Fast calibration method


I am using a "fast" calibration method to calibrate my CTs.
I know it works very well (accuracy<0,1%) with the continuous sampling sketch.
I would like to know if it can work too with the discrete sampling sketch.

Here is the method. During a couple of hours. I monitor the main power with the continuous sampling sketch with a CT round the main cable. Then, EmonCMS calculates the elapsed kWh seen by the CT. During the same time, another sketch counts the number of optical pulses emitted by my electric meter. This second value gives me the "real" elapsed kWh. In order to calibrate my CT, I just add a correcting factor in EmonCMS ( x 2nd value / 1st value).

I know it works with the continuous sampling sketch as the calibration of real power is done using a single variable (powerCal). Knowing that I don't care to measure the exact voltage, can I use this method with the discrete sampling sketch ? (my final goal is to get the real power)

I guess it's ok but I would like to be sure.

I hope it's clear ;)


calypso_rae's picture

Re: Fast calibration method

Hi Eric, I'm pleased to hear that my 'Continuous Monitoring' sketch is of use to you.

I'm not sure that I would claim quite such a high degree of accuracy, but my powerCal scheme is very straightforward to use for measuring real power.  This approach has not changed since I posted the first Mk2 PV Router sketch.  The small numerical value is because of the mismatch between Amps and ADC steps for the current sensor.  There are roughly 20 ADC steps per Amp, so the powerCal value needs to be the inverse of this, i.e. roughly 0.05.   The voltage sensor has roughly one ADC step per volt, so has little effect.

When calibrating any new hardware on the bench, I use a conventional supply meter that pulses every Watt Hour.  On the Atmega, I run my DigitalMeterEmulator sketch which can be 'calibrated' using exactly the same powerCal value.  When both systems are pulsing at the same rate, calibration has been achieved.  Finding the optimal powerCal value in this way takes just a few minutes. 

The discrete sketches have separate cal values for Voltage and Current, so it all gets a bit more complicated.  If your real power values are found to be incorrect, should you adjust Vcal or Ical or both?  Who knows, which is why I went for the simpler all-in-one approach.

dBC's picture

Re: Fast calibration method

I can see how how a linear multiplier like that could calibrate away any amplitude errors in V or I, provided they're in phase (i.e. a resistive load), but I'm struggling to see how it could work under other conditions, or how it could correct any phase errors in the measurements.


Actually, I take that back.  I think I was being overly pessimistic there.  Your single multiplier should work fine regardless of whether V and I are in phase.  Imagine a perfect world with no phase errors in your measurements, where you could precisely capture V and I at the exact same instant.  Then say you did decide to calibrate the individual readings so you could display real Volts and real Amps, and you came up with V' as your voltage multiplier and I' as your current multiplier.  To calculate Power, you'd then sum up all the:

V' * voltage-reading * I' * current-reading

Effectively, all you've done is come with a new multiplier, P' that equals V' * I'.   If you've no interest in displaying Volts and Amps, then you really don't care which multiplier was 3 and which one was 2, you just need to multiply the product of the readings by 6, which is effectively what your approach does.

calypso_rae's picture

Re: Fast calibration method

Isn't it better to calibrate the measurement system at source rather than when the data is being processed in emonCMS?

That's what the calibration values are for ...

Robert Wall's picture

Re: Fast calibration method

I will just note that phase calibration cannot be done in emonCMS. I have to agree with Robin that, as a general principle, it is indeed best to calibrate as close to the source of the error as possible. One obvious case is where you have an emonGLCD in addition to data logging with emonCMS. In that case, if you do not calibrate the emonTx, you would need to introduce one or more calibration constants into the GLCD as well as into emonCMS, and manage and maintain both. The only time I would calibrate at a distance from the source is when the source is inaccessible, for whatever reason.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.