Closed VincentSC closed 1 year ago
Solution: upgrade to DTU SW-version to V00.02.15 (V0.2.0F).
Would be a good thing for the documentation.
@VincentSC thank you for the feedback. I've added troubleshooting to readme based on your comments
@wasilukm and @VincentSC Just adding to this topic, recent experience. Even with latest firmware on the DTU it keeps locking up. Exact same message as above.
python 3.10 DTU Hardware rev - H09.06.1 DTU Software - V00.02.15
Even though the DTU is nice for info, the connection drops seem to be quite often. (Even without mod-bus)
Seems to be firmware, as the network of the DTU is still online -> e.g. ping is working 3 Solutions I see as potential solutions:
Do you have ideas (ever since)
Same experience right now: the DTU is super instable. Connections to the cloud drop a lot here - sometimes twice a day, sometimes it skips 1 or 2 days.
Oke, good to have the confirmation that I am not alone. Hope they will finally address these issues (but I am not hopeful looking at the messages online).
What I found as an alternative (and looks like they support the HMS): https://github.com/tbnobody/OpenDTU
Also supports MQTT I use HMS-2000-T
Mmm, is that effort different from this team? https://github.com/lumapu/ahoy and https://discord.com/channels/984173303147155506/993500210271629393
I decided to just buy the official DTU after I saw how slow they were progressing, and it was impossible for me to join in due to the lack of a structured approach. Other projects were very quiet, and assumed that would remain. I seemingly bet on the wrong team, as the Ahoy team still has "coming soon" :)
From what I see (and remember), building the hardware is relatively easy. So going to order parts this weekend if I did not do that already. Thanks for the tip!
Offtopic: if I'm reading TU
and Oke
correctly, I'm guessing you're also on Tweakers.net. Do you know of any electronics-makers-forum for such efforts like the Germans seem to have more often? https://www.circuitsonline.net/forum/search?query=hoymiles&mode=and&starter_user=&message_user=&topic_type=0&sort=0 shows less than Tweakers.
Looking at the github sites it are different projects (probably share some code).
For me the openDTU looks a bit more professional.
Looking for gear now so I can start with that. Mainly the antenna board sourcing seems to be challenging.
@VincentSC I did send a mail to Hoymiles and they answered today. I need to send my ID so a technician will contact me and will investigate the matter. In the end the opendtu solution is nice, but the original DTU is easier (will try this route first).
Did the same. Emailed them SW-version, HW-version and they asked for S/N. Now waiting for a magic solution.
So far no result. Almost have the feeling they have lockup in their TCP stack. Once a day it resets and comes back online. No change in using fixed IP.
Would be interesting if the RS485 port is working fine, but as always not a lot of time to test.
Same problem here. They even stopped responding.
I ended up connecting a zigbee-powerplug, that turns off for 5 seconds and back on, every hour during the day.
If this works well for a week, I will let the restart trigger by a lack of mqtt-messages. Far from ideal, but I see no other option.
For now disabled the modbus probing and it is more stable. However no data in HA... When more time I will most probably (if hoymiles will not come with a decent solution), switch to the opendtu.
Description
Just trying to connect for the first time. I get this error 100% of the time:
pymodbus.exceptions.ModbusIOException: Modbus Error: [Input/Output] Modbus Error: [Invalid Message] No response received, expected at least 8 bytes (0 received)
The Hoymiles app works well, and can read the data from all.
This software needs
libseccomp2
to be installed, else it won't work. I fixed that already.Restarted the device, but neither helped.
It seems to connect to the DTU correctly, as other ip-addresses give another error.
What I Did
Also tried the HM microinverter-type. When I add
--comm-retry-on-empty True
it keeps on looping.