UCD3138PFCEVM-026: Power management forum

It's not really within the scope of the E2E to do that kind of research for you.  Assuming you can get the C source of the library function, it's probably possible to implement something like it on the UCD.  Even if there is only assembly language source, you could probably also implement that.  It might require more processor resources than are reasonably available on the UCD, however.  It's a complicated issue.

The other issue is that our PFC code works fine without requiring that function.  I know that we have experimented with using the AC frequency in our algorithms over the years, and although attractive in principle, it's never worked out in practice.  Do you have a specific reason to need the frequency?  

We have a variable which measures the number of interrupts between zero crossings.   If you average that, I suspect that you could get a pretty accurate idea of what the AC frequency is with much less consumption of processor and programming resources.  

xiaoxiao, most AC signals are very stable, because the AC network needs to be synchronized.  However, the AC is often noisy because of the effect of noisy loads.  The noise is one of the reasons we don't used phase locked loops, because it can confuse them.

Also note that because of the way PFC works, there will always be an AC component on the bulk capacitor.  Since we draw current proportional to the input voltage, the capacitor voltage will always drop when the voltage is low during the cycle, and rise when the voltage is high.  

The variable which gives the frequency is the iv.half_cycle_counter_filtered.  This gives the time between zero crossings measured in 20 usec intervals and shifted left by 6.  So it's 20 usec/64, or .3135 usec for each count.  this is for a half cycle.