Repository navigation
PVS5 observations #21
Replies: 11 comments 10 replies
|
Thanks for the detailed investigation! This is really valuable info on PVS5 compatibility. On BUILD 5408: That's fascinating - SWVER "2025.10, Build 5408" means your firmware is from THIS MONTH (October 2025), It appears PVS5 uses a different BUILD numbering scheme than PVS6:
Critical question for integration compatibility: Our integration uses BUILD >= 61840 to detect new firmware with authentication. But your PVS5 might have authentication too, just with different BUILD numbers. Can you confirm:
On /supervisor/info: The integration has a fallback when /supervisor/info isn't available - it tries pypvs first, then dl_cgi device_list. Since you confirmed dl_cgi works, setup should succeed. On LAN connectivity: You mentioned unreliable home LAN connection. We recently discovered Raspberry Pi bridges/proxies don't work with new firmware (BUILD 61840+) because they can't handle HTTPS authentication on port 443. If your PVS5 has Testing request: Can you test the integration and share setup logs? Your PVS5 data would help us:
|
|
Ok thanks for the log very helpful!! Give me 30 minutes or so, and I will have a new version to try. Will reply again when it's ready. You'll need to install it via HACS as before, it will be a pre-release so only the couple testers will get it. v2025.10.17 |
|
v2025.10.17 is now ready! |
|
Good news, another user with a PVS5 is now up and working: #22 |
|
I could upgrade to 2025.10.17 and tried to add the integration. This time it succeeded in authorizing the connection to the PVS and then recognized the supervisor, meters, and inverters from the panels. So this looks good ! home-assistant_2025-10-23T13-06-19.510Z.log I think the integration may be using dl_cgi/devices/list. Let me generate a larger query of parameters available through the vars API on the system and share that as well. As you recognized the build numbering follows different schemes depending on the PVS type. It may be beneficial to use not just the build number but the build number together with the PVS type as a criterion if the new vars API should be used. |
|
New version is now live, actually released as a full release since it had other fixes and new features as well, not just PVS5 related. Look for v2025.10.18. Please let me know of any issues. |
|
I could retrieve all varserver variables which are published on my system. They are listed below. Out of the 68, 65 are listed in https://github.com/SunStrong-Management/pypvs/blob/main/doc/varserver-variables-public.csv and three are not:
Four are valued Two have Here is the full list: And here is the full list with some current values for my system: |
|
In case it's helpful, I've created a log of my new experiments with the PVS5 hardware under the SunStrong regime here: https://github.com/jeffkowalski/pvs/blob/master/HARDWARE.org It's part of the home-grown monitoring I've built using Influx, here: https://github.com/jeffkowalski/pvs/tree/master |
|
@jeffkowalski super. What is your PVS5 firmware build ? |
|
I could update to version 2025.10.18 and capture a debug log. pvpvs is using the LocalAPI (varserver). Looks good to me, I did not see errors. Here is the full log: |


Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Since the PVS5 is referenced in this repo a few times let me share a few observations I made experimenting with my early 2017 PVS5 (see krbaker/hass-sunpower#125 (comment) and krbaker/hass-sunpower#125 (comment)):
cgi-bin/dl_cgi/supervisor/inforeturns just an empty object{}cgi-bin/dl_cgi/devices/listworks and returns complete list and fields. First time I could see detailed panel data in 8 years.2025.10, Build 5408which I could not find mentioned anywhere else0.0.25.5408which looks offTBDwhich looks offDoes the integration assume that
supervisor/infois available ?Happy to look up more detail or background. Also saw the PVS6 Bluetooth vulnerability bulletin which may also apply PVS5.
All reactions