Interpretation of modbus data

[h0bb3]

As it seems it was more complicated to interpret the modbus data and there seems to be custom endianness and sizes of even the raw data.

One suggestion would be to interpret everything read from modbus as a 16bit number and name these Aa etc... and allow for a rearrangement of the the data. E.g. reading 4 registers and using the high bytes of each word to form a 32 bit integer big endian integer would then be ABCD using the low byte as little endian would give the string dcba etc.

This would at least give flexibility at the byte level.

Another idea would be to push this responsibility to the client and just allow for raw data reading.

[h0bb3]

advice wanted from @davmoz

[davmoz]

According to the modbus protocol, registers are always in big-endian format (see 1 and 2). So perhaps, we can assume that all inverters that we will be working with will adhere to the standard.

From experience, we know that the following is true:	Inverter model	Register (Word) endianness
Sungrow	Big-endian
Solaredge	Big-endian
Huawei	Big-endian
Growatt	-
Fronius	-
Ferroamp	-

"-" = We don't know

So i suggest we continue assuming that each register is transferred as big-endian. Do you agree?

That said, we should still have some interpretation on the Gateway for logging/diagnostics, so L28 probably need to be tweaked.

[h0bb3]

If big endian is what is in the spec then we go with that. The current implementation is flexible in regards to the higher level interpretation of the data, at least with big an little endian, though there may be variants there also i.e mid big endian, mid little endian for e.g. 32 bit values. But these can of course be added if needed.

I guess it is this part that should be big endian then? https://github.com/srcfl/srcful-gateway/blob/3cd427cb6d35fa7eba61ba5926ff6b5321148369/server/inverters/registerValue.py#L75

[davmoz]

I guess it is this part that should be big endian then?

Yup! :)

Fixed in b0fce8f9479d653579e23c96862e1d3e62deea48