Power factor calculation using microcontroller

[heavenlyrider]

hi,

I am using voltage divider network to measure voltage, and ACS712 (hall effect sensor) to measure current. I am using MCP3204 for ADC conversion.

Here is my code.

[Mike Walne]

We don't have your ADC code, which does not help.

Several approaches can be tried.

1)
You could record a series of current and voltage readings.
Then work out RMS values, real & apparent power and PF by hand.
Could be revealing.

2)
Seriously simplify, get rid of all the correction factors and fudges.
RMS values, real & apparent power will all be wrong but PF should be OK.

3)
Use a simpler load.
Say an X rated capacitor.
You SHOULD then get a PF of zero.

On UK mains 1uf gives RMS current of ~70mA.

Just a few things to try.

Mike

[Ttelmah]

Look through your samples to find the zero crossing point of the current. (interpolate between the samples each side of this point). Do the same for the voltage. Given you know your sampling rate, you can then calculate the time between these, and calculate the phase angle. Cosine of this is the PF.
Any other approach goes wrong if the waveshape goes non sinusoidal.....

[Mike Walne]

I gave up trying to decipher your schematic.
(You appear to be using drawing methods I abandoned half a century ago.)

You seem to be assuming that voltage & current readings of 2048 coincide with zero crossings.
That's possibly not the case and unneccessary.
You are creating an array for all your current and voltages but don't use them in any real way.
I'm guessing you are working with 50Hz mains.

There is a neater way to do all your calculations which:-

1) Uses less maths, and only unsigned integers except for the final stage.
2) Does not require the use of an array.
3) Does not require you to guess at (or make assumptions about) the zero crossing point.
4) Works for both sinusoidal and non-sinusoidal waveforms.

You simply process the raw current and voltage readings as follows.

Start by resetting all the running totals to zero then do the following:-

1) Calculate Square of Voltage (V2).
2) Calculate Square of Current (I2).
3) Calculate product of Current&Voltage (IV).

Keep running totals of:-
4) Sum_of_Voltages. Sum_V
5) Sum_of_Voltages_Squared. Sum_V2.
6) Sum_of_Currents. Sum_I
7) Sum_of_Currents_Squared. Sum_I2.
8) Sum_of_Current*Voltage, Sum_IV

Assuming you have n samples you then calculate:-

9) Voltage_Mean as (Sum_V)/n. (This is the zero crossing level for the AC Voltage)
10) Voltage_RMS as SQRT(( Sum_V2 )/n - Voltage_Mean^2 )
11) Similar for Current_Mean and Current_RMS.
12) Power_Real as ((Sum_IV)/n - (Voltage_Mean)*(Current_Mean)).
13) Power_Apparent and PF are then straightforward.
14) Convert to engineering units.

At this stage you have an arrays of all your values.
Take a complete set of values and process by hand by both your method and mine.
(Or transfer to Excel and let the PC do the work)
Both methods SHOULD yield the same results.

On the other hand you could try with a small set of simple numbers and use pencil and paper.
You could take samples for multiples of 100ms (5 cycles of 50Hz, 6 cycles of 60Hz).
That way you could work on either 50/60 mains.

Mike

[temtronic]

OK, I had a look at the schematic and it makes NO sense to me and like Mike I have several decades of design under my belt.
The glaring problem I have is in feeding an AC voltage into the ADC chip(MCP3204) that was only designed for +ve inputs.
The second ADC (for current ) isn't needed either.

What is missing is a zero cross detector. If that was wired in, it's dead simple to aquire the voltages and currents and do the 'math'. the 'math' of course would be done in scaled integers,easily in realtime.

Also missing is a TTL<>RS232 convertor for the PC interface. Typically a MAX232 device, the 2nd section of it can be used for the aforementioned 'zero cross detector'. That clever bit of kit was done for a 'Mains frequency meter' 30 years ago, using a PIC16C84.

Jay

[Ttelmah]

Aargh.....

I must admit I hadn't seen that there was a circuit link.
Whimper.

[dyeatman]

Here are several more simple zero crossing detectors.

http://www.bristolwatch.com/ele2/zero_crossing.htm
_________________
Google and Forum Search are some of your best tools!!!!

[Mike Walne]

The key advantage of my method of calculating is that you do not need a zero crossing detector at all.
If you do use one it can be used to synchronise your sampling to the incoming signal.
That way you make measurements over several complete whole cycles.

Mike

No response from O/P so assume this topic is dead.

[heavenlyrider]

Hi Mike,

[guy]

If you need to rectify the sinewave signal for the ADC without losing the Vf of the diodes, use a good Precision Detector circuit.

[Mike Walne]