Electronics > Power/Renewable Energy/EV's

What type of control method does this PFC is using ?


Image shows the control scheme used by Gan system for the control of their Bridgeless totem pole PFC reference design. In the paper, they states that they use average current control. But when I google it, I get results of average current mode control having an outer voltage loop that creates the current reference not the power reference. So is there another name for it ? and what's difference between these two methods ?

Seems to me that that is pretty much what you have there, it just has a scale factor applied for RMS supply voltage to change the gain depending on the supply voltage.

It is clearly average current mode control in the inner loop, the outer loop is clearly voltage and the 1/Vrms^2 term which seems to be the source of confusion is essentially just changing the loop gain to suit different input voltages and hence different operating points. note that all that really does is change the gain of the voltage PI controller, so it might be that this allows a faster voltage loop or something.

There are an unending number of minor tweaks to these things, and feedforward (Which is essentially what this is) is a fairly popular one.   

Right, the inner loop ends at the i_ref signal.  You could drive that signal from a whatever, and get a controlled proportional input current to your heart's desire.  In this case, they drive it from a separate outer loop, which is used to regulate voltage over long time scales -- note the outer PI unit must respond slow enough so that it doesn't try to regulate over mains-frequency ripple -- indeed, the output ripple is a critical aspect of a PFC, as the variation in C_L voltage reflects the variation in input power that's necessary for the PF ~ 1 condition from a single-phase source.  Because the right PI output is varying so slowly, we can vary it proportional to input -- by multiplying by |Vac| -- to get a current proportional to input, thus satisfying the PFC condition.

The rms multiplier isn't critical, but without it, the loop gain is all over the place (for one, varying with |Vac| of course) -- it serves to keep things stable and responsive as input voltage varies (mains voltage tends to dip and swell over time, as loads go on and off throughout the day).  The critical insight here is, the right PI's output is just whatever current it wants to keep its capacitor topped up, but the input current must be proportional to instantaneous voltage, and inversely proportional to rms -- because it must draw more current at lower Vin, but the output voltage loop doesn't know or care about input voltage.  Put another way, the output voltage is fairly stable (typically 400V), so power is proportional to current; but input current, at constant power, is inversely proportional to voltage, so we need to divide to get the right current.  And we divide twice (hence the square) because |Vac| is in there too.

So yet another way to look at it, is dimensionally: the right PI output could be seen in terms of output current (or rather, a voltage representing that current), but since its voltage is constant, it's also proportional to output power.  We multiply that power by |Vac| to get the waveform needed, and then divide by Vrms^2 to get current again -- the required input current.  (Again, expressed as a voltage.  Which, note also that the current loop is a transconductance amplifier: as an output, it controls mains current flow; its input is V(i_ref), so it has gain in units of A/V = S, [trans]conductance.)


Great answers from above, and also, if you run the attached LTspice sim, you can see the "current loop" of the PFC acting on its own........since in this case, the load is a battery and is automatically constant voltage, so no need for the outer voltage loop......so this one has a sinusoidal reference voltage,  (in phase with the mains) and simply scales that , then makes the input current follow the sine ref...and voila, you get a sinusoidal input current in phase with the mains voltage..............so this is kind of like the "inner bit" of your shown digram......


[0] Message Index

There was an error while thanking
Go to full version