TMS320F280049C: eQEP example question / C2000WARE 3.04

Hi,

I just wonder about algorithm used in the example code - below you can find a part which calculates the position difference for speed calculation.

Let's walk through the calculation:

If the QPOSLAT = 1000 and the mech_scaler = 4000, we get 4e6 in the tmp. Then it is masked and shifted to an integer range, so we get 1952. A mask of 0x7FFF does not make any difference.

The tricky part is newp=_IQ15toIQ(tmp); which should convert IQ15 to IQ (float). But the conversion doesn't go per-unit and we get  newp = 1952.0F.

Let's assume that we've got the oldp = 1852.0F. If we do a substraction of Tmpl = newp - oldp, it would give us Tmpl = 100.0F, which (based on the peroid of the calculation) results in the rotor speed.

But the issue here is the Speed_fr saturation done as if (Tmp1>_IQ(1)) etc. It always gets saturated and I'm always getting per-unit speed of +1.0 or -1.0. What am I missing?

Thanks

pos16bval=(unsigned int)EQep2Regs.QPOSLAT;      // Latched POSCNT value
tmp = (long)((long)pos16bval*(long)p->mech_scaler); // Q0*Q26 = Q26
tmp &= 0x03FFF000;
tmp = (int)(tmp>>11);           // Q26 -> Q15
tmp &= 0x7FFF;
newp=_IQ15toIQ(tmp);
oldp=p->oldpos;

//..
Tmp1 = newp -oldp;
//..

if (Tmp1>_IQ(1))
{
    p->Speed_fr = _IQ(1);
}
else if (Tmp1<_IQ(-1))
{
    p->Speed_fr = _IQ(-1);
}
else
{
    p->Speed_fr = Tmp1;
}