I worked through some of his sketches this weekend. Interesting stuff with the sine wave shown in Serial Monitor etc.
I think my problem has been forgetting to reset the emonTx after each flash. Starting to look much better. I knew I had to reset with the full sketch but thought the basic sketch was Ok without it.
I don’t understand that.
Generally, nothing is carried over from an old sketch to a new one.
Unless you’re using the EEPROM, then the sketch automatically resets everything when you restart it. The EEPROM doesn’t hold the calibration constants, so those are never carried over from one sketch to another anyway.
Even if the EEPROM is in use, values aren’t actually carried over, the new sketch is programmed to read them from EEPROM as part of its start-up procedure. So yes, they appear to be carried over, but it’s not automatic and not by default.
Scrap that.
The problem was probably trying to use CT2 / CT3 for very low power readings (50W) and moving to CT4 looks to be fine.
Will just try 2000W on CT4 and then go back and check 2000W+ for CT2 and CT3.
Edit: yes all good with 2KW on CT2 to CT4.
I am now moving on to whole house monitoring having previously tested single circuits.
With the fridge / freezer and a couple of laptops the property runs at around 500W.
Obviously it goes up when the kettle, air con and cooker etc are fired up.
My present favourite, emonTx Arduino shield, is showing around 4% high at 500W and 2.5% high at 4000W.
The 4KW is within the range of the higher accuracy CT4 on the emonTx but there isn’t a higher accuracy CT on the shield. Also from time to time it would go well over the 4.5KW allowed.
If I have understood the calibration correctly I could drop the default 60.606, for the shield, down to the bottom end of the anticipated range of 58.18 and it should correct the 4% high for low power usage and perhaps swing the higher power usage to 1 or 2% low.
Edit: dropped it to 58.50 and confirmed what I expected.
How accurate do most users find their shield and emonTx systems to be after a nominal amount of calibration?
If you comb the Archive, I think you’ll find claims of around, or better than, 1%. [Obviously, correlating kWh against the supplier’s meter is the only realistic way to compare.] Whether that applies to “most users”, I cannot say, as most users don’t report this.
You can convert an input to any range you wish by choosing the correct c.t. and, if necessary, changing the burden resistor. The design ‘rule-of-thumb’ is to aim for 1.1 V rms at the ADC input for an emonTx, and 1.6 V rms for the Shield.
1 Like