EM24 issues with Opcodes and negative power values

[korpa]

I tried to use mbmd to read data from an RS485 EM24. Sadly a few values where totally wrong. After digging into the documentation I was able to fix the Opcodes like this:

@@ -27,16 +27,16 @@ func NewCarloGavazziProducer() Producer {
                PowerL2:   0x14,
                PowerL3:   0x16,
                Power:     0x28,
-               CosphiL1:  0x2E,
-               CosphiL2:  0x2F,
-               CosphiL3:  0x30,
-               Cosphi:    0x31,
-               Frequency: 0x33,
-               Import:    0x34,
-               ImportL1:  0x40,
-               ImportL2:  0x42,
-               ImportL3:  0x44,
-               Export:    0x4E,
+               CosphiL1:  0x32,
+               CosphiL2:  0x33,
+               CosphiL3:  0x34,
+               Cosphi:    0x35,
+               Frequency: 0x37,
+               Import:    0x42,
+               ImportL1:  0x46,
+               ImportL2:  0x48,
+               ImportL3:  0x4A,
+               Export:    0x5C,

Now most values written to MQTT seem to be ok.

But power values never get negative, even if power flow is reversed through high solar power in the system. I could not verify if this issue is a setting issue within the EM24 or if the number encoding of mbmd (Int32LswFirst) never returns negative values.

[andig]

Looking at https://www.aggsoft.com/serial-data-logger/tutorials/modbus-data-logging/carlo-gavazzi-em24.htm it seems that the registers are currently wrong. Funny enough, they seemed to be right before https://github.com/volkszaehler/mbmd/pull/260. I can no longer reproduce why that change was made. I think we could just go back to the original register map which was copied 1:1 from the docs.

Would that work for you? Do you want to open a PR?

[korpa]

Why do you look at this aggsoft URL and not into the official documentation. I provides much more information than the aggsoft URL (e.g. the factors):

Regarding the Opcodes, I can provide the fix via PR.

But the issue regarding encoding of the power values. My knowledge about bit operations is only very basic. Therefore my question: Do you think switching back from RTUInt32ToFloat64Swapped to RTUInt32ToFloat64 will solve the issue, that power never gets negative?

[korpa]

By the way. The original documentation could be found here: https://gavazzi.se/app/uploads/2020/11/em24_is_cp.pdf

[andig]

Why do you look at this aggsoft URL and not into the official documentation.

Probably because I didn't find it at time of writing. Lets change the link the in file too to point to vendor docs.

Do you think switching back from RTUInt32ToFloat64Swapped to RTUInt32ToFloat64 will solve the issue, that power never gets negative?

I don't think so but you could just give it a try. If wrong you'll notice immediately looking at the 2-word decodings.

[korpa]

I tested RTUInt32ToFloat64. This returns totally wrong readings. So we need the swapped version.

As I said, I'm not very familiar with bit operations. I have no idea how to change Int32LswFirst that it supports negative values as well.

[andig]

As I said, I'm not very familiar with bit operations. I have no idea how to change Int32LswFirst that it supports negative values as well.

The meter is probably not sending them. Enable logging to see wire format.

[korpa]

I had to change the "application type" in the meter config from "b" to "h". Now it returns negative values.

=> Will send an pull request

[korpa]

Fixed in #298