How can I calbrate my power meter?

Question

I used to work for a company that made power meters but only worked on the desktop application (user interface) side of things and never really learned completely how some processes occur, like the calibration process. So in order to learn something that I've always been curious about I decided to make my own power meter

What I know and what I think I know

I know some regulations that define the number of points per wave (for example, @60hz must have 64 samples) given a desired precision. I know the equation for Vrms, Irms, apparent power, active power, power factor, real power and these things I can measure in the hardware I made (one TP and three TCs, I plan to do a software phase shift for the other two voltages, which I would guess is not good enough for any standard, but OK, or else I can add a phase delay with capacitors at the input of the ADC ... also TCs are really expensive).

The thing is that I know that the obtained values are calculated using a calibration constant, so ... Vrms = Kc * sqrt(1/N*(v1^2+v2^2+..+vn^2)) and the same for current, so the other calculations that derive from this are subject to the calibration constants. From what I've been reading this calibrations constant is Kc = (real high voltage - read low voltage)/(read high voltage - real low voltage) and there is also an offset, which is calculated in a similar manner., so the above Vrms =Kc * sqrt(1/N*(v1^2+v2^2+..+vn^2)) + Ko, where Ko = offset????

The problem I know I'm facing

The problem that I'm facing is how to implement the calibration process. According to a national standard O know that a device must be calibrated with another device (calibrator) that is at least 3 times more precise. So if I want to achieve 1% precision I'd need to calibrate it against a device that is at least 0.3%. I know that I'd need to calibrate it with PF=0, PF=1, PF=0.5 AT LEAST but I'm still unsure as to how I would calibrate it. My former company used a calibrator and then they had to write a program to send commands to it (change from delta to star configuration por example, change voltage, current, PF etc) and then they'd read both devices compare and generate a calibration constant. From what I remember it's not a very fast procedure, requires A LOT of math and I'm unsure as to how I'd learn about the necessary math.

What I've been told

Some people have told me that a calibration process is a well kept secret by companies, but that doesn't really add up, there must be a theory behind it and a base method that can be used and learned from books/internet and then after I gain the knowledge I'd be able to tweak the base method to achieve my goals.

What I want to know

Does anyone know about any good books/websites/articles about the calibration topic? I've been trying to research this but not having much success. And what about international standards for calibration? I know about ISO/IEC 17025:2005, but I think that only applies for laboratories that want to be certified and I just want to learn about how a real calibration process can be done for my meter. Or if you prefer you can answer me in detail in the answer, but I think that'd be a lot to task

PS: I have a notion of what I need to know, but there are certainly more things that I'd have to really know before going forwards. I'm an EE with a few years of desktop applications experience and then I quit that job and I've been working as a firmware developer (more my area of knowledge)

Answer 1

THere are 2 or 3 power phasors computed for each Line voltage and current: 2 lines and 1Delta , Assuming you are not in North America the line phasors angles are at 2phi Vac/120 deg and in North America split 240Vac/180 deg.

• Phasor power angles are straight forward

• PF = |cos φ| = 1000 × P(kW) / (√3 × VL-L(V) × I(A))

• Each line may be calibrated for resistive and reactive loads or some matched combo that is -45 deg then compute if any phase error.

• P(W) = |S(VA)| × PF = |S(VA)| × |cos φ|

• kW = √3 × PF × A × V / 1000 * k (scaling factors)
• you already have the RMS sample conversion formula
• As far as the ADC is concerned a stable Vref is used for conversion and it may have errors for ; offset and gain and missing codes (monotonicity error) due to digital noise crosstalk on the analog Vref or signal like 00011111 to 00100001 skipping a step
• good design self-checks calibration using a precision slow ramped test signal to test every code, linearity, gain and offset error. with correction factors ( sometimes done yet often ignored )

I used to work for a company that designed wireless 2 way meters for power, gas, water using the ISM band with multi tiered data repeaters at 928MHz in the mid 90's. 1 second response time but scheduled Mux's otherwise. Headend software integrated into Utilities database. But technology got sold out to Itron in Seattle. a prime supplier of US power meters after 7 yrs. I was there for the last 4. Iris Systems Ltd. Winnipeg.

Traceability to NIST standards and Utility Specifications are two topics for further discussion.

Cheers Tony

Answer 2

Use high standards even if the application is for general consumption (Makes it a better product or tool), I always use highest standards available. So go ahead and use what ever.