Hello,

I am working on 3 ph voltmeter.Hardware is like this 89c52,
10 bit adc chip , 4 seven segment displays , 3 keys.
mains voltage is given to isolation amp & then this ac vtg fed to adc pin which is biased to voltage adc ref/2.

In software I am thinking of following schem which I got from avr app note . the code needs lots of modification to suit my project but here is the schem -



//xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

struct
{
signed int Fresh, Previous;
signed long Filtered,PreviousFiltered,Calibrated;
}Sample;

struct
{signed long U2;} Accumulator;

struct
{
float U;
}Result;



__interrupt void ADC_ISR(void)
{
signed int TempI;
signed long TempL;

// Sampled data from ADC: ADCW Range=[0000...03FF] = 1023
// Copy of sampled data: Sample Range=[0000...03FF]
// Scaled copy of data: Temp Range=00000000...0003FB01]
// DC filtered data: Filtered Range=FFFE0280...0001FD80]

Sample.Previous = Sample.Fresh; // x[n+1] <- x[n]
Sample.Fresh = ADC; // x[n] <- DATA

// Apply filter for DC offset removal.

//y[n] =(0.996 * y[n-1]) + (0.996 * x[n]) - (0.996 * x[n-1])
//y[n] = (0.996 * prv fltrd ) + (0.996 * fresh adc) - (0.996 * old adc)

Sample.PreviousFiltered = Sample.Filtered; // y[n] <- y[n-1]
TempL = 255 * (long)Sample.Filtered; TempL = TempL >> 8;
TempI = Sample.Fresh - Sample.Previous;
TempL = TempL + (255 * (long)TempI;)
Sample.Filtered = TempL;

//For voltage measurements: accumulate square of samples
//Filtered data: Filtered Range=[FFFE0280...0001FD80]
//Prescaled data (>>5): TempX Range=[FFFFF014...00000FEC]
//Prescaled data (>>6): TempX Range=[FFFFF80A...000007F6]
//Multiplication result:(n/a) Range=[FFC09F9C...003F6064]

Temp = Sample.Filtered >> 6;
Accumulator = Accumulator + (Temp * Temp);

if (SampleCounter == NMAX)
{
SampleCounter = 0;
Accumulator = 0;
CYCLE_FULL = 1;
}




void main(void)
{

while (1)
{

// Voltage:U = SQRT [ (accumulated data) / (number of samples) ]
// All data processing must be complete before next ccumulation cycle is completed.

if (CYCLE_FULL == 1)
{
Sum = Accumulator;
Accumulator = 0;

TempUL =(Sum + OFFSET)* NORM;
Result = sqrt(TempUL);

Result = Result * CalCoeff;// Calibrate.
printf(", %8.3f",Result); // disp
}

}

//xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx




here are my some doubts

1. I would sample adc inputs at rate 200usec as mains freq
is 50hz.

Adc conversion will require @ 20usec

time for executing following iir filter code @100 usec
[I am not sure it may take even more time]

y[n] =(0.996 * y[n-1]) + (0.996 * x[n]) - (0.996 * x[n-1])

total Isr code will take @ 120usec , so only 80usec left
in which I should do Disp mux and some other tasks.

I am first time using this iir filter, is this schem will
give smooth readings ?

2. is there is any trik to avoid 32 bit calculations in ISR
due to iir filter.

3. is there any alternative method for getting rms reading with
out external hardware.


Regards