I’ve been working on a PCB for remote Meshtastic Routers based on the TI BQ25185 (solar charge controller) and Heltec HT-CT63 lately. The PCB isn’t the most beautiful thing around (it follows the reference design where required) but it seems to work for now.
(Picture is a bit of an older version. Latest revision got some clean ups but wasn’t yet tested and manufactured. But only smaller, non-functional changes happened in the PCB.)
Features:
Supports solar input up to 18V (24V survival)
Up to 1A charge current, 3A load current
Single-cell LiPos (3.7v) or LiFePo4 (3.6V) supported
Pseudo-MPPT (VINDPM) support
Power Path (being able to power the device directly from solar as soon the battery is full, mixing solar and battery power in the “twilight” zone, etc.)
Low power LDO (HEERMICR HE9073A33MR) to get from peak ~4.5V to 3.3V
Low bills of material. I hate soldering small stuff because of my shaky hands.
Seems like I can get 4-5 days out of the device without recharging and it nicely switches to direct pass-through (jumps >4.2V) once the battery is full.
Super. Ace to see a ground up design spec with super voltage data. How did you get the voltage data logged? Oh… I see you said in github… Wouldn’t it be amazing to have all this in one PCB… hang on… it IS all on one !!! This is really ace. Precisely what we need for mass production at low cost. Thanks.
Yes… money was bit of the motivation. I love the RAK4631 devices but spending 30-40€ per block was too expensive for me (heh, with all the test runs and PCB revisions my hardware should be factor 5 more expensive by now).
For the stats I’m using a hacked python script which parses the “–telemetry --dest …” output of the ‘meshtastic’ Python application. It’s not beautiful and not reliable as it often needs multiple attempts to fetch the data.
The better way would be to increase the telemetry timer of the DUT and then getting the data from MQTT.
One of my nodes shows abnormal behaviour… I have to track this down. Maybe a bug in the firmware:
I have been using the HT-CT62 myself for solar nodes. In router mode it seems to sleep well but not in some of the others.
Just switching from using a separate charger and LDO to this unit https://www.aliexpress.com/item/1005005175494273.html
Currently manually wiring to a board which is messy with the format ot the HT-CT module so lookign to design a super slim PCB.
All fits inside a 32mm water pipe and trialing different antenna designs.
And now of course they drop the new Mesh Node T114 at $27 which for sure is a bit more, but kinda does it all and super low power.
but i still dont understand why it has to be esp32, just to safe some 10 money…
if you want to have the ESP running on solar in winter time or cloudy times, you will have to spend alot more money for batterys and solar setup, than what you safe by choosing esp32 instead of nrf chip…
If you need more margin during the winter time you can can invest in a 18V panel which costs 2-3€ more.
Also Nordic based designs can profit from my PCB: the entire power chain can easily be adapted for another microcontroller. Just remove ESP32 and add the chip you like most.
I don’t think it that simplistic. Not that it ‘has’ to be ESP32, just that it a convenient platform.
… yes, the default mestastic firmware certainly seems better (power wise!) with the NRF MCU, but it doesnt mean it universally better. Also ESP32 has better support for some functioanlity, like MQTT/JSON, and Store/Forward.
That been said there are more NRF based modules coming out (including the mentioned Mesh Node T114) , so they may well overtake ESP.
Dont know where you based, but might be able to get cheaper elsewhere, if rather than direct from RAK
I still going to hold out hope that Heltec can do a ‘module’ version of the T114 - or at least a even more ‘cut down’ version of T114 for standalone routers - nice available without scrren and GPS, but even more could be cut off LEDS, much reduced IO, maybe even do away with bT/wifi antenaa, and the LEDs, and connectors.
Hi! Amazing design!
I have a question: why going for a LDO regulator and not a buck-boost like the TPS63020? It should have plenty of peak current for this project…
A buck-boost would normally need inductor(s) - that a nice chip in that only needs one, but still more than zero
… can be noisely depending on how well designed - not sure if that an issue here.
Also a LDO, seems like it ok for normal LI batteries.
