Raspberry Pi Setup

[jasonacox]

RPi Powerwall-Dasbboard

The Raspberry Pi is a great way to launch the Powerwall Dashboard. These little devices are low cost, high performing yet pull less than 15W for continuous operations. Additionally, these small board have both ethernet as well as WiFi connection, allowing flexibility in how users setup and run their Dashboard.

Benefits

• Low Cost ($80 for kit that includes latest RPi, power supply and SD card)
• Low Power (15-25W)
• Runs Debian Linux Variant (Raspberry Pi OS) 64bit
• Ethernet and WiFi
• Compact Form Factor: Measures 3.5" x 2.3" x 0.76" ( 885819.5mm )

Bridge Configuration

Requirements

I recommend buying a Raspberry Pi 4 with 4GB of RAM (or more). You will need a high performing, high endurance SD card (get the best rated version you can find). The SD card is the hard disk for the RPi and low cost SD's will wear out quickly, causing you grief. The entire Powerwall Dashboard stack can run on the RPI including the InfluxDB database. As with all installations, it is recommended that you have a backup plan in place to periodically copy a backup of the InfluxDB database (see backup tips).

Example Kit:

• Raspberry Pi 5 (16GB)
• USB-C Power Supply (25W)
• SanDisk 128GB Extreme microSDXC UHS-I Memory Card

Raspberry Pi - Setup

1. You will need to create a boot image on the SD card - Download and launch the Raspberry Pi Imager to create your boot SD. Select the Raspberry Pi OS (64-bit).

2. Click Next and Select "Edit Settings" and enter the details for your hostname, user and locale. If you plan to use the RPi to bridge between your LAN and the Powerwall Gateway WiFi (TEDAPI) access point, configure the wireless LAN and use the password for your Tesla system that is typically on the QR sticker on the Powerwall. The access point SSID is usually in the format TEG-xxx or TeslaPW_xxxxxxxx.

4. Make sure you selet Services tab and Enable SSH.

5. Save and allow the Imager to create the boot disk. Once it is complete, install it in the RPi, connect the ethernet cable to your LAN and power on the device.
6. If you are using the terminal on the RPi itself, you can skip this step. If you want to log in to the RPi from another computer, open a shell on that system an log in to the RPi remotely by using the user, password and hostname you select in the Settings:

ssh pi@raspberrypi.local

Simple test: If you want to ensure your setup is working correctly at this point, run these commands:

# Check wifi status with one of these
wpa_cli status 
iwconfig

# Check connectivity to Powerwall gateway
ping -c 1 192.168.91.1

# Check internet connectivity
ping -c 1 google.com

Example session:

  jason@lab:~/Powerwall-Dashboard $ wpa_cli status | egrep "Tesla|ip_"
  ssid=TeslaPW_XYZABC
  ip_address=192.168.91.238

  jason@lab:~/Powerwall-Dashboard $ ping -c 1 192.168.91.1
  PING 192.168.91.1 (192.168.91.1) 56(84) bytes of data.
  64 bytes from 192.168.91.1: icmp_seq=1 ttl=64 time=6.97 ms
  
  --- 192.168.91.1 ping statistics ---
  1 packets transmitted, 1 received, 0% packet loss, time 0ms
  rtt min/avg/max/mdev = 6.969/6.969/6.969/0.000 ms

Docker Setup

You will need to install Docker to run the Powerwall-Dashboard software. Since we are using th 64-bit OS, we will follow the instructions for installing Docker on a Debian system. See the Docker instructions here. Here are the instructions as of Apr 2025.

# Add Docker's official GPG key:
sudo apt-get update
sudo apt-get install ca-certificates curl
sudo install -m 0755 -d /etc/apt/keyrings
sudo curl -fsSL https://download.docker.com/linux/debian/gpg -o /etc/apt/keyrings/docker.asc
sudo chmod a+r /etc/apt/keyrings/docker.asc

# Add the repository to Apt sources:
echo \
  "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/debian \
  $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
sudo apt-get update

sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

Once docker is installed, you will need to add yourself to the docker user group and restart the RPi:

# Add your user to docker user group
sudo usermod -aG docker $USER

# Restart
sudo reboot

Install Powerwall Dashboard

You can use the quick start setup to get the Dashboard installed and running. The instructions are here:

# Install the Software 
git clone https://github.com/jasonacox/Powerwall-Dashboard.git
cd Powerwall-Dashboard
./setup.sh

Powerwall Dashboard (v4.6.5) - SETUP
-----------------------------------------
Select configuration mode:

Current: None

 1 - Local Access   (Powerwall 1, 2, or + using the Tesla Gateway on LAN) - Default
 2 - Tesla Cloud    (Solar-only systems or Powerwalls without LAN access)
 3 - FleetAPI Cloud (Powerwall systems using Official Telsa API)
 4 - Powerwall 3    (Powerwall 3 using the local Tesla Gateway)

Select mode: 4

If your RPi is set up to connect to your LAN with an ethernet cable and to the Powerwall with WiFi, you should select option "4" - this provides extended metrics and high fidelity (sampling every 10s). The setup will auto-detect the TEDPAI access point on the Powerwall and will prompt you for the password (same one used to connect to the WiFi).

Follow the rest of the instructions...

Timezone (leave blank for America/Los_Angeles)
Enter Timezone: 

Setting credentials for Powerwall 3...
Found Powerwall Gateway at 192.168.91.1
Use this IP? [Y/n] 

Congratulations!
Extended Device Metrics (vitals) are available on this endpoint via TEDAPI.
However, you will need the Gateway password to access them.
This password is often found on a QR code on the Powerwall.

Enter Gateway Password or leave blank to disable: XXXXXXXXX


Using America/Los_Angeles timezone...
-----------------------------------------

Attempting to determine your location...
   Your current location appears to be Latitude: 34.XXXXXXX, Longitude: -118.XXXXXXX

Enter Latitude (default: 34.XXXXXXX): 
Enter Longitude (default -118.XXXXXXX): 

Weather Data Setup
-----------------------------------------
Weather data from OpenWeatherMap can be added to your Powerwall Dashboard
graphs.  This requires that you set up a free account with OpenWeatherMap
and enter the API Key during this setup process.

Do you wish to setup Weather Data? [y/N] y
Forecast looks great!  Proceeding...


Set up a free account at OpenWeatherMap.org to get an API key
   1. Go to https://openweathermap.org/
   2. Create a new account and check your email to verify your account
   3. Click on 'API Keys' tab and copy 'Key' value and paste below.

Enter OpenWeatherMap API Key: XXXXXXXXXXXXXXXXXXXXXXXXXXXX

Enter your location coordinates to determine weather in your location.
   Your current location appears to be Latitude: 34.XXXXXXX, Longitude: -118..XXXXXXX

Enter Latitude (default: 34.XXXXXXX): 
Enter Longitude (default -118.XXXXXXX): 

Enter the desired units: M)etric, I)mperial or S)tandard where:
    M)etric = temperature in Celsius
    I)mperial = temperature in Fahrenheit
    S)tandard = temperature in Kelvin

Enter M, I or S: m
Units selected: metric

NOTE: The OpenWeatherMap key can take up to 2 hours to be valid.
      You may see errors or no data until it is fully activated.

Weather Configuration Complete

Running Docker Compose...

[+] Running 6/6
 ✔ Network powerwall-dashboard_default  Created                                                                            0.1s 
 ✔ Container pypowerwall                Started                                                                           23.7s 
 ✔ Container influxdb                   Started                                                                           23.7s 
 ✔ Container grafana                    Started                                                                            1.9s 
 ✔ Container weather411                 Started                                                                            1.8s 
 ✔ Container telegraf                   Started                                                                            1.8s 
-----------------------------------------
Waiting for InfluxDB to start...
... up!
Setup InfluxDB Data... ('already exist' errors harmless)
Executing single run query run-once-2.8.0.sql file...
Executing single run query run-once-2.8.3.sql file...
Executing single run query run-once-2.9.2.sql file...
Fetching local weather...
weather411
------------------[ Final Setup Instructions ]-----------------

Open Grafana at http://localhost:9000/ ... use admin/admin for login.

Dashboard Setup

Your system is running. The next step is to import the dashboard panels.

1. Navigate over to the Grafana dashboard: http://raspberrypi.local:9000 (replace raspberrypi.local with the local IP address of your Raspberry Pi or use the format name.local where name is the hostname you gave it during setup and your network supports mDNS) - You will be able to log in using admin/admin.

Troubleshooting: If you are having trouble getting to the Dashboard on your network, to find the local IP address of your Raspberry Pi, you can use the command: hostname -I or ip addr show eth0

2. From 'Dashboard\Browse', select 'New/Import', browse to /home/jason/Powerwall-Dashboard/dashboards
   and upload 'dashboard.json'.
3. During import, select the dropdown for data source (InfuxDB and SunAndMoon) and click Import.

5. The Dashboard should load and within a few seconds you should see the power flow animation and data starting to collect on the graphs.

Congratulations!

Upgrades

You can easily upgrade the Powerwall Dashboard software by using the built in upgrade.sh script:

./upgrade.sh

Upgrade Powerwall-Dashboard from 4.6.4 to 4.6.5
---------------------------------------------------------------------
This script will attempt to upgrade you to the latest version without
removing existing data. A backup is still recommended.

Checking Docker Compose...
Using Docker Compose V2, proceeding...

Upgrade - Proceed? [y/N] 

Troubleshooting and Tips

• The RPi is a great way to launch your Powerwall Dashboard. It has plenty of power and storage to last for years. However, as with any system, you should have a plan to safeguard the data. It is highly recommend that you periodically backup your InfluxDB database and store it on a separate devices, thumb drive or cloud location (see backup tips).
• Please comment below if there are troubleshooting or other tips you would like to see added.

Grafana Setup Help

This PDF show how to set up Grafana for the first time:
Powerwall Dashboard - Grafana Setup.pdf

[Nexarian]

@jasonacox A workaround for the SD card is you can buy an NVME hat (extension module) for the Raspberry PI: These may only be compatible for the PI 5 though.

YMMV of course, but I use https://52pi.com/products/p33-m-2-nvme-2280-poe-hat-extension-board-for-raspberry-pi-5 for both POE and a full-size 1 TB NVME. Mine doesn't even HAVE the SD card in it anymore.

[jasonacox]

Great suggestion!

Also, it is probably overboard, but I went with the Panda Wireless PAU0D AC1200 w/Dual Antennas for additional WiFi - https://www.amazon.com/dp/B0B2QD6RPX?ref=ppx_yo2ov_dt_b_fed_asin_title - It allows me to place the antennas away from the case for a direct line-of-sight to the Powerwall access point.

[JohnJ9ml]

I'm ordering the AC600 to try this as well. My PW3 installs are on the outside wall of the garage next to the service entrance. I'm likely going to have to put the Pi out in the garage to be close enough for a reliable signal. At one point I braced for this and researched, and the Pi is supposedly okay up to around 135F. That's probably a good thing: we set the all-time high here last year at 117.4. We may be taxing more than the AC with this setup!

[Nexarian]

That panda wireless looks awesome, I'm ordering it now, I may return the less substantial one I ordered.

@JohnJ9ml You make a very good point: Not every system can easily support wifi. I think it's reasonable to give feedback to Tesla that installers need a way to use Tesla One wired because it may not always be feasible to physically access the system on a support call.

[cwagz]

I just moved over to a RPi 5 and will be doing an NVMe hat this afternoon. I went all out and got the 16GB version and am mounting it on the wall next to my gateway 2. So far it is working great following the instructions above. I am still on 25.2.1 so I am hoping the firmware update will be seamless now that I am moved over to using TEDAPI. Thanks for the writeup, and all the work keeping up on this.

[Nexarian]

@cwagz I generally agree that when buying computers, you should get the highest tier your wallet will allow. My 1 GB RPi 3 is more/less useless. In 2017 when I got it, that was more than enough RAM, but times change!

[JohnJ9ml]

Looking over the instructions, I'd like to ask ahead of time on a slightly different approach:

My Dashboard instance is already running on an RPi 5 I bought in November specifically for this purpose. I just bought a Panda Wifi dongle to add to the device. My thinking is this:

• I configure the Wifi setup to connect to the primary PW3
• I remove the existing 192.168.91.1 route
• I restart the device
• My belief is that I can then run the Dashboard setup.sh, pick option 4, and things should work as above. Is that correct?

The one thing I don't quite see from the instructions above is where - if anyplace - you configured 192.168.91.1 to point to anything. Is that a missing step, or will setup.sh just work all this out and put the permanent ip route in place?

[Nexarian]

We can only speculate as to why our preferred way to get data was blocked. I agree it seems bad and forcing people onto wifi-only is a highly suspect decision. I assume that if there is a workaround, the community will eventually find it as this firmware goes out everywhere. I will certainly try my hardest (I was already updated), but first I want to get my dashboard functional again.

[cwagz]

I was running hard wire from the beginning but I must say this is working very well on the RPi 5 with WiFi for the gateway and hardwire connecting the Pi to my local network. After adding the NVMe drive everything feels just as fast or faster as my previous VM that was running on an older Xenon D-1520.

[JohnJ9ml]

I set this up today for the dashboard Pi. I could see the SSID but not connect to it from upstairs in the office. I moved things downstairs to the living room, wondering how it would play out. Signal was 36/70, and there were a lot of excessive retries.

It would also quit working after 15 or 20 minutes. I decided I was just too far away.

It's now about five feet and one wooden wall away from the PW in the garage, plugged into the same switch the PW LAN connection is. I expect it will be fine in the temp - or not. North facing wall, at least it will only be the temp of the room and not additional convection heating.

For a while, it would also work for 15 or 20 minutes, but then quit. It finally dawned on me that I'd not entirely taken the old 192.168.91.1 route out - the TEDAPI_routing.sh script was still running on boot! Oof!

[jasonacox]

Yes, you will need to remove that old route. 😉 Did you get it working?

As to temp, if the RPi is in a hot area, you may want to add an active cooler (fan) to the case. During normal operation, the Pi 5 can reach 65°C (149°F) at idle and may exceed 85°C under heavy load without cooling. Thermal throttling will begin when the CPU temperature reaches around 80°C (176°F). I just tested my no-fan RPi 4's (via vcgencmd measure_temp) running the dashboard and they are 54°C and 48°C in my ambient 28°C garage.

[JohnJ9ml]

Yes, it's all working fine, and it was just a laugher that I'd forgotten about the cron script. I now see I didn't make it entirely clear.

I'm going to monitor the temps on the box. The daily temps in the area are already starting to climb (time to wake those grapevines up!), and it will give me a good chance to monitor the temps as I go.

I've been buying a combo package such that all my RPi 4s and 5s have active fans in them. I'll track the temps as I go.

Today I move the other box that just runs my pypowerwall scripts and pushes the data. It already wasn't configuring quite the same way as the first one, so a bit more sleuthing. (Well, while I wait for that /py/XXXXX set of endpoints to show up in the proxy.... 😉 )

[Nexarian]

Here is how to do it if you have two independent gateways/site controllers from a single RPi with 2 wifi connections to each:

sudo ip addr add 192.168.92.100/32 dev lo
sudo ip addr add 192.168.92.101/32 dev lo

# Route for wlan0
sudo ip route add 192.168.91.0/24 dev wlan0 table wlan0table
sudo ip route add default via 192.168.91.1 dev wlan0 table wlan0table

# Route for wlan1
sudo ip route add 192.168.91.0/24 dev wlan1 table wlan1table
sudo ip route add default via 192.168.91.1 dev wlan1 table wlan1table

sudo ip rule add from 192.168.92.100 table wlan0table
sudo ip rule add from 192.168.92.101 table wlan1table

sudo iptables -t nat -A OUTPUT -d 192.168.92.100 -j DNAT --to-destination 192.168.91.1
sudo iptables -t nat -A OUTPUT -d 192.168.92.101 -j DNAT --to-destination 192.168.91.1

sudo iptables -t nat -A POSTROUTING -s 192.168.92.100 -d 192.168.91.1 -j SNAT --to-source 192.168.91.78
sudo iptables -t nat -A POSTROUTING -s 192.168.92.101 -d 192.168.91.1 -j SNAT --to-source 192.168.91.124

Where 192.168.91.78 and 192.168.91.124 are the IP addresses of each of your wlan interfaces respectively.

You would set PW_HOST to 192.168.92.100 and 192.168.92.101

I guess you would need my multi-gateway PR... I'll work on that :)

[Nexarian]

I fired mine up last night. I think we have to work on reducing the amount of requests that are going to the TEG network. Previously I believe this is what @mccahan was saying with "hammering." My dual inverter prototype is worse because I was carelessly loading up on duplicate calls. With a 100 megabit connection (I think that's all the inverters use), calls are cheap and I didn't notice. But with weak or dodgy wifi, calls are a limited precious resource and need to be used sparingly.

[JohnJ9ml]

I was talking about this suggestion: Powerwall API as proxy endpoints

Just putting a SOAP/REST style interface on the proxy for all the wonderful structures that come back from the Powerwall() API for me would just complete the picture. I know the hard work is done inside the library, since they can be fetched that way.

I will admit (and have previously done so) that this project did at least get me to write my first effort with a .py suffix.

[Nexarian]

If you'd like to try your hand at it, I recommend doing so. This project is a great way to learn.

That being said, I'm open to working on it, too, or even guiding you through it.

[JohnJ9ml]

I poked around a bit, and I can see the bulk of the work would e done in pypowerwall/proxy/server.py. I know the only real effort is making sure all the output was JSON payload, but I seriously am just an old Java programmer who never did Python, so I have no Python debugging chops, and even worse have really no real experience as anything other than a project cloner on github.

Blasphemy times six, I'm sure, but it's the state I'm in, and all this sideline fun probably just keeps me off the streets and out of the wine bars. I could even imagine sending someone a private copy of my initial hack, but I just wouldn't know what to do with something that wasn't dropping in an updated .jar file. This is admitting a lot for me!

I really like this project, and this group of people. I'm just not much of an actual contributor, truth be told.

[JohnJ9ml]

Okay, okay....

So, it turns out this was pretty easy to do.

I'll look around to see that I got everything covered, and then I'll work on how to submit it to be "improved".

[JohnJ9ml]

For some of us, "git" is a word Walter Brennan said a lot in The Real McCoys.

I finished this up and it works great locally. I can't get through the permissions errors trying to submit the changes back, and haven't located anything online that seems to capably get me past my stuck point.

It does work nicely for me, though, and lets me call through port 8675 and get all the same JSON data structures I was getting out of the Powerwall() API.

[Nexarian]

Ok so I'm learning a lot about wifi and raspberry pi.

I ordered this and tried it for a few days: https://www.amazon.com/dp/B0B2QD6RPX

• Result: It's "OK." The problem is that this is likely a wifi adapter for the masses, not for nerds who need REALLY SOLID network connections. It will occasionally glitch, and when it glitches, because I have 2x connections, the wifi adapter will hop to the other inverter, completely destroying my dashboard.

• Right now I am using 2x and they haven't glitched: https://www.amazon.com/dp/B0BY8GMW32, however they are newer chipsets with Beta-ish Linux drivers, so you may have better luck with:

-- https://github.com/morrownr/USB-WiFi/blob/main/home/Recommended_Adapters_for_Kali_Linux.md

Words that consistently come up with Linux & USB & Wifi: MediaTek and ALFA

For my inverters the antenna is on the top. I recommend you figure out where you antenna is and snuggle your adapters as close as possible to it. Right now I'm at -29 dBm for one inverter and -49 dBm for the other. How I found where the antennas were on mine? I took a picture of the PCB and then asked ChatGPT where the antenna connector was.

Use iwconfig to check wlan nerdy metrics.

[pgatty]

FWIW I'm using an 8GB Pi 5 running Ubuntu 24.04 and I haven't noticed any gaps in my data on the Dashboard despite the 14k excessive retries. The Pi is about 15ft from the lead Powerwall 3 with one uninsulated sheetrock wall in between. I occasionally notice that the Power Flow graphic gets an error, but this isn't any different than when I was routing traffic though the LAN IP of my Powerwall.

iwconfig

wlan0     IEEE 802.11  ESSID:"XXXXXX"
          Mode:Managed  Frequency:2.462 GHz  Access Point: XXXXXXXXX
          Bit Rate=39 Mb/s   Tx-Power=31 dBm
          Retry short limit:7   RTS thr:off   Fragment thr:off
          Encryption key:off
          Power Management:on
          Link Quality=51/70  Signal level=-59 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:13912  Invalid misc:0   Missed beacon:0

ifconfig

wlan0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 192.168.91.78  netmask 255.255.255.0  broadcast 192.168.91.255
        inet6 fe80::e65f:1ff:fe00:5a1f  prefixlen 64  scopeid 0x20<link>
        ether e4:5f:01:00:5a:1f  txqueuelen 1000  (Ethernet)
        RX packets 23162583  bytes 6580071423 (6.5 GB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 25775003  bytes 10563749912 (10.5 GB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[jasonacox]

It would be good to get a list of reliable WiFi adapters for the RPi.

Similar to @pgatty, my RPi 4 has had solid wlan0 connection using the built-in WiFi (and using eth0 for LAN) an no additional adapter. Could it be interference issues with having more than one wifi radio?

[jgleigh]

Are the WiFi adapters required or do they just help if you're far from the Gateway and need to grab a weaker signal?

[cwagz]

Mine is working fine with the built in WiFi. I did mount it close to the gateway and am using a plastic case. I would assume the built in WiFi would be adequate for many people as long as the case has enough plastic to allow the signal through.

[mccahan]

I didn't want to move my Pi to be closer to my PW (and eventually I'm going to move off the Pi to my regular Docker server anyway), and I definitely didn't want to try working through different USB adapters and networking configs on the Pi, so I went with a $35 travel router bridge configuration.

Then you've got two options:

• If you have an ethernet port close, put the travel router there, set the TeslaPW network as the WAN with the WiFi clientmode, and make sure you disable the travel router's own DHCP server. Connect the ethernet to a LAN port.
• If you don't have ethernet close (like me), you can do a WiFi-to-WiFi bridge by setting the TeslaPW network as the WAN and then adding another interface on the second radio and just plug it in to power where it can see both the PW network and the home network.

In both cases, your router will now have an IP address on your network that you can use to create the static route for your Pi to access 192.168.91.1.

I made a mini write-up elsewhere.

N.B. the WiFi-to-WiFi works fine for me because I can use the 2.4GHz radio for the PW (and letting the router broadcast its own network for management, but will probably disable that for interference avoidance) and the 5GHz radio for my home network. If you're running everything on 2.4GHz, I think it might work if everything's on the same channel.

@Nexarian out of curiosity, do both of your inverters operate on the same channel?

[Nexarian]

It also occurs to me that, if we're able to get confirmation from Tesla that no path to a wired LAN will ever be re-enabled and we must use wifi, we should probably think about putting wifi stability metrics as an optional section on the dashboard, or create a wifi stability metric dashboard standalone.

This will not be trivial if we want to support Windows and MacOS in addition to Linux.

[jasonacox]

Agree, likely needed. We could have a bash or python script that telegraf runs simliar to how we import project version data:

[[inputs.exec]]
  commands = ["/etc/telegraf/ver.sh"]
  name_override = "powerwall_dashboard"
  data_format = "influx"

[cwagz]

I modified a case from Printables and designed a wall mount. This was a great opportunity for multi material printing on my new Prusa XL. The RPi 5 is working perfect with Ethernet to my network and the built in WiFi to the gateway. I will probably upload this as a remix on Printables and can post a link here if anyone is interested.

[jasonacox]

Very nice!! I want one! 😁

[JohnJ9ml]

The Powerwall as a yard equipment battery charger stand. Genius! :-)

[pgatty]

Argon 40 makes a basic NVME case for the Pi 5. I've been running several for nearly a year now and they've been great. It includes the M.2 NVME expansion board. Amazon sells them for $38

Argon NEO 5 M.2 NVME PCIE Case for Raspberry Pi 5

[cwagz]

Is this a plastic case? I was having trouble finding a plastic one so as to not block the built in WiFi. This looks like a really nice case if it’s plastic.

[pgatty]

Is this a plastic case? I was having trouble finding a plastic one so as to not block the built in WiFi. This looks like a really nice case if it’s plastic.

The top part is aluminum and the bottom part is plastic. The WiFi metrics I posted here are for a Pi in the older version of the Argon NEO 5 M.2 case. It's also an aluminum case with a plastic bottom.

[lignumaqua]

Small update. The Raspberry Pi4 I installed last week is still working fine. It's indoors, about 20 ft from the PW3s which are outside. It does get a fair number of logged 'excessive retries' but they don't seem to actually cause a problem. One point, the daily backup takes a LONG time to run on the Pi. That isn't an issue but worth noting. I'm using Syncthing on the Pi to copy the backup back to my main house server.

[anderep]

It would be nice to get influxdb to sync just the prior days data in a network share (appending the data to a network share database file). Which would make it quick.

[Nexarian]

So I have a lot more information on the way my dodgy inverter is behaving. I want to record this publicly for anyone who may face similar issues. I'm not necessarily looking for troubleshooting help (though if you have it, it's welcome). I'm mainly writing this down so that if someone in the future sees this, they know they are not crazy and this is a real thing that can happen with Tesla hardware.

One of my regular solar inverters works fine. It has never had any issues whatsoever. All 4 strings work, the Neurio is solid purple (connected and operating properly) and never changes after setup, and I can always connect to the wifi with a laptop/rpi/Tesla One.

The other one; however, appears to have a bad wifi card. When a Neurio is connected, the light will be solid purple for a time, and then start blinking blue/purple, and then eventually go to solid blue (not connected) or blinking blue (waiting for connection). It lasts for only a few hours max, and then the SSID stops broadcasting. Tesla One is also therefore gone. The only known way to repair it is to power cycle the inverter. No amount of changing the antenna on the Neurio, changing the direction of the antenna, etc has helped.

On the pypowerwall side, this is what that looks like:

04/21/2025 01:13:35 PM [pypowerwall.tedapi] [ERROR] Error fetching status: HTTPSConnectionPool(host='192.168.92.101', port=443): Max retries exceeded with url: /tedapi/v1 (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7fff876a4cd0>, 'Connection to 192.168.92.101 timed out. (connect timeout=5)'))

These errors will gradually get more and more frequent until the same thing happens: The SSID stops broadcasting. This takes about a day-and-half (so 36 hours). If I restart the thing nightly I don't have problems in the daytime as long as I leave the Neurio disconnected. The Tesla power animation, with its very frequent calls, seems to speed up the decay. It's as if usage of this card is radioactive. Note: This is regardless of signal strength. I've seen this happen at -20 dBm all way to -50 dBm.

I did not have any issues like this when the wired/static route approach was working.

[s0schen]

Thanks for the guide! Also switched my setup from a VM on a TrueNAS to a Pi5 because of the TedAPI ethernet lockout. Running the Pi in my server closet since I found an old unused Asus Router that has Merlin firmware and was able to extend the TEG wifi transmission.

[jasonacox]

In #621 (comment) @JonMurphy recommended a TP-Link AC1300 Dual Band WiFi Adapter -Archer T3U plus, for the RPi [Amazon] - Reporting good results.

[jasonacox]

Also @goodoldme in #651 (comment) recommended: BrosTrend 650Mbps Linux Compatible WiFi Adapter and updated drivers - https://linux.brostrend.com/

[Nexarian]

My experiences on wifi adapters thus far:

After a decent amount of testing, I can highly recommend the ALFA AWUS036ACH-C -- You do have to install the drivers manually (as of this writing this is the best source: https://github.com/lwfinger/rtw88).

I tried the ALFA AWUS036AXML, but the MediaTek chipset in it is just not ready for prime-time on Linux, the drivers and firmware are too flaky. It was fine but you really don't need the Wifi 6E for this purpose, and the additional complexity seems to be a hindrance. It doesn't report signal strength properly and I'm unsure if it reports errors well.

With the AWUS036AXML, I got -30 dBm (good). But with the AWUS036ACH-C? -10 dBm with no changes to the antenna configuration!

I also tried the Panda adapters, but the number of Tx excessive retries and other errors was off the charts, not so with the Alfas.

[jasonacox]

Nice. Thanks @Nexarian - I may try to keep a table of some of the findings from the community:

WiFi Adapter	Recommendation	Notes
ALFA AWUS036ACH-C	Highly recommended	Manual driver installation required
ALFA AWUS036AXML	Not recommended	MediaTek chipset is flaky on Linux
Panda Wireless PAU0D AC1200	Recommended	Dual antennas for improved signal strength but may see excessive retries
TP-Link AC1300 Dual Band WiFi Adapter - Archer T3U plus	Recommended	Good results reported
BrosTrend 650Mbps Linux Compatible WiFi Adapter	Recommended	Updated drivers available

[jeffbeagley]

I feel like this is probably obvious but I run my own grafana setup on a separate stack apart from this. Was super simple to just add the influx DB as a separate connection and repoint the dashboard to the pi setup. Worked perfectly. Thank you for this project.

[jasonacox]

Thanks @jeffbeagley! 😊

[Dave-Jones-1963]

Loving this and especially being able to scrape the data from the website for home assistant. Would it be at all possible to be able to change settings such as the backup reserve initially and ideally to read and write the utility rate plan data both rate and times?

Then I could completely drop the tesla integration from my home assistant setup as that really doesn't do what I need - especially as without cloud access you cannot get the battery reserve via that - this works perfectly thank you so much!

[jasonacox]

Thanks for the kind note @Dave-Jones-1963!

What are you are needing to scrape for HA? The pypowerwall proxy container exposes most of the data available from the Powerwall via API URLs (see https://github.com/jasonacox/pypowerwall/blob/main/proxy/API.md). Reserve and other key data is available directly via http://localhost:8675/csv/v2?headers (remove headers param to just get values).

Grid,Home,Solar,Battery,BatteryLevel,GridStatus,Reserve
-18.00,829.50,2760.00,-1910.00,31.39,1,20

I'm not sure what you mean by "read and write the utility rate plan." You can see utility TOU data locally via the http://localhost:8675/tedapi/config API. However, the only way to change that is via the cloud API.

[Slopecat]

Let me start by saying I am really loving this project. I have had it up and running for a couple of weeks and I have learned a lot so far. Thank you for sharing this with all of us.

My apologies if this is not the right place for this comment. I am new to Github and a great many other things that and needed to make all of this work. My install on my Raspberry Pi 4 went pretty well. Now everything just up and died on me and my Pi went in to a continuous reboot loop. I have no idea why. I am looking for help in getting it working again. Again, if this is not the correct place for this, can someone direct me to where I need to go?

I have the system back up, but the docker container would not start. I was getting some kind of networking error. After hours on Stack-Overflow I attempted to resolve this my removing the docker network. I did the following, perhaps not in this order. I followed whatever were in a couple of posts I found.

sudo rm -rf /var/lib/docker/network
sudo ip link set docker0 down
sudo ip link delete docker0

I eventually got to where docker was running and I attempted to restart my containers with compose.

user@raspberrypi:~/git-repos/Powerwall-Dashboard $ ./compose-dash.sh start
Running Docker Compose...
[+] Running 0/2
 ⠸ Container pypowerwall  Starting                                             0.4s
 ⠸ Container influxdb     Starting                                                 0.4s
Error response from daemon: failed to set up container networking: network xxxxxx not found

Well as best I can understand this, they won't start because the network they are looking for was deleted by me. My previous very basic knowledge of Docker had me thinking the compose script would likely be creating the network its containers needed.

What can I do to get things working again?

Thank you - any help would be much appreciated.

[jasonacox]

Thanks for the kind words and sorry your Pi decided to go gremlin mode. 😅 I have seen that using cheap SD cards in the past, but usually after several months, not days.

The “network … not found” error usually means Docker’s bridge/network state on the host is out of sync with what Compose expects. Since the /var/lib/docker/network directory was removed, we need to let Docker/Compose recreate things cleanly.

# Restarting Docker engine
sudo systemctl restart docker

# Pruning stale Docker networks
docker network prune -f

# Bring the stack down cleanly (no volume deletion)
./compose-dash.sh stop 
docker compose -f powerwall.yml down --remove-orphans 

# Bring it back up
./compose-dash.sh up -d 

[Nexarian]

If the SD card is going bad, you can get a PI hat that puts an NVME drive on instead. Something like this (I use the POE + NVME version)

https://www.walmart.com/ip/52Pi-N04-M-2-NVMe-to-PCIe-Adapter-for-Raspberry-Pi-5-Support-M-2-NVMe-SSD-2230-2242-2260-2280/5332432080

[Slopecat]

Thanks so much for the quick response! I did get things working again with your help.

I did have to provide the compose file name to bring everything down. It took me a bit to figure out which file to call. In case anyone finds themselves having to do this, here is the one change I made:

docker compose -f powerwall.yml down --remove-orphans

I would hope my sd card isn't the problem. I did try to find a good one for this project. The card was brand new when I set this project up. I went with a Samsung Pro Endurance 128GB. It claims to have to "High Endurance for Video Monitoring up to 70,080 hours". It appears to be quite popular in dash-cams and video monitoring applications. I thought it would be a good choice for this where data is frequently being written to the database.

Thanks for the info on the NVME drive too.

[jasonacox]

Thanks, @Slopecat ! I updated my instructions to include your correction. Glad you are back in business!

[Slopecat]

Unfortunately something has gone wrong on my end I and need to call on the community for help. It appears that my pypowerwall container is unable to get data from my powerwall. I have spent the last six hours trying to figure out what is going on and I am running out of ideas.

First of all I have three separate computers set up to run this. Its kind of a long story. I had two running and one shut down. Both of the two that were running stopped pulling data at the same time yesterday so I suspect something changed on the powerwall side of things instead of one of the computers. I fired up the third one that had been off and it is behaving the same way. I made no changes to my powerwall. I have no idea if our installer did or if the system updates firmware or anything else on its own.

I can connect to the powerwall (v3) just fine with the Tesla app on my phone. I was also able to use the Netzero app to connect and download a CSV file so its not dead.

Here is was I get when I run the verify.sh script:
`
Checking pypowerwall

• Config File pypowerwall.env: GOOD
• Container (pypowerwall): GOOD
• Service (port 8675): ERROR: Not Listening
• Version: Unknown
• Powerwall State: ERROR: Not Connected - Firmware: Unknown
• Powerwall Gateway TEDAPI: Not Available (192.168.91.1)
• Cloud Mode: NO
  `

Here is some of the log ouput from the pypowerwall container:
11/01/2025 12:50:13 PM [urllib3.connectionpool] [WARNING] Retrying (Retry(total=2, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f9f3dd240>, 'Connection to 192.168.91.1 timed out. (connect timeout=5)')': / 11/01/2025 12:50:26 PM [urllib3.connectionpool] [WARNING] Retrying (Retry(total=1, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f9f3dd3c0>, 'Connection to 192.168.91.1 timed out. (connect timeout=5)')': / 11/01/2025 12:50:47 PM [urllib3.connectionpool] [WARNING] Retrying (Retry(total=0, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f9f3dd5a0>, 'Connection to 192.168.91.1 timed out. (connect timeout=5)')': / 11/01/2025 12:50:52 PM [pypowerwall.tedapi] [ERROR] Unable to connect to Powerwall Gateway 192.168.91.1 11/01/2025 12:50:52 PM [pypowerwall.tedapi] [ERROR] Please verify your your host has a route to the Gateway. 11/01/2025 12:50:52 PM [pypowerwall.tedapi] [ERROR] Error Details: HTTPSConnectionPool(host='192.168.91.1', port=443): Max retries exceeded with url: / (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f9f3dd6c0>, 'Connection to 192.168.91.1 timed out. (connect timeout=5)')) 11/01/2025 12:50:52 PM [pypowerwall.tedapi] [ERROR] Failed to connect to Powerwall Gateway

All of the other containers pass all tests.

When running the NetZero app on my phone it reports the Tesla Gateway version as "25.34.3 82272c3b"

Here are some of the other steps I have taken:

• reboot each of the three computers
• rerun the setup.sh script
• used docker compose to bring the stack up and down - no errors there.

What else do I need to be looking at? It does appear that the WiFi connection to the Powerwall on each system is up and working.

Thank you

[Nexarian]

On a computer that is connected to the powerwall wifi, run this:

python3 -m pypowerwall.tedapi --gw_ip 192.168.91.1

Enter the wifi password when prompted for Enter Powerwall Gateway Password. Do you get any output?

The other thing that occurs to me is you may need to reset the powerwalls themselves.

[Slopecat]

It seems I did not have pip or pypowerwall installed. I created a virtual environment and install it with pip. Here is the output:
`
pyPowerwall - Powerwall Gateway TEDAPI Reader [v0.14.2]

• Connecting to https://192.168.91.1... FAILED

ERROR: Unable to connect to Powerwall Gateway 192.168.91.1 on port 443.
Please verify your your host has a route to the Gateway.

Error details: HTTPSConnectionPool(host='192.168.91.1', port=443): Max retries exceeded with url: / (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f7ed13cb0>, 'Connection to 192.168.91.1 timed out. (connect timeout=5)'))
`

I am not sure how to verify that I have a route to the gateway. However, if I run arp -a one of the lines of output is the following:
(192.168.91.1) at de:2e:97:a1:01:4c [ether] on wlan0

wlan0 is up and running. It has an ip address of 192.168.91.44. I get no response when I ping 192.168.91.1 which I believe is the Powerwall gateway.

I am seeing the same thing from two completely different servers that were working fine for several weeks until yesterday.

[jasonacox]

@Slopecat Are you able to connect to the Powerwall with the Tesla One app? The fact that you can't even ping the 192.168.91.1 endpoint would indicate that somehow it is not really connected or the Powerwall access point is in bad state. I would try connecting with the Tesla One app (or does NetZero still connect) and if that works, it would indicate something off about how your systems are connecting to the access point. I would disconnect all the systems fromt he wifi, reboot my WiFi router in case it is hosting some stale ARP data, and try to reconnect the WiFi again.

If the Tesla One app doesn't work or the above fails, I would try a full Tesla system restart (of Powerwall).

[Slopecat]

After fooling with this for about 12 hours today, I think I am back in business. In the end, I think a power cycle of the "TACO" is all that really mattered. I figured out where the Gateway was outside of the house and turned off both breakers (one 60A and one 30A) and waited for an hour as suggested. I turned them back on and then the power switch on the Powerwall itself. Within 5 minutes the data started flowing back into pypowerwall as shown in Grafana without touching a thing on the computer(s).

Thanks again for all of the help. Every time so far it has been quite the learning experience.

[Slopecat]

After doing this hard reset and bringing it back online I can now connect to the Powerwall with the Tesla One app as well. This is the first time using it and it is handy to see the additional data it can access.

[jasonacox]

Awesome! I'm glad you were able to get it working!

My old PW+/PW2 system has rock solid WiFi but, sadly, the PW2 faulted and Tesla is going to replace it. They suggested replacing all of my system with a PW3. I confess, part of me would like that to help with the project as a majority of issues come from the PW3 and it's shady TACO, but also for that same reason, I would rather get a PW2 and keep my solid WiFi. 😉

[Nexarian]

@jasonacox I continue to delve deeper and deeper into debugging my TACO. I found these articles to be interesting. One thing to know is that Tesla engineers are EXTREMELY lazy. The hardware inside these things is effectively from a Tesla vehicle and a ton of the engineering is reused from product to product and generation to generation. As an example: The antenna on a Gateway 2 is identical to that of the antenna on my inverter.

• https://energylibrary.tesla.com/docs/Public/EnergyStorage/Powerwall/3/InstallManual/BackupGateway/2/en-us/GUID-A8866096-18CD-499C-8766-59ED3CD3E3FC.html

• https://energylibrary.tesla.com/docs/Public/EnergyStorage/Powerwall/3/InstallManual/BackupSwitch/en-us/GUID-570320A6-25C3-48F1-B96D-7CFB4E1FF573.html

[jasonacox]

The Load Shedding Feature is pretty nice! Has anyone set that up?

[heynorm]

I have set this up on my PW3. Specifically, I have the PW3 load-shedding relay connected to a Shelly wifi relay/switch that is monitored thru my Hubitat home automation hub. The Shelly device just looks like a switch device to the Hubitat (Grid is either On or Off). I have some rules in the Hubitat to shut off certain loads if a grid outage occurs (which I am still refining). It was only a few PW firmware releases ago that the internal PW load shedding relay started working (and I am still suspect that they have finalized and fully test the behavior). Since it started working, I have noted things like the relay will open briefly during some firmware upgrades (as the article says). So my load shedding rules must see the grid down for 10 min before actually shedding any loads. It is a useful feature once it is stable.

…

________________________________ From: Jason Cox ***@***.***> Sent: Sunday, November 2, 2025 7:59 AM To: jasonacox/Powerwall-Dashboard ***@***.***> Cc: Subscribed ***@***.***> Subject: Re: [jasonacox/Powerwall-Dashboard] Raspberry Pi Setup (Discussion #607) The Load Shedding Feature is pretty nice! Has anyone set that up? — Reply to this email directly, view it on GitHub<#607 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/APUXY3GZJAMFP77ZUT523CL32YL5ZAVCNFSM6AAAAAB2RJ4SRCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTIOBVGE2TSMY>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[jasonacox]

Thanks @heynorm ! Fascinating use of that and also interesting that it is just now coming online with the firmware update.

For my setup, I have been monitoring http://pypowerwall:8675/csv/v2?headers=1 and when GridStatus goes to zero, I have automation that turns on a red "GRID Offline" light. Our neighborhood occasionally has power outages and we never even knew we were on battery. With the bright alert light, my family knows not to fire up the oven or stove, for instance. I haven't seen any issues (false positives) but I'm wondering if your automation may be more reliable.

The fact that it goes "offline" for firmware upgrades may be a way to preempt any power related anomalies of having a heavy startup load (like a heat pump) fire up while the system is trying to come back online. Or as @Nexarian observes, maybe just lazy coding. 😆

[edk333]

I bought the hardware a while back (Raspberry Pi 5 with an NVME hat instead of an SD card) and finally broke down yesterday and ran through this process of installing the necessary pieces to get the Powerwall Dashboard running. I know just enough about the Linux environment to be dangerous, so I tried to follow the instructions as closely as possible. I did a clean install of Powerwall-Dashboard and a clean influx database. Everything went surprisingly well. The only thing that is not showing up on the dashboard is the inverter power section. All other areas of the current dashboard are being populated.

My system consists of a Gateway 2, a Powerwall + and a Powerwall 2.

I poked around in the inverter power section of the dashboard and it looks like it's getting it's data from the strings area for field inverter1. I ran http://localhost:8675/strings for both the windows setup and the raspberry pi setup. It's still early here so not any real data currently but I don't see anything about inverter1 in either of the outputs, so I may not be understanding it all:

From the Windows setup
{
"A": {
"Connected": true,
"Current": 0.01,
"Power": 0,
"Voltage": 0.6000000000000014
},
"B": {
"Connected": true,
"Current": 0,
"Power": 0,
"Voltage": 0.40000000000000036
},
"C": {
"Connected": true,
"Current": 0,
"Power": 0,
"Voltage": 0.8000000000000007
},
"D": {
"Connected": false,
"Current": 0,
"Power": 0,
"Voltage": 0.9000000000000004
}
}

From the RPi setup:
{
"A": {
"Connected": true,
"Current": 0,
"Power": 0.0,
"State": "PV_Active",
"Voltage": 0.6000000000000014
},
"B": {
"Connected": true,
"Current": 0,
"Power": 0.0,
"State": "PV_Active",
"Voltage": 0.3000000000000007
},
"C": {
"Connected": true,
"Current": 0,
"Power": 0.0,
"State": "PV_Active",
"Voltage": 0.7000000000000011
},
"D": {
"Connected": false,
"Current": 0,
"Power": 0.0,
"State": "PV_Disabled",
"Voltage": 0.8000000000000007
}
}

I ran the verify.sh on the RPi setup:

I answered yes to the display of the last 10 log lines for each container. The messages appear to be from prior evening/earlier this morning:

I'm not sure if it would be helpful to run a verify on my windows setup to help troubleshoot or not.

Anybody have any ideas on where I may have made a misstep to just be missing this one piece? Thanks in advance.