This is part three of six of the power factor correction circuit basic seminar. In this section, we'll be discussing some performance considerations with respect to the critical conduction mode boost converter. OK, now let's take a step back and look at overall how the switching frequency is affected by these different switching schemes.

Here's the zero valley. And one of the advantages of this is just really ease of implementation. But you can see at the zero crossings, we definitely stop that inductor frequency from going so high. So we've clamped it in this case just about at 400 kilohertz.

By going to the first valley, we get an additional advantage in the current, or the frequency gets stopped from going quite as high. Now the second valley does it even more. We reduce the switching frequency even more. And the advantage of the second valley is now we can go to a third valley or a fourth valley, if for some reason we need to limit that switching frequency further.

Next, let's take a look at how these different switching mechanisms, these different decision points for turning on affect the overall distortion. After all, that's what we're adding the power factor correcter in there for in the first place. The first plot shows the zero valley, and we get a pretty good power factor but pretty significant distortion.

You can see what the input voltage looks like here in blue and the corresponding line current and the distortion associated with that. The individual harmonics are listed below. One of the things you'll notice is that there's this large flat area on the input current. This is a direct result of the fact that, at some point in time, there's just not enough energy in the inductor to resonate up to deliver current through the output diode.

When that happens, the average current through the inductor essentially becomes zero. And we get this dead zone. And that dead zone is one of the biggest things that contributes to the distortion.

The first valley, similarly, has a dead zone. But it's not quite as big. And the reason why it's not quite as big is if you remember, we talked about the fact that zero valley starts its on time when the current is negative. First valley starts when the current is zero. The result is we get a little bit more current in the inductor that helps us correct for that error. And then the second valley, you can see the results that are shown there, pretty similar to what you get with first valley. But some of the additional distortion comes about from the fact that we've introduced this discontinuous conduction time.

Some other things that affect the distortion are the fact that the body diode of the MOSFET actually burns some power. This is going to introduce some non-ideal behavior into the system as well. In addition, there are other sources of switching loss that may create some minor distortion as well.

OK, now let's talk about, how do we compensate the critical conduction mode PFC? I've got a high level schematic here shown in the upper left. And basically, what this thing does is we've got a voltage loop and a voltage loop only.

And remember, if you remember, if we hold the on time to the MOSFET constant, we're able to basically generate good power factor. So if we create a voltage loop such that VC varies very little across the whole line voltage, we can then generate a constant on time. And that constant on time will give us a very good power factor.

One of the things that can be very helpful for these systems though is if we have input voltage feed forward. And the idea here conceptually is pretty simple. We want to make the output of this op amp instead of directly proportional to on time, we want to make it proportional to the power demand.

And then we introduce another block between here, between this VC and this VC connection where that power demand gets translated into the constant on time command. And then that translation involves the input voltage. So those feed forward terms can make it much easier to derive one set of compensation values that works over a very wide input voltage range.

Another thing to keep in mind is that power factor correctors often supply power to constant power loads. This is really the basic nature of a downstream DC to DC converter that takes the output of the PFC and then converts it to a constant voltage to deliver power to its load. In order for it to do that efficiently, its input essentially looks like constant.

So what is the overall strategy for compensating a PFC? Well, like any compensator, there's a series of tradeoffs that you make. And that's no different for power factor correctors.

So first of all, good power factor requires a slow control loop. That comes about from the fact that we don't want to correct the output voltage for disturbances that are due to the line voltage variation. These are things that occur at-- because we're rectifying the input, these are things that occur at twice the line frequency. They're 100 Hertz or 120 Hertz. So we want to make sure that our control loop is much slower than that, so that we can't influence those line harmonics. Because when we do, we'll just corrupt them and make our harmonic distortion worse.

Now on the counter side, we want good transient response. Because if there's any kind of a load transient applied to our output, we want to correct for that quickly. So one way to solve this is to have some kind of nonlinear gain that's added to the amplifier, so that the bandwidth of the loop drastically increases when the output deviation is large. But for small deviations, in other words, deviations that are just due to normal ripple voltage on the output, we don't respond to those. And we respond with a very low bandwidth.

OK, now let's take a look at the Bode plot shown on the right. Basically, what we're looking at here are the compensators shown in red, the overall loop shown in blue, and then the gain and the plant, which in this case is Vbulk or Vout over VC. You can see that the plant has a very integrator type look to it. And that comes as a result of the fact that, one, we're driving a constant power load and that the nature of the PFC is a constant power source. Those things clash a little bit and give you this integrator like effect that you need to compensate around.

Now we need another integrator in the system in order to get good DC regulation. And you can see that in my compensator here at very low frequencies. And then we introduce a zero here around the crossover frequency in order to achieve good phase margin. And then we want to roll off the gain at higher frequencies, so that we don't introduce any additional distortion in our control loop. Overall, you can see that we've got a band width of around two Hertz in this case, apple phase margin, and ample gain margin. So this is a very well-compensated loop in terms of stability and overall power factor quality.

OK, now let's put this whole thing together. Let's look at our overall boost converter here. We've got our input bridge. We've got our controller. We've got our feedback. And we've got our boost converter.

There's a new element up here that we haven't talked about yet. And that's this auxiliary winding that's placed on the boost converter. This winding is there to help us detect when the zero crossings are. Basically, it gives us a signal that allows us to determine when to turn the switch on in order to achieve critical conduction mode PFC.

Down here at the bottom, we've expanded what's inside that PFC controller block. Starting from the left, we've got our sensed input voltage, which corresponds to this red waveform, and that gets filtered. Oftentimes, it's just averaged. But really, the goal is to create the RMS voltage.

And then down here, we've got our air amplifier, essentially the same air amplifier that I showed you on the previous slide. And that goes into the B input. Now the output of this air amplifier is the power demand. It's commanding how much power needs to get delivered to the output.

B divided by A squared basically takes our power demand, divides it by the input voltage to create a voltage that's proportional to on time. So that's our on time command. And then that goes through our on time generator. And then once that triggers a pulse, that shuts off the MOSFET.

Down here in the bottom, we've got the circuit that determines when the MOSFET turns on. We've got our aux winding sense. And when the voltage on our aux winding goes above our programmed reference threshold, it sets the flip-flop and turns on the pulse and initiates the next cycle.

So now just a few comments to wrap up the critical conduction mode PFC. Basically, one of the biggest reasons this solution is so popular is due to its simplicity and its overall low cost. The solution, in general, is applicable to lower power outputs, probably sub 300 watts. And this is because you've always got to have large ripple current in it.

The large ripple current is basically it's always going to be 2x the average. And that's going to result in larger RMS currents. It can result in larger losses in the system.

If you remember, we also get valley switching with this topology. This gives us some efficiency advantage. We also get the advantage of zero current switching in the boost diode, which allows us to use lower cost diodes for that.

So all in all, you get a pretty good power factor, low cost, OK THD. There are some techniques out there for improving it. But this is definitely not going to be your very top of the line THD solution.