This discussion has been locked.

You can no longer post new replies to this discussion. If you have a question you can start a new discussion

FIR filterin in Fixed Format

Hi,
In My application I am sampling 3 phase current at sampling frequency of 800Hz, currently I am
using the below code and could get a steady rms without any problem.. I am using a bipolar ADC
with 2's complement output format with 12 bit resolution including the sign bit

if((rCurrent&0x0800)==0x0800)
                        {
                                ut =((~rCurrent)+1)&0x00000FFF;
                                redInputCurrent[sampleCount]=(((float)ut)/2048)*-2.5;
                        }
                        else
                         {
                          ut = rCurrent&0x00000FFF;
                                redInputCurrent[sampleCount]=((float)ut/2048)*2.5;
                         }
                        if((yCurrent&0x0800)==0x0800)
                        {
                                ut =((~yCurrent)+1)&0x00000FFF;
                                yellowInputCurrent[sampleCount]=(((float)ut)/2048)*-2.5;

                        }
                        else
                        {
                             ut = yCurrent&0x00000FFF;
                             yellowInputCurrent[sampleCount]=((float)ut/2048)*2.5;
                        }
                         if((bCurrent&0x0800)==0x0800)
                        {
                           ut =((~bCurrent)+1)&0x00000FFF;
                           blueInputCurrent[sampleCount]=(((float)ut)/2048)*-2.5;
                        }
                        else
                        {
                          ut = bCurrent&0x00000FFF;
                          blueInputCurrent[sampleCount]=((float)ut/2048)*2.5;
                        }
                        if((eCurrent&0x0800)==0x0800)
                        {
                          ut =((~eCurrent)+1)&0x00000FFF;
                          earthInputCurrent[sampleCount]=(((float)ut)/2048)*-2.5;
                        }
                        else
                        {
                         ut = eCurrent&0x00000FFF;
                         earthInputCurrent[sampleCount]=((float)ut/2048)*2.5;
                        }



                                        sampleCount++;
                                        if(sampleCount>15)
                                        sampleCount=0;



                        arm_rms_f32(&redInputCurrent[0], 16, &rCurrentValue);
                        arm_rms_f32(&yellowInputCurrent[0], 16, &yCurrentValue);
                        arm_rms_f32(&blueInputCurrent[0], 16, &bCurrentValue);
                        arm_rms_f32(&earthInputCurrent[0], 16, &eCurrentValue);


                        CurrentValue[RED_CURRENT_CHANNEL] = rCurrentValue * R_CURRENT_RATIO;
                        red_i_index = (int)CurrentValue[RED_CURRENT_CHANNEL]/5;
                        CurrentValue[RED_CURRENT_CHANNEL] = CurrentValue[RED_CURRENT_CHANNEL] + red_offset[red_i_index];

                        CurrentValue[YELLOW_CURRENT_CHANNEL] = yCurrentValue*Y_CURRENT_RATIO;
                        yellow_i_index = (int)CurrentValue[YELLOW_CURRENT_CHANNEL]/5;
                        CurrentValue[YELLOW_CURRENT_CHANNEL] += yellow_offset[yellow_i_index];

                        CurrentValue[BLUE_CURRENT_CHANNEL] = bCurrentValue*B_CURRENT_RATIO;
                        blue_i_index = (int)CurrentValue[BLUE_CURRENT_CHANNEL]/5;
                        CurrentValue[BLUE_CURRENT_CHANNEL] += blue_offset[blue_i_index];

                        CurrentValue[EARTH_CURRENT_CHANNEL] = eCurrentValue*E_CURRENT_RATIO;
                        earth_i_index = (int)CurrentValue[EARTH_CURRENT_CHANNEL]/5;
                        CurrentValue[EARTH_CURRENT_CHANNEL] += earth_offset[earth_i_index];

but the moment I try to convert these samples from float to q_15 format for filtering purposes
the range reduces by half.. i.e. without conversion and filtering as above, I could
measure current from 180mA to 43A with 1:1000 ratio CT.. but when I introduce FIR filtering the
reading range reduces from 800mA to 19A... I have tried bypassing the filtering after the q15
conversion and calculating the rms of the signal and the problem still persists.. Can anybody
help on this regard.