Finally some good news emerges from the heat-driven meltdown. UMESH 5 down near the high school sits on a residential rooftop, as most do. However, it’s mostly shaded in the morning and get direct sun for only about half the day when the trees have leaves on them. Got up there and it was OK, still going strong.
After that, I went to do some surveying around UMESH 12 on the east side of the University of Illinois campus. Yippee! It’s still on and that’s a good thing since it was our most technically challenging install, requiring UMESH 9 to exercise his solid mountaineering rope skills. He’ll be happy he doesn’t have to repeat that climb so soon after the install. It’s somewhat more exposed than UMESH 5, but similarly has shady mornings and sun in the afternoon.
These findings suggest that it may not be just the top temp recorded, but the daily length of exposure to those temps while the relay and its solar panel in full sun. My guess is the two I’ve found working so far came very close to having the same thing happen that killed the 4 others I’ve found dead. I still need to search for a better solution to spec the support battery but I can make some minor adjustments and probably avoid the meltdown that happened this time.
One of the general lessons/tips to take away is that leaves aren’t always the enemy of the mesh. If you position things to take advantage of their summer shading from the sun, when the leaves fall you’ll gain extra exposure to get you through the wintry season with its reduced solar availability. And you may need that solar shading to keep the mesh on during summer every bit as much as you needed them to go away to get more sun in the winter on your solar panels.
So six are back on and five remain to be checked, not all of which may prove down. Most will, but it could be a lot worst. Basically, the heart of the system near downtown Urbana is working. Farther from the city center, there may be issues. Let us know if you are missing service, as that provides some guidance in what order to focus on which relay to address the needs of those who are using the network.
Also another reminder that we can still use more relay node hosts sited in among and between existing relays, as more relays tend to mitigate failures by routing around them based on what survives. Just contact me through a direct message here or via other contact info listed above in several locations and I’ll arrange to get it placed at no cost.
@MikeL You are encountering the main problem I had with outdoor installations. Heat. I tried a variety of boxes but even the ones with the air pressure release valve couldn’t keep the temps down. Now I did have a cell phone in the box with the gtm which also had a temp gauge in it and things basically shut down over 120F. I’ve since backtracked to where the gtm’s are inside the house on the top most floor but no longer in the attic as temps got too high in there too. I lose about 250-500m depending on the direction but it is what it is.
It is a problem, but a solution is in site. We have reports from 3 more relays today. One was a reboot, one was good as it was, and one is being checked as I write. They’re not all dead, even as bad as it was, so I think my basic design is sound. In fact, as it turns out, the one that was the source of the graph was a survivor!
Now I assess the mesh breakdown, because coverage was very spotty, was more a matter of not having enough redundancy built into the system yet. I’m working on that and people are moving back after a long summer away or arriving here for the first time for a lengthy stay. They need only contact me to end up with a nearby relay node one way or another. We’ll have things back up sooner than expected from the heat and then I think the weather will leave us alone until the deep snows come…we hope
The host of the one we’re waiting on the report will also be adding an enviro sensor to 9, so we’ll soon have two sets of data on that. I think leaving the foam in and using a big enough case helps. That’s with the current battery chemistry.
I’ve been looking into LiFePo4 batteries as an alternative. Sure enough, the high end of the temp goes up from 50 C to between 60 and 80 C. But these would cost more and I know you need experience in using batteries in a solar application before you can really say it works all around.
There are a couple of ideas kicking around to add a little cooling to the case. One is passive by essentially building a box within the box that is designed to vent heat. Another is to solar power a small vent fan that would work without a battery only when it had enough sunlight. That way the top end of the daily heat rise is always tempered, even though you might still get some vent through after dark if careful design were used.
I think now that we have good data that we’re almost there in this climate. There are places hotter already and this place won’t be cooling before the next Ice Age, so in the longer run we’ll need to do more than just get our nose across the line, even if it suffices for right now.
Acurite did that with their Atlas set of PWS. We put one up a few months ago and I can hear the fan running during the heat of the day to try to dissipate the heat to not mess up the temp reading. Might be worth looking into for a solution. I just ran out of copious free time to pursue any further investigation.
Did not realize that kind of cooling was involved with one of those personal weather stations. There’s one a block away from us. But I’m thinking using salvage solar panels and computer cooling fans. A thermostatic switch would be neat, but not absolutely necessary, as there’s only a couple of months out of the year that absolutely need cooling from the looks of the data.
Updating on the third relay from yesterday (9) we were awaiting word on. Turns out it was still good, just was unable to link to other nearby relays after they went down. Also (probably) confirmed another relay (11) was up and operating, although we didn’t physically examine it.
With that, it appears that all of the UMESH relay’s are back online and ready to serve you.
After. brief rain delay, I’m pleased to announce that UMESH 13, the 12th goTenna Mesh relay node serving Urbana, has gone on the air. This puts UMESH at ~33% complete.
UMESH 13 is a relatively easy climb to about 25’ AGL. There is some shading from trees, but a good view of the sky in the afternoon, so pretty ideal for our needs. The solar panel faces west.
Andrew, the node host, is a ham and spends a lot of time in Florida, so has an interest in learning more about goTenna Mesh’s usefulness during hurricanes and other emergency situations.
If all prove productive, that takes the inventory of nodes in reserve down to just 2. We get those two new homes, also, and we’ll be to 50% meshed in Urbana. Can we do that by September? Maybe.
Hey Mike, I responded to your post on Nextdoor. It’s much easier to type here. I live in the Savannah Green subdivision in east Urbana and I would be interested in hosting a node. I already have an antenna mast on my roof (I am a ham radio operator, KC9MGZ). Is hosting the node something you provide? I see that this is a kit form with a solar panel that you have created. If so, you are more than welcome to host a node on my roof, and you can have access to my roof anytime you need to do maintenance.
Hi Andy,
Yes, this is better than NextDoor for communicating, but it’s been productive in bringing in leads for me.
Yes, UMESH will supply the node completely packaged, ready to go and it costs nothing. goTenna supplies the GTM devices that are the heart of the relay node and we supply the case, solar panel, and battery pack. Here’s a pic of the outside and the inside.
Outside
Inside
It’s just a matter of getting it up on your roof safely, then it just sits there, making it’s own power. Here’s a pic of the one on the roof from overhead.
Here at home, we separated the GTM and mount it up on a mast, like this:
If we need it, it’s good to know that the extra height is available on a mast, but we’ll probably just try the standard way first.
Typically, we only rarely need access. The main issue we run into is that if the GTM loses power, it needs to be rebooted manually. Mostly, this can happen in the short days of winter when the powerpack may not get enough sun, perhaps blocked by snow on the panel. There have been a fee cases where the battery pack exceeded its 122 degrees F operating limit and shuts down. We’re working on solutions to these issues, but things are good to go generally
You can use this site’s private message system or the one at NextDoor’s to reach me or via my phone number (voice only, no text). I’ll be out that way on Thursday morning to put up one on S. Smith Road, but we can arrange a time to meet and talk and maybe get one up on your roof at a time convenient to you.
We put up UMESH 15 today. It will be on an updated map soon. UMESH 14 will go in Thursday morning at 10am. I suspect 15 will need to wait until we get 14 on the air before it really becomes useful, but his GTM devices will arrive later on Thursday and should be able to link up then with the rest of the network.
The mesh has been troublesome over the last few days. After an initial set of chats with Andrew W at UMESH 13, things went silent. Turns out that both UMESH 1, my home node, and UMESH 2 were down. They’ve been rebooted. Not sure if this is some kind of heat hangover or what, but I’m hoping this solves most of these minor issue.
I’m curious on that note, any sort of automated reboot system to handle the myriad of issues like that that could be handled with a simple reboot? I can’t think of anything that doesn’t require hardware modification to the unit itself, but a simple automated button pusher connected to the internet could probably handle that. Or to do without internet, a cheap android phone could probably detect if the GTM was offline and trigger the button pusher via BLE. I’m fond of SwitchBot, but I don’t actually know how well that would work in that scenario. Alternatively I’m betting an Arduino setup could work too, but I don’t have any experience in Arduino coding myself and very little in hardware so I can’t say how tricky that would be. SwitchBot works via BLE directly with a device or via the hub to access over a network, as workaround I remote access my Android phone I have dedicated for SwitchBot control so I don’t have to buy a hub.
Thanks for the thoughts. There is a viable system or two that others have come up to deal with button pushing. I like the possibility of it, but hoping that in the long run GoTenna will come up with an integrated system.
The main thing is I don’t like to add complication if it can be avoided. So far, the hardest part is figuring out whst is down. In most cases, it’s easy enough to make the climb, but there are a couple that are real PITA.
Then comes the numbers needed. Even when stuff is inexpensive, once you are talking the multiples we need it turns into significant totals.
And that’s why I have no antennas on my roof. More because we get bad storms regularly and a hurricane typically once a year so I’d rather just keep it high in the attic or a window so I don’t have to worry about it. Still I wish there weren’t so many things that are designed without the ability to turn on automatically. My computer, turns off for any reason, power outage for instance, I need to go push a physical button to turn it back on. Otherwise I just remote access it all the time.
When I was a kid, everything stayed where it was according to the switch. Off stayed off and most like ON stayed on if that’s where the switch was.
Computers of any kind are more finicky.
Here’s some thoughts on locating dead nodes once the mesh starts to thicken up. It’s as much art as science at this point, but here’s what I have in terms of ideas that are easy to convey.
My own relay was down today before I rebooted it, people may have been having troubles, there were a couple of nodes down, UMESH 2 was also down. We may still be feeling some effects of the heat. I’ve been watching where I send messages from and how well and quickly they are Confirmed or not. Sometimes this results in just not getting through.
Other times, a message will get through but take a long time/many more hops than usual. That’s why having more nodes helps avoid issues by routing around where possible. There is some of that built into the network already, but having more nodes improves the chances of you being in a place that works well and reliably because it has good sight of multiple possible receivers to feed its network.
Obviously it would have an increased cost, but what about a cheap android device at each GTM, or at least those that prove problematic, that can send a “ping” of sorts, an automated message to a specific user via the network periodically, like once a day, or two days, depending on how quickly we should detect a malfunction. Personally, if I were setting up the software I’d use Tasker on a rooted device set up to run from device boot up and be fully automated from boot up through sending the messages periodically. A tasker setup would be somewhat device specific, if I had such a device to use I’d create a tasker program to handle it but I don’t know what sort of specific device may be appropriate. An android device that turns on when it begins charging may also be beneficial.
Just brainstorming here. No idea how practical this is or if there’s a much simpler way to detect that a device is offline. Perhaps a power meter, something to check power draw from the battery? Coupled with a simple external LED on the case…
Those sound like great ideas, but mostly go right over my head, sorry to say. I’m the sort of guy who can barely keep a RaspberryPi alive to run my train layout’s auxiliary services.
But there are others here who are equipped to deal with the problem on the level you want to. I encourage you to do some reading on those topics in the archives here. Lots of interest in such things, and some success.
I think many of the problems have simple solutions, it’s a matter of coming up with something simple to solve this problem in light of the fact that goTenna is undoubtedly taking this issue to heart and will solve it next time out. That may be what is restraining third-party solutions, but if they get solved righteously they will be adopted in the interim because they solve a problem that is sometimes given more psychological value than presenting a realistic threat. Just be realistic about initial placement, because yoiu’ll always need to go back, you just want the situation to be reasonable for tempering everyoie’s expectations. Don’t put one where you have to hire a crane to get to it, for instance.
I tend to document things pretty extensively and I now have quite a few nodes. When an issue comes up I tend to go for solutions, but need to get the data to support change, just to be thorough. It may require a little scrambling once in awhile, but this project has been a real down to earth experiment in making things work. That’s how you achieve better operations, by engaging with the process of fixing the few things that do not work while fostering even better performance of the things that do work. Consider the long periods when little is reported as Operations Nominal.
There are enough people using it that notice degradations and report it along with my own long term experience in knowing where things work and where they don’t and why it matters, which I try to pass along in various ways to the many who have helped me. I could not do this without the able assistance of a number of others who mostly prefer to remain anonymous. That’s what intrigues many about goTenna and they are rewarded with a full suite of choices in that regard.
I think we’re on our way to a solution to the heat issue if we need it in the Li-Fe-Po4 battery chemistry. Turns out that they’re used in some solar yard lighting, at Home Depot, for instance. I just haven’t had the time to check on that angle yet, but will be buying stuff soon, so it will happen then, it the other issues like charging and packaging can be worked out.
Checked on UMESH 15 this morning and am happy to report it already connects with the mesh without aid from UMESH 14. It will take a few days to get around to an updated coverage map.
