support phase currents, and three-phase with enphase envoy

[pdeliot]

Added support phase currents, and three-phase with Enphase envoy #16744

[andig]

Not sure we can do this. This will return empty measurements on devices not supporting this?

[pdeliot]

Yes, this was discussed with ivoks on the issue discussion.
Originally I was returning 0 when phase is not available.
ivoks suggested returning empty and tested it on its one phase system successfully.

This is the choosen way to implement 3 or 1 phase support.

[andig]

We cannot simply return „nothing“.

[pdeliot]

It is "empty" because the phase 2 and 3 may not exist :-)
And it is working well.

But if you prefer having "0" and this is not creating issue for systems with only 1 phase.. we can do that.

[andig]

We just cannot have this interface for 1p meters

[pdeliot]

I've pushed version return 0 when value is not available instead on "empty".

[andig]

@premultiply was meinst du? Gute Idee?

[pdeliot]

What do you prefer?
-Return "empty"
-Return "0"
-Another way to deal with 1p and 3p

[andig]

Imho we should not have 3p measurements on 1p devices at all, so 3rd option. Would any of the measurements (I guess current?) be missing if this was a 1p device?

[pdeliot]

With 1p, as far as I know "powers", "currents" and "voltages" will not be supported.
But I have no docs about evcc requirement (if you have a pointer on the specifications, feel free to provide :-) )

I'm pushing a new version, with an optional parameter "phases".
Default is "1" then no behavior change for existing installations.

And you can set "3" to gain 3 phases support and linked capabilities.

[ivoks]

I'm worried that this would silently break existing 3p setups.

IMHO templates should collect as much info as possible and core engine should be able to cope with the collected data. If the template returns 'value none none' it should be obvious that it's a single phase system. If it returns 'value 0 0', that's a 3p system. I believe that using 0 in this patch is a wrong approach as you just can't know if the 0 is because there is no current or because there's no phase. Who knows what the future brings; evcc might eventually want to know if it's a 3p or 1p setup.

[ivoks]

ivoks suggested returning empty and tested it on its one phase system successfully.

I've tested it on a 3p system without CTs ;)

[andig]

To decide @premultiply:

• remove powers (only needed if currents not signed)
• remove voltages (only needed for chargers)
• allow phase measurements for 1p devices (and remove config option)

See evcc-io/docs#650

[ivoks]

@pdeliot to work with both single and three phase system, you should really check the length of the array. For example, for phase in production:

curl http://${envoy_ip}/production.json?details=1 | jq 'if (( .production[] | select(.measurementType == "production").activeCount >= 1 ) and ( .production[] | select(.measurementType == "production").lines | length >= 1 )) then .production[] | select(.measurementType == "production").lines[0].rmsCurrent else 0 end'

but then for the second and third line (change the line 1 to 2):

curl http://${envoy_ip}/production.json?details=1 | jq 'if (( .production[] | select(.measurementType == "production").activeCount >= 1 ) and ( .production[] | select(.measurementType == "production").lines | length >= 3 )) then .production[] | select(.measurementType == "production").lines[1].rmsCurrent else 0 end'

And then similar for consumption. What do you think? I think this would cover all requirements; it would work only on CT enabled systems and it would work with both 1 and 3 phase systems without any parameters.