Forum Replies Created
-
Posts
-
Autumn in Bonham: Rest Stop #3 Recap
Our team at Rest Stop #3 had a busy and rewarding Saturday, with a total of 106 riders coming through.
RS3 was set up at the Bois d’Arc Lake Operation Center. I arrived around 7:30, leaving early since I wasn’t familiar with the location, and that turned out to be a good call. I did have trouble finding the spot, and a few riders had the same problem during the event. Most of the confusion came down to signage: a standard school-project poster board was the only marker for the turnaround. That’s something worth passing along to the organizers for next year.
The event staff went all out with their Hawaii-themed decorations, which gave the stop a fun, welcoming feel.
We had a few mechanical issues along the way, and our SAG operator transported two riders back to the starting location. The famous 513 rider made it to RS3 but couldn’t continue because of a punctured rear tire, and mechanical trouble kept another rider from finishing his 40-mile ride. Our last rider of the day told us she had taken a wrong turn and ended up on the wrong route. We helped her get her bearings, and she continued on.
She came in just before the rain started, and we were able to get the site packed up in time. Shortly after, Net Control called asking us to bring RS3’s leftover supplies to RS12 in Windom, at the intersection of Highway 56 and FM 1743. After a quick directions search, the crew loaded everything into the truck, and we arrived at RS12 just as riders were showing up. The timing couldn’t have been better.
All in all, it was a great day and a blast. Thanks to everyone who helped make RS3 and the event a success!
73,
Chris, KI5HSQAnd the second one is painted and put together. This one went together in about half the time.
The full print time is about a day and a half with my Ender 3 Nero. Painting takes a full day to dry. Then putting it together is only about 30 minutes of work.
Our grant actually goes against the NextNav policy proposal. We just need to create awareness of it to limit their intrusion into the band. The +4 watts would raise the noise floor and drown out all non—commercial applications.
HRCC has a similar build for an aluminum-body F-150 or a glass-topped roof. It’s the same type of print just uses suction cups instead of the magnetic feet. In his video, he does uses the slim, low-profile 3D print without the fan or switch.
Two Weeks In — What I’ve Learned (Follow-up to my Tower/Truck Buildout post)
Figured I’d do a proper lessons-learned writeup since a few of these caught me off guard and might save someone else some time. Long post, organized by category.
Preset choice actually matters, a lot
Confirmed via back-to-back field testing: Long Range – Fast reached the Loop consistently from 9 miles out; Long Range – Turbo fell short by almost 2 miles on the identical route. Same spreading factor (SF11) on both — the only difference is bandwidth (250kHz vs 500kHz), and that ~3dB sensitivity gap translates directly into real-world range. If you’re optimizing for reach over throughput, this is not a minor detail.
The catch: the app itself flags Long Fast as non-compliant with US bandwidth rules and pushes Long Turbo as the compliant option. Real tradeoff between range and being fully by-the-book — still working through where I land on that, and it’s the main open question in my last post.
The coverage estimator is only as good as what you feed it
Learned this the hard way — my first coverage estimate was wildly conservative because I had the wrong antenna height (6.5 ft instead of my actual 33 ft) and wrong gain (2 dBi instead of ~5 dBi net) loaded into the tool. Once corrected, predicted range roughly doubled. If you’re using one of these estimator tools, double- and triple-check every input against your actual hardware before trusting the output — garbage in, garbage out.
Also: turn on high-resolution terrain if it’s an option. The jagged “starburst” fingers you see on a good terrain model aren’t a rendering glitch — that’s real line-of-sight clipping from hills and tree lines. A smoothed/low-res terrain model will lie to you by averaging that away.
RSSI showing 0 doesn’t mean what you’d think
Spent a good while chasing this before figuring it out: nodes 17 feet apart were showing RSSI 0dB in the node list, right alongside genuinely distant nodes showing real negative RSSI values. Turns out RSSI is an optional field in the packet structure, and when it’s absent, at least the iOS app has a known bug where it displays “0” instead of showing it as unknown/missing. SNR is the field to actually trust — it was populated correctly and consistently across every node regardless of the RSSI display issue. Don’t let a 0 RSSI reading convince you a nearby link is broken or fake.
A visible node isn’t proof of a working link
Mark’s ThinkNode M6 at ~13.4km on 33rd SE showed up in my tower’s node list, but wouldn’t complete a return Traceroute. First instinct was “maybe MQTT is making this node visible without a real RF path” — worth checking, but confirmed neither is running MQTT, so more likely explanation is an asymmetric or marginal real RF link (one direction hearing better than the other), or a relay path that doesn’t complete cleanly on the way back. Traceroute, not just node-list visibility, is the only way to confirm an actual working link — still running this down.
Device role: Client Base beats Router/Repeater for fixed infrastructure
Current firmware guidance actively steers away from Router/Repeater roles unless you’re maintaining serious community backbone infrastructure — apparently they’re more prone to causing collision/congestion issues than helping. For a home base station like my tower, Client Base (paired with marking your other nodes as Favorites) is the supported pattern for “this fixed node should prioritize relaying my other gear.” Wish I’d known this before defaulting to Router out of habit.
Antenna shopping: watch for missing gain specs, and match the vehicle to the antenna
Two separate lessons here:
If a listing doesn’t state a gain figure in dBi, that’s a red flag, not an oversight. Compared a rigid fiberglass N-connector antenna (clearly stated 5dBi) against a flexible rubber “improved signal” antenna with zero gain spec — pulling the FCC filing on a similar unit showed the flexible style typically comes in under 1dBi despite the marketing copy. Rigid fiberglass construction is generally the tell for genuine gain; flexible rubber duck style usually isn’t.
Higher gain isn’t automatically better on a moving vehicle. Higher-gain omnis achieve that gain by narrowing the vertical beamwidth — great on a fixed, always-vertical tower, bad on a truck that tilts on rolling terrain. Landed on the 5-6dBi range as the sweet spot for the truck rather than chasing the highest number available.
Enclosure material: PLA+ is not a Texas-summer material
This one applies whether the part sits in direct sun or inside a sealed enclosure — sealed doesn’t mean cooler, since there’s no airflow to shed the heat a black case absorbs. PLA/PLA+’s heat deflection point (~56°C average) is well within range of what a black, sun-exposed part will actually hit. Switched my truck node enclosure to PETG (paint + UV-stable clearcoat over PLA won’t fix the underlying softening risk) and I’m printing all outdoor battery holders in PETG too, since sustained heat plus continuous spring-tension load is a recipe for slow creep/deformation, not just a warping risk on a hot day.
Also: don’t rely on printed plastic for the actual battery spring contacts, regardless of filament — use real metal contacts. Plastic isn’t a good long-term spring material under thermal cycling.
Little things worth flagging for anyone replicating this
RAK19007’s solar port uses JST ZH 1.5mm, not the far more common JST PH 2.0mm — cost me a wasted order before I caught it. Check your base board’s actual connector pitch before ordering pigtails.
Wire parallel 18650 packs in parallel, not series, if your board expects single-cell 3.7-4.2V input — series would overvolt the charge controller.
A cooling fan tied to a thermal switch on a solar-charged setup is mostly self-solving power-wise — it only draws current when the enclosure’s hot, which is also when the panel’s producing the most.Last but not least
Building is way better than buying, I could have just bought an off the shelf node that was pre-flashed and installed them in the truck and on the tower but where is the fun in that. Between the 3D printing of the enclosures, building out the wisblocks and heltecs, flashing the units, and then working through all of the setting, I’ve learned a lot. Mostly what not to do but still useful information along the way. An the whole FCC compliance thing has me running in circles, on one hand everyone is running Long Fast, and the new regulations that came out this month say we should all switch to Long Turbo. It directly limits rural areas like use while little to no impact is felt in major cities with lots of nodes. Personally I don’t think everyone is going to switch over, over-night, but it is frustrating.
The Good the Bad and the Ugly
Good:
What I love about the whole 900 band is that it stretches your concepts and challenges your RF skills to the max — everything you’ve learned, or thought you’d learned, in tech/general now gets put to the test in real-world conditions. You’re dealing with 1W of power, with something that can radiate further than WiFi or Bluetooth. Reach can’t be overcome with more power — antenna gain concepts become more practical, and a hill or a valley becomes a real challenge.
The Bad
Hills and valleys (never would have thought we had so many in Lamar County) — even though 900MHz’s longer wavelength gives it better range and obstacle penetration than WiFi or Bluetooth, it still travels essentially line-of-sight at this frequency, so every change in elevation can affect the signal. Roll and pitch of a truck, combined with antenna gain, plays a big factor in receiving a signal.
The Ugly
Meshtastic relies on a lot of nodes, but is hindered by rising congestion and reliability issues as hop count climbs. The trick is getting antennas that are above a certain elevation with the proper signal gain for coverage, all while trying to minimize congestion on the mesh. It’s not just about drawing circles on a map — it’s about direct line of sight to a node and the proper antenna gain to send and receive signals. These past two weeks have been challenging to say the least, but have been the most fun I’ve had in a long time.
Still open
Confirming the ThinkNode M6 marginal-link mysteries
Deciding on Long Fast vs Long Turbo for the whole fleet given the compliance/range tradeoff
FM 1497 repeater placement — still scouting for the right elevated spot
Happy to go deeper on any of these if it’s useful to someone else’s build. Learned most of this the slow way, so hopefully this shortcuts it for the next person.73,
KI5HSQ
Tower + Truck Buildout Update — Paris, TX (Lamar County)
Wanted to share where things stand on my home/truck mesh setup and get some feedback from folks who’ve done similar builds out this way.
The setup so far:
Home tower (KI5HSQ_Tower):
- RAK4631 (WisBlock) on a mast at 33 ft AGL
- 5.8 dBi N-connector omni antenna, ~20 ft of LMR400 (net gain after cable loss: roughly 5 dBi at the antenna)
- 30 dBm (1W) TX power, RX Boosted Gain enabled
- Region: US, running Long Range – Fast (250 kHz Long Fast flagged non-compliant by the app for US bandwidth rules, Long Range – Turbo is now suggested)
- Device Role: Client Base, with the truck/personal/MQTT nodes set as Favorites
House bridge (KI5HSQ_MQTT):
- Heltec V3 on WiFi, bridging to the public broker (mqtt.meshtastic.org)
Map Report / Ok to MQTT enabled on this node only — everything else stays off the public map intentionally
Truck node (mobile):
- WisBlock RAK4631, currently testing antenna options (5 dBi baseline, considering 6-8 dBi N-type for a real A/B comparison)
- New power build in progress: 2x 18650 3500mAh (wired parallel, 7000mAh @ 3.7V), 5V/6W solar panel, thermal-switch-controlled cooling fan (independent circuit, kicks on at 40°C)
Personal/handheld node (KI5HSQ):
- Heltec V3, carried on me once I’m out of the truck — pairs to my phone for messaging/traceroute on foot
- Same channel/PSK and Long Fast preset as the rest of the fleet, set as a Favorite on the tower’s Client Base config
Coverage estimate vs. reality
Ran the terrain-based coverage estimator with corrected antenna height/gain/sensitivity for Long Turbo and Long Fast, and the predicted footprint showed the usual jagged “starburst” pattern you’d expect from terrain clipping — strong core out to several miles, long fingers down open corridors, short stubs where tree lines/river bottoms get in the way.
Been field-validating with Traceroute rather than trusting the map blind — logging SNR and hop path at intervals while driving the predicted fingers, using Route Recorder to track the GPS path alongside it.
Results so far: Ran the same test route on both presets for a direct comparison. Long Range – Fast consistently reached the Loop from 9 miles out — solid, repeatable link at that distance. Long Range – Turbo fell short by almost 2 miles on the same route, which tracks with the ~3 dB sensitivity difference between the two (Turbo’s wider 500 kHz bandwidth trades range for speed). Given the compliance flag on Fast, still weighing that range advantage against the US bandwidth rule the app warns about — open to how others in the area are handling that tradeoff.
Also picked up Mj33’s node on 33rd SE on the way — showing up on my tower’s node list, but Traceroute isn’t returning a confirmed path back. Still working out whether that’s a one-way visibility thing (MQTT or otherwise) or a real-but-marginal RF link — will update once I run down what’s going on there.
Next steps
Updating the Mobile enclosure to a Simi permanent truck mount. I’m printing it in PETG and plan to paint it the color of the truck.
Concept picture of the finished product.
I’m also printing some 18650 holders for backup batteries for the tower unit and mobile nodes.
Question for the group
Anyone running nodes along FM 1497 or in Lamar County already? Curious what preset you’re on and whether there’s a coordinated local channel I should be matching, rather than assuming Long Fast or Long Turbo in isolation. Also open to repeater-placement advice for anyone who knows that stretch of road.
73,
KI5HSQ
