edstobi wrote: ↑14 Nov 2022, 23:51
Hi
I am using PZEM-004Tv30-Multiple on ESP_Easy_mega_20221105_energy_ESP32_4M316k_ETH
For Energy_WH I get the reading eg. 9.2 Wh from Espeasy
In my opiniun that output of the espeasy is wrong and shoud be 9.2
kWh
For a load of 9 Watt to use 9 kWh, it must be running 1000 hours.
Still, running for 72 hours would not make it to 0.7 Wh.
To make sure where the error comes from, I suggest to add some purely resistive load for some time to see how that's being measured.
Also 9 Watt is nearly at the lowest possible end the device can measure, so a single count higher or lower on the ADC will probably jump quite a bit in reported values.
So it is best to run some higher load (purely resistive), like a 60Watt light bulb or 230V halogen lamp not using electronics.
After running this for some time (e.g. 10 - 60 minutes), you can add your other load parallel to it and see how much the reported energy consumption increases.
With a resistive load, which is also more significant in energy consumption compared to the other device, you make it easier for the meter to determine a more useful cos-phi.
Are you using a shunt resistor, or a Rogowski coil?
According to the
datasheet the coil version (100A version) starts measuring at 20 mA.
With a load of 9 Watt, the average current should be about 39 mA.
Not sure how its behavior is upto the stated 20 mA. It could be that this is considered to be the noise floor and measurements below this are simply ignored, or it might be possible the entire current measurement has some "DC offset" and thus every measurement should be compensated by some offset.
The stated resolution of 1mA doesn't mention whether it applies for the 10A version or the 100A version.
But given the fact that a 0.001A resolution on a 10A range already needs a max. count of 10'000, I think we can safely conclude the 100A version doesn't have a 0.001A resolution.
In order to achieve a count of +/- 8192 (both positive and negative), you already need a 14 bit ADC.
I really doubt there will be a 16 bit ADC in this sensor and even 14 bit would already be more than I'd expected.
Let's assume it has a 14 bit ADC, then for the 100A version the actual resolution is roughly 10 mA, which may even be worse when there is some calibration being used.
This means at a load of 9 Watt, the current will be below the 20 mA threshold for 1/3rd of the time.
But if the connected device is drawing current like this:
You can be sure the computed current and cos-phi will absolutely not make any sense in the low end of its measurement range.
By adding some pure resistive base load in parallel, like I suggested (using a 60 Watt Tungsten light), you'll lift the measurement more into its usable range.
You could also try measuring multiple of these devices in parallel, to get some idea of the true cos-phi. But this will only work if their current draw is showing exactly the same pattern for each device.
