MPP3 working for SH15T and SH25T

[RafAustralia]

Hey,

I have prepared these sensors for my MPPT3 and they are working on my end:

  - name: MPPT3 power
    unique_id: sg_mppt3_power
    unit_of_measurement: W
    device_class: power
    state_class: measurement
    availability: >-
      {{ 
      not is_state('sensor.mppt3_voltage', 'unavailable') 
      and not is_state('sensor.mppt3_current', 'unavailable' ) 
      }}
    state: "{{ (states('sensor.mppt3_voltage') | float * states('sensor.mppt3_current') | float) |int }}"

  - name: MPPT3 voltage
    unique_id: sg_mppt3_voltage
    device_address: !secret sungrow_modbus_slave
    address: 5014 # reg 5015
    input_type: input
    data_type: uint16
    precision: 1
    unit_of_measurement: V
    device_class: Voltage
    state_class: measurement
    scale: 0.1
    scan_interval: 10

  - name: MPPT3 current
    unique_id: sg_mppt3_current
    device_address: !secret sungrow_modbus_slave
    address: 5015 # reg 5016
    input_type: input
    data_type: uint16
    precision: 2
    unit_of_measurement: A
    device_class: Current
    state_class: measurement
    scale: 0.1
    scan_interval: 10

[Gnarfoz]

Do you have a new firmware or something?
Nobody seems to have accomplished it this (default) way before.

[RafAustralia]

No. Its not a firmware update. Instead, I have worked directly with code writer from Sungrow R&D china team to enable the missing registers. And just to mention… It’s been achieved before but via use of winet extractor add on in HA. I however in my latest setup do not use winets. So that option was not available to me. I have my inverters sitting behind a logger1000a via com port. Winets are not even installed hence why I worked with them to resolve these registers. Message ID: ***@***.***>

[RafAustralia]

Hi Christian
At the time I was working with Sungrow in parallel on so many issues that everything is sort of a bit of a blur…
We were doing everything from checking their spelling and missing code and mistakes in protocol…
When I say that – it was intense 10-20 emails a day for 3-4 months process.
What I have found and attached below is the last email I found in the link from vice director of code for sungrow, on that topic, and after that my sensor just worked… and I actually mean – my next reply (posted below) is – it works!...
Perhaps they have pushed some change to the code after 100 failed attempts,
Or there can only be one more explanation.
My equipment is an extremely differently connected bundle of inverters and batteries.
Quite rare amongst other users...
I do not have winets anymore, and I do not have ethernet cables either.
Chint meter is also not connected to the inverter, so it is whole another story all together…
Everything is connected via com1 and com2 cables to the LOGGER1000A, and the logger communicates with the HA, and perhaps and just perhaps amongst all that these registers read better via com port connection? I am not quite sure…

See below my last and final piece of communication with sungrow on that topic:
(it has some other info you may find useful – shows how many topics we covered)
I will go back and have a look through my notes and my code how did we make all of it happen... but like I said - there were 100 failed attempts in between...

Hope that helps to explain this process somewhat...

[RafAustralia]

Hi Christian,

Few more snips from my HA, checking the sensors, as I really could not remember how we got there...

I think because my installer has installed two strings on mppt2 and mppt3 of my SH25T, I was stuck with mpp3 not working (once winet was gone) and the only difference in its settings I can see are these highlighted in yellow

Thats what I found so not sure if that helps at all...

(I have scrolled million messages on discord as well but could not get anything better on that topic... will look more later)

One thing you may wanna check... and I vaguely recall playing with Sunspec as the way to bring inverter registers across, and I use it as the way of importing registers for my other gear... such as my chint and my logger

but anyhow - hope all of this is helpful.

[Gnarfoz]

Everything is connected via com1 and com2 cables to the LOGGER1000A, and the logger communicates with the HA, and perhaps and just perhaps amongst all that these registers read better via com port connection? I am not quite sure…

This is the key to why this works for you. Using a RS485-to-Ethernet converter on the COM2 "logger" ports (A1 / B1) is a common recommendation for SHxT users and achieves the same result. See for example #298 (comment).

[RafAustralia]

hey Christian,

the issue there you describe in 298 is very similar in essence but not the same...

Yes it is Sungrow's direction to connect multiple inverters via logger, but it is not a common occurrence here.
They are moving away from master slave setup overall so they introduced a logger to manage their EMS.
As far as I know my setup was first they have done in residential market here hence why they were keen to do it themselves.

this set up has other issues associated with logger being in place... a lot of sensors disappear, cos chint is removed...
So it is a two way street in a way - some benefits overall and some serious complications for HA, such as inability to measure import and export power sensors... we had to create complex formulas to mimic the isolarcloud behaviour.
Logger at this point in time cannot be controlled via HA, in a way of initiating actions or issuing commands... logger can be read, but thats about it... so logger has helped and has ruined a lot of things also... thats why I have put it in open-loop mode (pass through) and resorted back to pushing on with HA to make it work again... its been a journey... :)
I now run 4 different modbus files and the secret there is that it is recommended each modbus gains its own port... hence why 502, 503, 504, 505 etc...

[Gnarfoz]

Thank you for elaborating, but the root issue still seems to be the same. The MPPT3 registers can't be directly read via WiNet-S (only in bulk with others) but they can via the COM2 interface. In your case that's through the Logger1000, for other people it's a simple converter. The result is the same: it only works if you're accessing these registers via COM2.

It's something that Sungrow needs to fix in the WiNet-S firmware. Known workarounds:

• Using Modbus via COM2 via for example adapter or logger1000
• Using WiNet REST API extractor to avoid Modbus altogether
• Reading the registers as part of a bulk read operation

[RafAustralia]

Hi Christian Fair enough comment – but along the way – in process - I did try to pull sensors through third party integration (e.g. Sunsync) so wondering if it played the role in exposing them better… but its been such a long time since I really cannot tell… maybe it is just com2…