Orno1p: wrong opcodes?

[andig]

@ochykysh could you kindly double-check https://github.com/volkszaehler/mbmd/blob/master/meters/rs485/orno1p.go? It seems strange to have 3p values in the opcodes map for a 1p meter? Also seems as if not all opcodes are referenced by the producer?

Stumbled upon the by user report of evcc-io/evcc#4576

Originally posted by @andig in #129 (comment)

[ochykysh]

(copy of comment in the original issue)

@andig The register codes are from the original PDFs

https://b2b.orno.pl/download-resource/26063/
https://b2b.orno.pl/download-resource/26064/

And yes, they do have 3p registers on a 1p meter...no idea why. I decided not to poll for registers that will be empty, so that's why data is not being collected from phase 2 and phase 3.

We could remove the references if you think it would be better.

[andig]

Thank you, seems the confusion is due to the vendor. Main problem is that Power won't work on the 1p meter. Seems we should make Power an alias of PowerL1 here?

[ochykysh]

If we do that, probably makes sense to do it for

VoltageL1 -> Voltage
CurrentL1 -> Current
PowerL1 -> Power
ReactivePowerL1 -> ReactivePower
ApparentPowerL1 -> ApparentPower
CosphiL1 -> Cosphi

Do we have some alias implementation already? Don't want to poll all of those registers two times unless I have to.

[andig]

Do we have some alias implementation already?

We don't and it would probably not help. Problem seems to be that for the "official" registers the L1 measurements use different registers from the total measurements.

As the implementation is targeted at L1 meters my current approach would be:

• comment all registers as "offical definition"
• rename all L1 measurements to totals

Downside: we will no longer show L1 measurements in the UI. OR we duplicate them, but that would generate additional reads.

Seems we need to think about a proper solution a little more.

[ochykysh]

Just a note: if we just rename them, we will break existing installations.
Too bad we can't return the set of values as a whole, that way we could easily add the alias implementation.

[RichieB2B]

Too bad we can't return the set of values as a whole, that way we could easily add the alias implementation.

We and we should, as proposed in #255