Closed ianchi closed 2 months ago
I'm using a PG04 (UK Plug), and testing with a 1kVA AC source and a Yokogawa WT310E (power meter) and this seems to work for me, except the voltage is way off (which it also is in the original code).
Hence I was just about to raise a different PR to add a linear filter to the voltage, and the same filter seems to work for me both with and without your changes. Can you try it out on your PG03?
- calibrate_linear:
- 89.4 -> 80.0
- 100.7 -> 90.0
- 111.8 -> 100.0
- 123.0 -> 110.0
- 134.2 -> 120.0
- 145.4 -> 130.0
- 235.4 -> 210.0
- 246.5 -> 220.0
- 258.2 -> 230.0
- 269.0 -> 240.0
- 277.5 -> 250.0
Hi @AlexCPU, have you tried adjusting the voltage_divider
parameter instead of applying a filter?
With your numbers, it seems that you could try with 1550
and check how it affects power (it will decrease in the same proportion).
When you tested against the reference, only voltage was off? Current and power where correct?
In my case your filter would shield too low a voltage.
Thanks, that's improved it, nice update :)
1550
seems to give me better results. My voltage was showing about 269V
with the default config which is WAY off!
Sorry, I'm not very expert But I can no longer update ESPHome, it gives me a space error How can I solve it?
My yaml
`substitutions: name: presa-235-athom-smart-plug-v2 friendly_name: Presa 235 Athon V2 packages: Athom_Technology.Smart_Plug_V2: github://athom-tech/athom-configs/athom-smart-plug-v2.yaml esphome: name: ${name} name_add_mac_suffix: false friendly_name: ${friendly_name} api: encryption: key: Bch2rF8jfi9HtSJc5bolF8xv+Qpcj8nNJ0ZGnIpH7s=
wifi: ssid: !secret wifi_ssid password: !secret wifi_password
# Optional manual IP
manual_ip: static_ip: 192.168.1.235 gateway: 192.168.1.254 subnet: 255.255.255.0
mqtt: broker: !secret mqtt_host port: !secret mqtt_port username: !secret mqtt_username password: !secret mqtt_password id: mqtt_client`
It seems that the actual chip used en PG03 is not HLW8012 but BL0937. I haven't open the plug, but it is apparent from the readings. In that case, using the new
model
option, the multipliers are correctly computed and no linear filter is needed. This way all measures are correct, while the current implementation shows an incorrectenergy
if it is used directly (instead of the integration). I calculated the voltage multiplier empirically. You could provide the actual resistors value.