that we intend to keep adding to.

The short version: a complete payload landed from 544 m (1,785 ft) through timber at peak leaf-out, the link we most needed to work closed with real headroom at ~410 m with the antenna at fence-post height in the rain, and an alert-sized photo crop takes about 20 seconds to arrive whether it is sent from 22 m away or 426 m away.

That is the part that fits in a headline. The rest of the walk is more useful.

The mean curve is not the deployment number

Dense summer canopy cost us a very well-behaved 50 dB per ten-fold increase in distance — five probes spanning 30 to 360 m, every one within 3.2 dB of the fitted line. Extend that line and it says the radio runs out of sensitivity somewhere around 890 m, which would look tremendous on a product page.

We are not going to use it, because the same walks found obstruction shadows 22 dB deep, and the worst one was at 162 m — well inside the working radius, nowhere near the edge. Read the stops in order and the link appeared to die at 162 m; then 135 m further out it came back twelve decibels stronger. Allow for one such hole anywhere on the property and the planning figure drops from 890 m to about 350 m through timber. That is the number we site cameras with.

Which way you point beat how far you went

Grouping every probe in the 395–545 m band by compass bearing produced the largest single effect in the data: 17 dB between the wooded bearing and the bearing across open water, same range, same afternoon, same instrument. At this site's canopy slope, turning a link 15° costs more than doubling its length. Two probes 20 m apart on the same bearing came in 11 dB apart.

You cannot plan around that from a map, which is why MeshCam surveys a deployment with a probe instead of quoting a coverage radius. Walk the spot, press the button, put the post where the measurement says.

The instrument in your hand lies at the edge

Out near the limit, a payload lands at the gateway and the acknowledgment coming back does not survive the return trip, so the handheld reports a failure for a photo that arrived perfectly intact. On the second walk, two of five apparent edge failures had in fact delivered complete payloads. The true reach of one leg was 544 m, not the 480 m the device claimed.

The lesson is which end you believe. The outbound direction, gateway to camera, is the weak half by a measured 13 dB, so every margin on the numbers page is quoted in that direction. Quoting the strong half would have doubled our headroom figures on paper and changed nothing in the woods.

The most expensive thing on the mesh is chatter

The two sites differed in one way we could not have isolated deliberately: one channel was quiet and one had two other nodes checking in every thirty seconds. Same probe, same payload, same radio settings, and a 2.95× airtime penalty on the busy one — a median of 59 seconds against 20. That is larger than any gain available from transmit power, antennas, or radio tuning, and it moved keeping nodes quiet between events to the top of the list.

And the number we are not publishing

The obvious question about a solar camera is what it draws asleep. We have a design budget for it and an indirect bench check that is consistent with that budget, and we are not going to print a sleep-current figure until it comes off a current meter. An ordinary bench multimeter cannot take this measurement: its own shunt browns the board out at the instant the radio wakes to transmit, so the act of measuring destroys what is being measured. The right instrument follows microamps to a full amp in one capture, it is next on the bench list, and when it produces real numbers they will go on the numbers page, method attached.

A page that quotes range to the metre and hand-waves the power budget is telling you which one it actually measured. We would rather have a hole in the page with a date on it.

The numbers page is here, with the tables, the chart, the caveats, and a list of everything these two walks do not prove. Corrections and hard questions are welcome: hello@getmeshcam.com.