Since you asked for it, here is my KiCad project as a zip archive (Google Drive link): https://drive.google.com/file/d/1FoM3wkLLuqLHVw9ZMFcP-gA8kyjV1oky/view?usp=drive_link
Try setting Router_Client not repeater.
Maybe that help.
Not Router only… but Router_Client.
And if you use a 5volt - 5watt solar panel (like me)
then don’t use to many batteries, 2 x 18650 is enough.
RAK4631 and 2x18650
Solar 5volt - 5watt.
Adding more batteries is asking for trouble because the won’t charge all the way up.
With this configuration i get 90% full and that’s fine.
Batteries are each 1300mAh so about 2600mAh in total.
I want to mount a solar node somewhere where I might not have access again. The T-beam and RAK 4631 have the problem that they do not re-start automatically when the power drains the batteries. They need manual assistance to re-start. Is there a known solution or better node hardware for a “place and forget” solar node? The only solution I can think of is a relay that switches on the power when the battery is full and thereby simulates a manual usb re-connection. In my car I have a T-beam plugged into the dash USB and the node switches on when the car switches on. No problem. That is what I need on a solar node: it should switch off when the battery is at a certain level due to long period of lack of sun, and when the battery is full again, just get reconnected “like when the car starts”. Any sensible way of doing this or similar?
From another forum this product was suggested as a solution as it switches the power on and off depending on battery voltage and who knows, that might cause clean voltage switch that make the devices (T-beam and or RAK) wake up when the sun has charged the battery again. https://www.waveshare.com/solar-power-manager-d.htm?sku=26363
Would the following help?
2.5-9V Fine-Adjust Step-Up/Step-Down Voltage Regulator w/ Adjustable Low-Voltage Cutoff S9V11MACMA
Product Description:
Specs:
If I understand this correctly one could put this regulator between a max 9V solar panel and the Wisblock solar connector and adjust it to:
- the desired output voltage (standard operating voltage of the Wisblock; see footnote 4 in the specs) and
- the desired low voltage cut off value (3.6V?).
Correct?
I changed my enclosure design so it is more watertight. Everything now sits on a caddy that slides into the enclosure.
I put some STLs on my google drive, if anyone wants to have a look. This uses the RAK 19007 baseboard and the pcb LORA antenna.
The gasket is printed in NinjaFlex.
7 Likes
Whoa, I did some tests, so the ASA plastic plus the spray paint leads to a signal loss of 10dB !!! So I guess this design is out. Will incorporate a standard SMA antenna then. At least climbing trees is fun.
So, don’t do this then, looks cool, but the army green paint must have lead or barium or whatnot in it that totally kills the signal.
Although the ubiquitous stubby SMA antennas are kinda bad without a counterpoise - maybe i need to incorporate a counterpoise into the enclosure somehow.
Take care that your darkgreenish spray can heat up you box a lot.
LiPos must be kept between 0-45°C for charging.
A trick might be to put them behind the solar cell with some spacers (panels tend to heat up a lot too) so it indirectly can act as a sun shield.
I made a RAK node that does re-start in tests once flat. Not sure if that situation occurred in the field yet. This is it: Solar node for top of tall building using the Waveshare D solar power manager
This is a bit off topic in terms of the node itself (since this is about the RAK systems), but I have now make a “restart testing kit” using a solar panel (Polysilicon Solar Panel (18V 10W), 10Wp Power Photovoltaic Panel, High Conversion Efficiency) and the Waveshare D Solar Power Management board (Solar Power Manager Module (D), Supports 6V~24V Solar Panel and Type-C Power Adapter, 5V/3A Regulated Output | Solar Power Manager (D)) and three 18650 LiIon protected cells (4000 mAh) (Keeppower 18650 - 4000mAh 3,6V - 3,7V Li-Ion-Akku mit Schutzelektronik (Pluspol erhöht) P1840C | akkuteile.de).
The idea is to be able to test any node for re-start after brownout. The T-echo was really stubborn - it did not want to switch off at all, even when the LiIon batteries attached to the Waveshare D were drained to their minimum, the T-techo went right on and started up and ran fine for a further couple of days. Eventually after 2 days the T-echo was completely unresponsive to button presses and not contactable via the mesh.
The outcome I will post here.
The setup can be used to test any node arrangement. It would be nice to be able to plug in different sorts of batteries, and the Waveshare D does allow for this.
My aim here is testing the T-echo for re-starting after a total power out due to having drained the three Li-Ion batteries and its own battery. I got very good radio results with the T-echo and this antenna (Taoglas FW.86.B.SMA.M e.g. from FW.86.B.SMA.M | Taoglas Antenna, ISM, 850 ... 890 MHz, 240mm, 3.5 dBi, Male SMA, Screw | Distrelec International) so was keen to use this one with a broken screen as a solar node… if it can re-start of its own accord.
The test board is made with perspex (plexiglass) sheet to mount the parts, and then the sheet is fixed to the solar panel.
The T-echo did wake up automatically when the pannel was left in the sun.
Thus far the batteries of the Waveshare have not charged, but the T-echo did wake up, having been previously drained to the point where the buttons caused not lights to light and no message from nor traceroute to nor nodeinfo was available for the device. (This result is consistent with reports of nodes that use a T-echo in a tupperware box and a solar panel with USB C output plugged into it.) Which is all fine… provided tomorrow when the sun shines the batteries of the Waveshare D also charge… so far so good…
BUT the T-echo needed a button press to send and receive so it cannot be used as a solar node
Now testing waveshare D with T-beam S3 Supreme… with no battery in the Supreme.
I looked up the cells shown in the picture because I could not believe the low capacity number you stated. Turns out these are 2900mAh, so your resulting capacity is 5800mAh total.
Duh, I feel stupid now. But thanks for stating the not so obvious.Why am I faffing around trying to stuff everything into the dumb enclosure with the RAK baseboard when I could so easily put the battery with the charger+voltage supervision board behind the humonguous solar panel? I wasn’t seeing the forest for all the trees.
No worries.
I learned this the hard way with spending a lot of money and time into dysfunctional overheating designs 
After reading this thread for a while and having same issues with brownouts i decided into getting a bit into PCB design.
i started with @michimartini design and after the charge chip was not available at jlcpcb, i started a new revision and another one after reading the datasheets 
Anyway currently rev3 is with jlcpcb to get build, with shipping i guess 14days from now, i will test it.
I also plan to opensource the designs on github, in fact i already started the repo, its just private atm.
@michimartini are you ok with me putting your design there too?
I also have plans for other improvements, but i want to test everything first before designing and ordering another test batch.
Yes, very cool of you to improve on my design and making it more accessible. Make sure you get a charger IC that does have a charge indicator pin, so you can hook up an LED, so you’ll know when it is charging. Also, if it can charge with more than 500mA that would be beneficial for people that use 10w solar panels as those can source up to 2A.
Anyhow, excited to see what you come up with.
1 Like
Yeah, but this would need some way to configuration i guess, maybe some kind of jumper.
I’m currently waiting for my “rev3” board
Basically:
rev2: (never built)
- Switch to MCP73812T-420I-OT
- 500ma
rev3:
- Switch to MCP73831T-2DCI-OT
- 500ma
- Added charge LED
possible rev 4:
- Possible switch to CN3065
- Charge + charged LED
- USB connector for charging
The CN3065 is interesting because it supports up to 1000ma and also temperature sensing.
Could be directly on the board i guess, since the whole node will be in a box.
Other features i want to explore are:
- custom base board for RAK4631, via cpu port (fairly complex regarding footprint etc)
- possible directly add a INA219 instead of the need of a RAK16000
2 Likes
Also have a look at the BQ25185. The chip super cheap and supports Power Path’ing.
It only needs 3x Cs and 2 Rs.
@dm5tt yeah saw your thread, you wrote it tends to overheat. in the MCP* Datasheet it also mentioned adding a bunch of via’s to fix that.
How did you fix your overheating issues? I’m fairly new to PCB design Just hoping my changes actually work jlcpcb just shipped it, so i’m eager to test ^^
The power pathing/load sharing is an interesting feature. I dont think it will be a big problem with the RAK chips, since they draw so little power but i saw this article (+part2) a while back.
As far as i understood it, if the charger does not support it, it might be overcharging the lipo, since the load causes a voltage drop.
Edit: i just hope bms in battery+low draw mitigate this
In my version v0.6 I 've made large ground areas with many vias to the back. Problem seems to be fixed. The chip still gets warm - but this heat is now dissipated and distributed very well.
Yes - you will need a charger with a dedicated “Load”-Pin where you can connect your RAK to. I’ve seen weird effects when the load is directly connected to the battery which is getting charged by a controller.
Ok, the rev3 board came today and i did some measurements and its now connected to a rak, solar panel and battery.
I will do tests over the next couple of days to see if the rak wakes up again after the battery runs flat, if battery is actually charging etc.
Its the first batch of tests, since currently the battery also has a bms and the panel has a boost converter to 5v. But without the board, it never came back with those components before.
Anyhow, i also now made my github repo public:
It contains rev1 (@michimartini pcb), rev2 and the now build and in testing rev3.
1 Like