Expose external energy meters through mqtt

[jan--f]

Hi,
I have a Pulsar with a Carlo Gavazzi EM340 attached via modbus for load balancing. I want that data in HA and going through the wallbox seems the path of least resistance.
I think this could be useful generally for other people too. Folks who's HA already has an energy meter integrated can simply ignore these new sensor fields.
The redis instance already has the relevant data under the m2w key. Here is an excerpt from my box:

137) "Carlo3P_addr1.energy_wh.value"
138) "7.7789e+06"
139) "Carlo3P_addr1.line1.current.valid"
140) "1"
141) "Carlo3P_addr1.line2.current.valid"
142) "1"
143) "Carlo3P_addr1.line3.current.valid"
144) "1"
145) "Carlo3P_addr1.line1.current.value"
146) "5.372"
147) "Carlo3P_addr1.line2.current.value"
148) "5.148"
149) "Carlo3P_addr1.line3.current.value"
150) "5.265"
151) "Carlo3P_addr1.line1.voltage.value"
152) "231.1"
153) "Carlo3P_addr1.line2.voltage.value"
154) "231.8"
155) "Carlo3P_addr1.line3.voltage.value"
156) "231.1"
157) "Carlo3P_addr1.line1.power.value"
158) "1231.7"
159) "Carlo3P_addr1.line2.power.value"
160) "1190"
161) "Carlo3P_addr1.line3.power.value"
162) "1210.6"
163) "PBO.energy_wh.value"
164) "7.7789e+06"
165) "PBO.energy_wh.valid"
166) "1"
167) "PBO.line1.current.valid"
168) "1"
169) "PBO.line2.current.valid"
170) "1"
171) "PBO.line3.current.valid"
172) "1"
173) "PBO.line1.current.value"
174) "5.372"
175) "PBO.line2.current.value"
176) "5.144"
177) "PBO.line3.current.value"
178) "5.265"
179) "PBO.line1.voltage.value"
180) "231.1"
181) "PBO.line2.voltage.value"
182) "231.7"
183) "PBO.line3.voltage.value"
184) "231"
185) "PBO.line1.power.value"
186) "1231.9"
187) "PBO.line2.power.value"
188) "1188.7"
189) "PBO.line3.power.value"
190) "1210.3"

I'd be happy to contribute the code. However I'm not sure which field to use. I imagine the PBO and Carlo3P_addr1 prefixes are the same device (I have only the one meter installed in my setup) and was wondering if I can consider PBO as the generic "external meter" key under which other meter models would show up to.
Maybe other know more and can hint to more relevant info?

I think the wallbox can also read two meters in a PV setup. Not sure though what that would look like but I assume that would be interesting data too.

[tronikos]

See https://community.home-assistant.io/t/wallbox-pulsar-plus-integration/200339/751
That just forwards the internal mqtt topic so it's faster and I assume less resource intensive than having the script in this repo poll redis and publish to mqtt.

You might also find https://community.home-assistant.io/t/energy-disaggregation-or-non-intrusive-load-monitoring-nilm-blueprint/619672 useful.

[jan--f]

fwiw here's a snapshot from my box with power meter.

[jan--f]

Gave #50 a quick test

Albeit this being poor measurement, at least in this sample of 1 the PR has no detectable impact.

[jagheterfredrik]

I did notice high memory consumption and CPU spikes, probably from gc. I hope they’re fixed with #51

[tronikos]

@jagheterfredrik since you don't have a power meter, I'm curious what does your CPU usage look like across all processes? If it's much less than us, it might be worth figuring out a way to stop writing all these power meter measurements into mysql without breaking power management functionality.

[jagheterfredrik]

It's not too different I suppose, the point is that the bridge is consuming next to nothing compared to m2w and mysql.

[jagheterfredrik]

Added in #50 thanks to @martonbtoth

[tronikos]

I just tried it. There is no noticeable resource impact. But there is a noticeable delay of up to 5 seconds. See

Not sure why it's more than the polling interval of 1 sec. I guess the values on redis aren't updated as often as the ones in mqtt. I'll personally continue forwarding the internal mqtt topic via https://community.home-assistant.io/t/wallbox-pulsar-plus-integration/200339/751 since it's faster and likely more resource efficient.

[martonbtoth]

MQTT publishing is rate limited on the application side, the values in redis are updated much more frequently. I guess its the same frequency as new values are published to the internal mqtt broker.

[tronikos]

If you see the screenshot, the first increase of 1A was delayed by 5 seconds. A few other changes were also delayed by that much. I don't think it's due to the rate limiter. The rate limiter is configured at 0.2 threshold.

[jan--f]

I have been running this change since last night. I can't see any delays.

Every now and again the two values are slightly different though. Not too much to worry about I think but its curious nonetheless.

[tronikos]

Can you compare them in home assistant history http://homeassistant.local:8123/history over a 5 minutes time window?

[jan--f]

Here is roughly the same time frame in home assistant.