Poor nRF52 thread got sidetracked, so let’s start a specific one for this subject of eternal controversy.
The requirement is concurrent charging of one lithium or lifepo4 cell while simultaneously powering the meshtastic device.
About linear chargers:
I think that linear chargers such as the bq24210 or the adafruit board would do good enough in this use case. If you have the right solar panel (MPP around the charging range of a lithium cell, say 3-4V) and and a charger chip that doesnt get confused by the varying output conditions, you don’t need switching topologies and the whole MPPT voodoo. Regular linear lithium chargers tend to get confused by small solar panels and low light conditions. With small panels connected to 18650s current limiting is inherent to the panel itself.The most important thing is actually termination voltage, as discussed earlier.
About Pseudo-MPPT chargers with switching topologies:
These are mostly buck converters that keep the solar panel at a pre-set mpp voltage. I found the CN3791 to work quite well while being cheap and available. One problem is that it always charges until the maximum fixed voltage which should be OK if you use something power-hungry like the T-beam. That way the battery will not be topped up all the time. Plenty of chips with adjustable termination voltage are around.
About energy harvesters.
I own some boards based on the AEM10941 which is a boost mppt solar energy harvester for a single cell. These work very well but are hard to get and somewhat expensive. They also contain 2 LDOs and are therefore overkill. I have a BLE solar sensor node running on this setup reliably for some time now. Works indoors as well. Due to this positive experience with solar energy harvesting I found the SPV1040 appealing. It’s been around for a few years now https://youtu.be/UkFLnVSquu4
The SPV1050 seems to be even more fancy. DFROBOT sell a readymade board for 12 bucks.
I’ve been collecting various higher power solar lights which are readily available, weather resistant, have quite a bit of extra room inside, and often run on common 18650s. For example, TTGO LORA32s are the perfect size to fit inside these lights I found at Harbor Freight, among many other retailers using different branding.
This particular model has a switch on the bottom to deactivate the LED array, but they’re easy to desolder.
I used some flush cutters and a round file to remove a section from the bottom of the case for the stubby antenna to protrude, and soldered the included battery leads (red to red and black to black) into the wires coming off the included (but rather crappy capacity) 18650. This powers the board, but I haven’t done any longer term testing. For a low power relay node that spends most of its life asleep, I imagine battery life will mostly depend on software anyway.
I came to the same conclusion, that MPPT in this application provided no benefit over linear but required more components.
Nearly anything labeled “energy harvester” has a max current too low. The SPV1050 is 70mA max.
The SPV1040 looked good to me too but needs to be combined with a charger IC as well. It also is limited to panels upto 5.5V
The BQ24210 seems to be a good option for these relatively low power repeaters. Although as was discussed, a 4.1V termination voltage is a good idea to prolong the battery life.
I like the idea of the power pass charge IC’s as it means the system draw does not affect the charge termination current sensing.
Wow that’s a great idea! It is a rather low capacity 18650 though so you may need to swap that out. I’d also hazard a guess that the panel may be a little small. Can’t beat a real life test though so keep us posted!
Nice find! I already bought batteries, but for experimenters, this could be a solution for a case, and initial battery at least to start. I like that the light is adhesive attached to a surface. With the right duty cycle I could see this being a dual use platform, light at night, and repeater mode during the day… I wonder if clipping out some of the LEDs would increase the battery life? It has eight of them, removing the central T pattern would leave you with the four corners.
My thoughts exactly. You could probably fit 2 18650s inside giving you a total capacity of up to 7000 mAh, but that’s probably a bit excessive given the size of the panel. According to a previously posted spreadsheet, the built in 1500 mAh battery should probably be enough for 4 days of LoRa Rx and CPU mostly sleeping with no sun.
With these cheap solar lights, it’s important to double check that it has the appropriate charge controller on the board. Some other versions appear to be designed for 3S NiCad batteries lacking overvoltage protection, but are assembled with lithium batteries anyway, and the constantly overcharged lithium cells die quickly. https://www.youtube.com/watch?v=ruEWFdlZnmA
In addition to being a cheap plug and play deployment solution, another reason I was attracted to these lights was their ability to be a relatively inconspicuous concealment device. These are a common sight around lawns, fences, and buildings, and aside from the protruding antenna from this particular example (which others might hide internally), there’s nothing that screams ‘interesting illicit infrastructure’ or ‘possible pipe bomb’ (looking at you @sexycyborg ) about the box itself.
Btw - here’s some of my (super rough - not yet measured wrt solar power outputs) guesses on solar panel sizes needed for NRF52 based router nodes. The NRF52&SX1262 power draws I measured (in bold) but other values are only from datasheets currently.
Any corrections are welcomed. (These numbers are not nearly as mature as the ESP32 measurements, but for router configs I’m personally mostly interested in the NRF52 because it has much lower power requirements)
LEDs are in series string config (?). Best is to remove the string entirely (might consume even more!) or change the string current via a resistor, usually these security lamps have combined IC for charge and night time boost/buck function, so the LED string boost converter will kick in in low light situations.
This particular model has a switch on the bottom to turn off the LEDs, while (I would assume) still keeping the battery topped up with the solar panel. To be honest, I have zero interest in using them for actual illumination, but the fewer modifications, the better.
One important aspect of this is solar reliability, somebody in this group mentioned the use of two solar panels instead of sigle solar + MPPT, I tend to agree. Feeding solar to a linear charger via a diode-or can provide great morning and afternoon solar harvesting like no other setup can. If you have a repeater installed in tree (trees are everywhere!) you will definitely need two panels as most of the solar harvesting occurs with ‘horizontal’ light (sunrise and sunsets), the perpendicular light is efficiency shaded by the tree.
Also, dunno if worth mentioning here, but the ideal size for a panel + main PCB seen as a 1/2 wave or greater for of 868/915Mhz bands for greatest RF efficiency. My point is that the T-beams/other boards can be longer in size (or overall bigger dimensions) for best LORA results.
The final shape of the outdoor enclosure will determine the ease of use positioning the solar panel(s) and I have no magic bullet for the enclosure shape, however a tube-like enclosure for 18650 + elongated PCB is very appealing from waterproofing & antenna ergonomics. Imagine a plastic tube with Ant on top and power cables at the bottom, natural waterproofing, even if bottom is open for moisture venting. I designed waterproof enclosures from die-cast alu, some LORA gateway customers like it, but price is too high (https://resources.mini-box.com/online/ENC-WRAP-2A1E/ENC-WRAP-2A1E-specs.pdf), they are great but a "bullet type tube’ is even friendlier for mounting, IMHO!
The spreadsheet @geeksville lists in the posted above has widely optimistic solar daily 280kwh per square meter panels (for the michigan value). Since the meshtastic device probably will NOT have a sun-tracking system, the odds of the panel in full sun perpendicular are very low.
I have panels in my michigan location and during winter months we average cloud coverage for 9 out of 10 days. The winter output efficiency averages less then 10% of the typical full sun rating. Additionally, during winter the hours of sunlight with direct exposure to the panels is reduced.
My best guess… design should assume worst case input power to the batteries is one hour (1 hr) of direct sunlight every 10 days and trickle amounts for other daylight hours during that 10 day period.
If the solar light panels are rated at 6 watts that doesn’t give us much of a power budget.
+1 Solar is almost always not as good as the calculations suggest. Increasingly it is cheap though. I’d just go for a 10w glass panel and be done with it. As it degrades over time you’ll get a couple more years out of it that way.
I think the TI bq24650 does real MPPT for not crazy prices. Here’s a listing for 2A, 4A, 6A, 10A starting at $8.89 and adjustable voltage. I have a similar module, but haven’t yet tested.
I own a bq24650-based board for a bigger solar setup. It’s quite efficient buck converter-wise (high switching freq. and synchronous topology) but still not real MPPT. One of those “I keep the panel voltage at a pre-set value” things.
I wonder if a true MPPT controller is really justified for charging a meshtastic device. I was searching aliexpress for "Solar Panel USB 5V DC " and I find panels with a voltage regulator manufactured onto the back of the flexible 100W or 50W rated panels with 5V and/or 12V .
I have to ignore the ‘rated’ power and looked ‘100W’ which are 540mm x 280mm (about 21" x 11") and only have a controller output max rating of usb: 5v@2.4a, or 12v@1.2a which calculate to a maximum output of 12 to 14 Watts. Pricing was around $30 US.
The advantage… it is ready to plug into the meshtastic device… just add a usb-a to micro-usb cable
That sounds approximately like my crummy solar test node. This is a not very scientific test I’ve been running outside my house for the last three weeks. It is just a 5V solar panel USB thing I had from hiking, powering a heltec board with a 700mah lipo. I’ve also used the “meshtastic --setpref” command to configure the node to never go to deep-sleep but to leave bluetooth off unless the screen is on. (i.e. the node stays in the light-sleep state and happily participates in the mesh).
When I slapped this node outside the lipo was nearly empty. At least in my current norcal sun it happily stays running forever.
Here’s another option for a solar node. For the last couple days, I’ve been running a Heltec Wireless Stick off of one of these solar panels designed for WiFi cameras. Only time will tell how dependable it is…
The panel has a power bank built in with two 18650s in parallel. I’m powering the Heltec board straight from the panel’s USB with no battery plugged into the Heltec’s JS connector. The panel appears to output the cell voltage from the 18650s, and not a regulated 5v, so I didn’t trust it to provide enough voltage to charge another battery, but so far the Heltec seems stable.
Looking for building some autonomous remote mesh nodes. I’m new to this and have a couple of development boards coming, including the Heltec CubeCell. I wonder if this would be a good self-powered mesh node? I guess it can’t hurt to play…
) about the box itself.
