WEBVTT 00:00:00.000 --> 00:00:02.000 align:middle line:90% [SQUEAKING] 00:00:02.000 --> 00:00:04.500 align:middle line:90% [RUSTLING] 00:00:04.500 --> 00:00:06.000 align:middle line:90% [CLICKING] 00:00:06.000 --> 00:00:11.622 align:middle line:90% 00:00:11.622 --> 00:00:13.080 align:middle line:84% PROFESSOR: Before diving into this, 00:00:13.080 --> 00:00:15.750 align:middle line:84% I'd like to remind you of what we 00:00:15.750 --> 00:00:21.220 align:middle line:84% talked about earlier in the course, which is switching loss. 00:00:21.220 --> 00:00:23.370 align:middle line:84% If I took-- for example, let me just 00:00:23.370 --> 00:00:26.400 align:middle line:84% take my friendly neighborhood buck converter here. 00:00:26.400 --> 00:00:28.290 align:middle line:90% so I have V IN. 00:00:28.290 --> 00:00:29.633 align:middle line:90% And I might have a switch. 00:00:29.633 --> 00:00:32.369 align:middle line:90% 00:00:32.369 --> 00:00:34.470 align:middle line:90% And maybe I have a diode. 00:00:34.470 --> 00:00:38.040 align:middle line:84% And I'll think of these as ideal. 00:00:38.040 --> 00:00:42.390 align:middle line:84% And actually, perhaps let me just leave 00:00:42.390 --> 00:00:43.905 align:middle line:90% myself a little more room here. 00:00:43.905 --> 00:00:44.655 align:middle line:90% I'll have a diode. 00:00:44.655 --> 00:00:47.190 align:middle line:90% 00:00:47.190 --> 00:00:50.530 align:middle line:84% And maybe for the moment, I'll think of this as this ID, 00:00:50.530 --> 00:00:52.920 align:middle line:84% this almost constant current into the output 00:00:52.920 --> 00:00:56.550 align:middle line:84% that I'm controlling by PW-ing my converter. 00:00:56.550 --> 00:01:02.070 align:middle line:84% And we can say, OK, well, I'm interested in the losses 00:01:02.070 --> 00:01:03.310 align:middle line:90% in the switch and the diode. 00:01:03.310 --> 00:01:05.550 align:middle line:90% Let's focus on the switch here. 00:01:05.550 --> 00:01:08.340 align:middle line:84% So I might have some switch voltage 00:01:08.340 --> 00:01:09.645 align:middle line:90% that I'll call the switch. 00:01:09.645 --> 00:01:12.780 align:middle line:90% 00:01:12.780 --> 00:01:16.920 align:middle line:84% And I might have some switch current that I'll call I switch. 00:01:16.920 --> 00:01:20.730 align:middle line:90% 00:01:20.730 --> 00:01:24.660 align:middle line:84% And I might say, well, when my switch is on, 00:01:24.660 --> 00:01:28.050 align:middle line:84% I have I switch squared R on for the switch. 00:01:28.050 --> 00:01:31.360 align:middle line:84% And that's how I figure out my conduction loss on my switch. 00:01:31.360 --> 00:01:34.440 align:middle line:84% We also said you have to worry about switching losses 00:01:34.440 --> 00:01:35.880 align:middle line:90% in the device. 00:01:35.880 --> 00:01:43.050 align:middle line:84% And we might approximate switching loss this way. 00:01:43.050 --> 00:01:46.410 align:middle line:84% We might say, OK, let me just focus in 00:01:46.410 --> 00:01:48.360 align:middle line:84% on the tiny part of the time cycle 00:01:48.360 --> 00:01:51.510 align:middle line:84% when I'm, say, turning this device on and off 00:01:51.510 --> 00:01:53.320 align:middle line:84% and turning this device on and off. 00:01:53.320 --> 00:01:55.600 align:middle line:84% I'm switching between the two devices. 00:01:55.600 --> 00:01:58.200 align:middle line:84% In an ideal world, that happens instantly, 00:01:58.200 --> 00:01:59.890 align:middle line:84% but in the real world, it doesn't. 00:01:59.890 --> 00:02:02.640 align:middle line:84% And it can take, depending upon my switching frequency, maybe 00:02:02.640 --> 00:02:05.400 align:middle line:84% a percent of the cycle or something like that. 00:02:05.400 --> 00:02:05.970 align:middle line:90% OK. 00:02:05.970 --> 00:02:07.530 align:middle line:90% So let's just think. 00:02:07.530 --> 00:02:13.450 align:middle line:84% What happens when I turn this switch off? 00:02:13.450 --> 00:02:16.570 align:middle line:84% Suppose this switch is carrying ID. 00:02:16.570 --> 00:02:19.750 align:middle line:90% I would like to turn him off. 00:02:19.750 --> 00:02:22.450 align:middle line:84% I can worry about what does v switch do and I switch do. 00:02:22.450 --> 00:02:23.950 align:middle line:84% And if I just thought of this switch 00:02:23.950 --> 00:02:27.070 align:middle line:84% as an ideal current source ramping 00:02:27.070 --> 00:02:30.940 align:middle line:84% from carrying high current down to zero, 00:02:30.940 --> 00:02:32.870 align:middle line:84% maybe that would do something like this. 00:02:32.870 --> 00:02:33.915 align:middle line:90% So here's I switch. 00:02:33.915 --> 00:02:36.500 align:middle line:90% 00:02:36.500 --> 00:02:40.978 align:middle line:84% And at some point, it's going to fall from high current 00:02:40.978 --> 00:02:41.520 align:middle line:90% down to zero. 00:02:41.520 --> 00:02:43.670 align:middle line:84% Now the question is, does the switch voltage 00:02:43.670 --> 00:02:47.580 align:middle line:84% rise before or after the switch current falls? 00:02:47.580 --> 00:02:51.940 align:middle line:84% And to the extent that we can think of this diode as being 00:02:51.940 --> 00:02:59.520 align:middle line:84% ideal, the diode cannot carry any current until the diode has 00:02:59.520 --> 00:03:01.602 align:middle line:90% zero volts on them. 00:03:01.602 --> 00:03:03.060 align:middle line:84% He's not going to carry any current 00:03:03.060 --> 00:03:04.310 align:middle line:90% till he has zero volts on him. 00:03:04.310 --> 00:03:07.740 align:middle line:84% That means that the switch voltage must go up to V IN 00:03:07.740 --> 00:03:10.410 align:middle line:84% before this diode can start picking up current. 00:03:10.410 --> 00:03:13.950 align:middle line:84% So I might expect then that the switch voltage, 00:03:13.950 --> 00:03:16.500 align:middle line:84% if I'm going to have my switching start here, 00:03:16.500 --> 00:03:20.790 align:middle line:84% the switch voltage has to rise from 0 to V IN at which point 00:03:20.790 --> 00:03:23.250 align:middle line:84% the current in the switch can start falling. 00:03:23.250 --> 00:03:23.910 align:middle line:90% OK. 00:03:23.910 --> 00:03:27.570 align:middle line:84% So I'm going to have some switching transition. 00:03:27.570 --> 00:03:32.740 align:middle line:84% And I'm approximating these transitions as linear. 00:03:32.740 --> 00:03:34.180 align:middle line:90% What do they actually look like? 00:03:34.180 --> 00:03:36.660 align:middle line:84% That depends upon the device, and the device driver, 00:03:36.660 --> 00:03:37.780 align:middle line:90% and a lot of other things. 00:03:37.780 --> 00:03:42.210 align:middle line:84% But approximating those linear is often not a bad idea. 00:03:42.210 --> 00:03:46.980 align:middle line:84% And I might say that this time period takes some, 00:03:46.980 --> 00:03:49.560 align:middle line:90% say, time t fall. 00:03:49.560 --> 00:03:51.690 align:middle line:84% And in fact, if you look in datasheets, 00:03:51.690 --> 00:03:54.300 align:middle line:84% you can find expressions for t fall 00:03:54.300 --> 00:03:56.490 align:middle line:84% under some nominal drive conditions. 00:03:56.490 --> 00:03:59.440 align:middle line:84% It'll give you some approximate notion 00:03:59.440 --> 00:04:01.330 align:middle line:90% of how long that will take. 00:04:01.330 --> 00:04:04.360 align:middle line:84% But we can control it by the driver we choose 00:04:04.360 --> 00:04:05.990 align:middle line:90% and other things. 00:04:05.990 --> 00:04:06.490 align:middle line:90% OK. 00:04:06.490 --> 00:04:07.580 align:middle line:90% But here's the point. 00:04:07.580 --> 00:04:10.270 align:middle line:84% During this time period, the switching transition 00:04:10.270 --> 00:04:14.920 align:middle line:84% when the voltage on the switch is rising 00:04:14.920 --> 00:04:17.950 align:middle line:84% and the current on the switch is falling, 00:04:17.950 --> 00:04:21.140 align:middle line:84% I have a big overlap in voltage and current on that switch. 00:04:21.140 --> 00:04:21.860 align:middle line:90% So right. 00:04:21.860 --> 00:04:24.752 align:middle line:84% So if I looked at the power in the switch, 00:04:24.752 --> 00:04:26.210 align:middle line:84% the power dissipated in the switch, 00:04:26.210 --> 00:04:28.580 align:middle line:84% yeah I had some i squared R going on here. 00:04:28.580 --> 00:04:31.510 align:middle line:84% It wasn't 0 but it was pretty darn small. 00:04:31.510 --> 00:04:34.870 align:middle line:84% Here I may have, if this is 100 volts and 100 amps, 00:04:34.870 --> 00:04:38.180 align:middle line:84% maybe that's like 5 kilowatts peak dissipation. 00:04:38.180 --> 00:04:41.140 align:middle line:84% It's 5 kilowatts for a very short period of time 00:04:41.140 --> 00:04:42.770 align:middle line:90% that just goes-- 00:04:42.770 --> 00:04:44.740 align:middle line:84% actually, I haven't drawn that very well-- 00:04:44.740 --> 00:04:48.040 align:middle line:84% that just goes over this duration. 00:04:48.040 --> 00:04:51.413 align:middle line:84% But I get a big pulse of power in the switch. 00:04:51.413 --> 00:04:55.240 align:middle line:90% 00:04:55.240 --> 00:04:59.380 align:middle line:84% And this is taking place over a very short time period, 00:04:59.380 --> 00:05:03.790 align:middle line:84% Maybe if I was switching a megahertz, 00:05:03.790 --> 00:05:06.080 align:middle line:84% it would be a microsecond period. 00:05:06.080 --> 00:05:08.690 align:middle line:84% And maybe this takes 10 nanoseconds. 00:05:08.690 --> 00:05:10.760 align:middle line:84% So maybe that's a percent of the cycle. 00:05:10.760 --> 00:05:14.720 align:middle line:84% But on the other hand, this is a pretty big instantaneous power. 00:05:14.720 --> 00:05:17.230 align:middle line:84% So I actually have to care about figuring out 00:05:17.230 --> 00:05:19.180 align:middle line:90% that power dissipation. 00:05:19.180 --> 00:05:22.540 align:middle line:84% If I waited for the other transition-- 00:05:22.540 --> 00:05:25.060 align:middle line:84% this is when I went from the switch conducting. 00:05:25.060 --> 00:05:26.988 align:middle line:84% The switch turns off and the diode turns on. 00:05:26.988 --> 00:05:28.780 align:middle line:84% If I waited in the other direction, I said, 00:05:28.780 --> 00:05:30.930 align:middle line:90% what is the switch loss? 00:05:30.930 --> 00:05:32.480 align:middle line:90% What's going to happen here? 00:05:32.480 --> 00:05:34.200 align:middle line:90% The opposite. 00:05:34.200 --> 00:05:40.920 align:middle line:84% Before the switch can pick up any current 00:05:40.920 --> 00:05:43.000 align:middle line:84% or for the switch to pick up current, 00:05:43.000 --> 00:05:45.030 align:middle line:84% as long as the diode is carrying any current, 00:05:45.030 --> 00:05:47.350 align:middle line:84% the switch voltage has to stay high. 00:05:47.350 --> 00:05:50.010 align:middle line:84% And so the switch current will come up 00:05:50.010 --> 00:05:52.180 align:middle line:84% before the switch voltage collapses. 00:05:52.180 --> 00:05:54.240 align:middle line:84% And if I looked at the other transition, 00:05:54.240 --> 00:05:55.680 align:middle line:90% I would see something like this. 00:05:55.680 --> 00:05:57.630 align:middle line:84% Later at the other end of the cycle, 00:05:57.630 --> 00:06:01.860 align:middle line:84% I would see something like the switch current rises 00:06:01.860 --> 00:06:06.100 align:middle line:84% and then, after the switch current rises, 00:06:06.100 --> 00:06:08.830 align:middle line:90% the switch voltage can fall. 00:06:08.830 --> 00:06:13.425 align:middle line:84% And I might call this time period t rise. 00:06:13.425 --> 00:06:15.970 align:middle line:90% 00:06:15.970 --> 00:06:21.820 align:middle line:84% And again, I will get a big power pulsation 00:06:21.820 --> 00:06:23.330 align:middle line:90% during that time period. 00:06:23.330 --> 00:06:25.390 align:middle line:84% And then if I use this simplified model, 00:06:25.390 --> 00:06:33.130 align:middle line:84% what I'd get is a switching loss that was equal to V IN, 00:06:33.130 --> 00:06:37.000 align:middle line:84% because that's the peak voltage, times I d, that's 00:06:37.000 --> 00:06:45.220 align:middle line:84% the peak current, times tr plus tf over to times the switching 00:06:45.220 --> 00:06:47.180 align:middle line:90% frequency. 00:06:47.180 --> 00:06:54.580 align:middle line:84% This is the area under each of these times tr. 00:06:54.580 --> 00:06:56.990 align:middle line:84% And I'd get this total power dissipation. 00:06:56.990 --> 00:06:58.380 align:middle line:90% Any questions about that? 00:06:58.380 --> 00:07:01.390 align:middle line:90% 00:07:01.390 --> 00:07:04.780 align:middle line:84% So I care about this because, you know, 00:07:04.780 --> 00:07:07.030 align:middle line:84% we're motivated to turn up our switching 00:07:07.030 --> 00:07:11.620 align:middle line:90% frequency until it hurts. 00:07:11.620 --> 00:07:13.150 align:middle line:84% If I have a low switching frequency, 00:07:13.150 --> 00:07:15.442 align:middle line:84% I have big components, I want my components to be small 00:07:15.442 --> 00:07:18.130 align:middle line:84% so I'm going to keep turning up my switching frequency until 00:07:18.130 --> 00:07:20.870 align:middle line:84% something tells me that I shouldn't. 00:07:20.870 --> 00:07:22.100 align:middle line:90% And here you have it. 00:07:22.100 --> 00:07:23.665 align:middle line:84% You get frequency-dependent losses. 00:07:23.665 --> 00:07:25.790 align:middle line:84% And one of those losses is the switching frequency. 00:07:25.790 --> 00:07:29.740 align:middle line:84% This is proportional to how many times a second I switch. 00:07:29.740 --> 00:07:36.220 align:middle line:84% And we can control that to some extent by making t fall 00:07:36.220 --> 00:07:40.810 align:middle line:84% and t rise short but that also has EMI implications 00:07:40.810 --> 00:07:44.690 align:middle line:90% so we can't do that arbitrarily. 00:07:44.690 --> 00:07:48.050 align:middle line:84% So we're stuck with some amount of switching loss. 00:07:48.050 --> 00:07:51.020 align:middle line:84% And I'm going to pick my switching frequency in part 00:07:51.020 --> 00:07:55.910 align:middle line:84% to balance the switching loss against the conduction loss 00:07:55.910 --> 00:07:58.970 align:middle line:90% or get my total loss down. 00:07:58.970 --> 00:08:05.550 align:middle line:84% And this is a good way to proceed. 00:08:05.550 --> 00:08:08.700 align:middle line:84% You figure out some balance between the size 00:08:08.700 --> 00:08:10.860 align:middle line:84% of the device you're going to use this r on, 00:08:10.860 --> 00:08:16.010 align:middle line:84% how fast you can switch it, and get some total loss. 00:08:16.010 --> 00:08:21.530 align:middle line:84% We also did, however, talk about how might I slow up 00:08:21.530 --> 00:08:26.480 align:middle line:90% these kind of transitions. 00:08:26.480 --> 00:08:28.310 align:middle line:84% We said sometimes those transitions 00:08:28.310 --> 00:08:30.140 align:middle line:90% can cause a lot of EMI. 00:08:30.140 --> 00:08:33.590 align:middle line:84% And one of the tricks that we showed earlier in the term 00:08:33.590 --> 00:08:37.380 align:middle line:84% was we said, OK, well, I could come in here. 00:08:37.380 --> 00:08:40.220 align:middle line:84% And if I put, essentially, a circuit that 00:08:40.220 --> 00:08:44.320 align:middle line:84% looked like it did something like this, 00:08:44.320 --> 00:08:48.590 align:middle line:84% I'll put a capacitor across the device, ideally. 00:08:48.590 --> 00:08:51.040 align:middle line:84% But what I will do is I will make it 00:08:51.040 --> 00:08:55.650 align:middle line:84% so that instead of putting it directly across the device, 00:08:55.650 --> 00:08:57.780 align:middle line:84% I'll do it so that in the forward direction, 00:08:57.780 --> 00:08:59.970 align:middle line:84% it looks directly across the device 00:08:59.970 --> 00:09:04.410 align:middle line:84% and, in the reverse direction, I have a resistor here. 00:09:04.410 --> 00:09:07.560 align:middle line:84% If I did that, what that would do 00:09:07.560 --> 00:09:09.810 align:middle line:84% is when I turn the transistor off, 00:09:09.810 --> 00:09:12.240 align:middle line:84% I have another path for current to go through 00:09:12.240 --> 00:09:15.030 align:middle line:90% and I can charge the-- 00:09:15.030 --> 00:09:18.480 align:middle line:84% I can charge up this capacitance so voltage doesn't 00:09:18.480 --> 00:09:20.790 align:middle line:90% need to rise so quickly. 00:09:20.790 --> 00:09:26.700 align:middle line:84% So if I do this, if I put this kind of circuit in, 00:09:26.700 --> 00:09:31.034 align:middle line:84% I can delay the voltage rise in the device. 00:09:31.034 --> 00:09:33.220 align:middle line:90% It might do something like this. 00:09:33.220 --> 00:09:36.390 align:middle line:90% So I kind of push this out here. 00:09:36.390 --> 00:09:38.830 align:middle line:90% And I reduce the VI overlap. 00:09:38.830 --> 00:09:43.500 align:middle line:84% And I also, instead of a fast rise, I get a slow rise. 00:09:43.500 --> 00:09:47.730 align:middle line:90% And that works. 00:09:47.730 --> 00:09:51.120 align:middle line:84% And it does reduce the loss of turn off. 00:09:51.120 --> 00:09:52.740 align:middle line:90% That's very nice. 00:09:52.740 --> 00:09:56.010 align:middle line:84% The penalty I pay is I put a bunch of energy 00:09:56.010 --> 00:09:58.280 align:middle line:90% in this capacitor. 00:09:58.280 --> 00:10:01.400 align:middle line:84% And when I go to turn the switch back on, 00:10:01.400 --> 00:10:04.530 align:middle line:84% that energy ultimately gets dumped into this resistor. 00:10:04.530 --> 00:10:07.010 align:middle line:84% And in practice, typically, I make 00:10:07.010 --> 00:10:09.650 align:middle line:84% the efficiency of the converter worse, not better. 00:10:09.650 --> 00:10:13.370 align:middle line:84% I've reduced the loss in the device, 00:10:13.370 --> 00:10:17.240 align:middle line:84% but I've more than offset that with loss in the resistor 00:10:17.240 --> 00:10:19.340 align:middle line:90% I get here. 00:10:19.340 --> 00:10:20.640 align:middle line:90% OK. 00:10:20.640 --> 00:10:23.340 align:middle line:84% I could do, if I had a device that needed it, 00:10:23.340 --> 00:10:26.490 align:middle line:84% I could do the same kind of thing at turn on. 00:10:26.490 --> 00:10:29.790 align:middle line:84% I could come in here and I could say, OK, 00:10:29.790 --> 00:10:33.030 align:middle line:84% let me add an inductor here, and then I 00:10:33.030 --> 00:10:37.440 align:middle line:90% will add a resistor and a diode. 00:10:37.440 --> 00:10:44.010 align:middle line:84% So when I go to turn the device on, this will hold the voltage 00:10:44.010 --> 00:10:47.730 align:middle line:84% and keep the current from rising very quickly so instead 00:10:47.730 --> 00:10:50.670 align:middle line:84% of the current rising like this, the current rising 00:10:50.670 --> 00:10:51.545 align:middle line:90% might do this. 00:10:51.545 --> 00:10:54.460 align:middle line:90% 00:10:54.460 --> 00:10:57.070 align:middle line:84% And I will again reduce the overlap 00:10:57.070 --> 00:11:00.130 align:middle line:84% between the voltage and current by delaying the current rise. 00:11:00.130 --> 00:11:03.060 align:middle line:84% But I store a bunch of energy in this inductor. 00:11:03.060 --> 00:11:04.990 align:middle line:84% When I go to turn the switch off, 00:11:04.990 --> 00:11:06.880 align:middle line:84% that current will circulate in here 00:11:06.880 --> 00:11:08.590 align:middle line:90% and dissipate in the resistor. 00:11:08.590 --> 00:11:13.270 align:middle line:84% So again, I have reduced the overlap at the switching 00:11:13.270 --> 00:11:16.030 align:middle line:84% transition, but I have lost energy 00:11:16.030 --> 00:11:17.740 align:middle line:90% in the process on balance. 00:11:17.740 --> 00:11:19.750 align:middle line:84% But I might want to do it for reasons 00:11:19.750 --> 00:11:25.100 align:middle line:84% of changing, say, the di/dt or the dv/dt's over here. 00:11:25.100 --> 00:11:27.210 align:middle line:84% So that's all a little bit of a review. 00:11:27.210 --> 00:11:30.880 align:middle line:90% Any questions about that? 00:11:30.880 --> 00:11:31.510 align:middle line:90% STUDENT: Yeah. 00:11:31.510 --> 00:11:35.890 align:middle line:84% Are we expected to look into snubbers for the project? 00:11:35.890 --> 00:11:37.990 align:middle line:84% PROFESSOR: That's a very good question. 00:11:37.990 --> 00:11:40.117 align:middle line:90% So, no. 00:11:40.117 --> 00:11:40.617 align:middle line:90% Right? 00:11:40.617 --> 00:11:42.110 align:middle line:90% Why? 00:11:42.110 --> 00:11:44.170 align:middle line:84% Because the devices that you're using, 00:11:44.170 --> 00:11:46.390 align:middle line:84% MOSFETs and Schottky diodes, tend 00:11:46.390 --> 00:11:49.300 align:middle line:84% not to be very susceptible to failure 00:11:49.300 --> 00:11:54.130 align:middle line:84% of having a safe operating area limitation or a dv/dt 00:11:54.130 --> 00:11:56.920 align:middle line:84% or a di/dt limit in their operation. 00:11:56.920 --> 00:12:02.920 align:middle line:84% If I were working with giant gate turn off thyristors, 00:12:02.920 --> 00:12:04.810 align:middle line:84% then that would be a different story. 00:12:04.810 --> 00:12:06.770 align:middle line:84% And I would have to put snubbers on them. 00:12:06.770 --> 00:12:08.860 align:middle line:84% But for the kind of converter you're designing 00:12:08.860 --> 00:12:11.200 align:middle line:84% and the design project, pretty much you 00:12:11.200 --> 00:12:12.640 align:middle line:90% can get away without snubbers. 00:12:12.640 --> 00:12:16.180 align:middle line:84% In the real world, if you were to build one, you might build it 00:12:16.180 --> 00:12:19.300 align:middle line:84% and then say, oh, I'm getting a little bit too much EMI 00:12:19.300 --> 00:12:21.820 align:middle line:84% and I associate that with the slew rate of the voltage, 00:12:21.820 --> 00:12:22.880 align:middle line:90% for example. 00:12:22.880 --> 00:12:25.000 align:middle line:84% And then I might go in, throw in a small snubber 00:12:25.000 --> 00:12:27.050 align:middle line:84% to slow down the voltage transition. 00:12:27.050 --> 00:12:31.190 align:middle line:84% So sometimes people will leave provision 00:12:31.190 --> 00:12:33.407 align:middle line:84% to put a snubber in if they need it in practice, 00:12:33.407 --> 00:12:35.240 align:middle line:84% but it's not something you would necessarily 00:12:35.240 --> 00:12:38.600 align:middle line:84% have to design upfront unless the types of devices 00:12:38.600 --> 00:12:41.220 align:middle line:90% you were using required it. 00:12:41.220 --> 00:12:44.600 align:middle line:84% And that would be typically very high power devices 00:12:44.600 --> 00:12:47.690 align:middle line:84% that are a bit more fragile for MOSFETs and Schottky 00:12:47.690 --> 00:12:48.860 align:middle line:90% diodes and stuff. 00:12:48.860 --> 00:12:51.080 align:middle line:84% Typically, you don't have to worry about it. 00:12:51.080 --> 00:12:53.690 align:middle line:90% Excellent question. 00:12:53.690 --> 00:12:54.320 align:middle line:90% OK. 00:12:54.320 --> 00:12:57.080 align:middle line:84% So that brings me to the topic I wanted to tell you about today 00:12:57.080 --> 00:13:01.920 align:middle line:84% and I only really wanted to give you a flavor of today. 00:13:01.920 --> 00:13:05.700 align:middle line:84% But I want to give you a notion of how this works. 00:13:05.700 --> 00:13:07.630 align:middle line:90% This kind of poses a question. 00:13:07.630 --> 00:13:10.290 align:middle line:84% I said putting the inductor and the capacitor 00:13:10.290 --> 00:13:14.490 align:middle line:84% in parallel with the device here does reduce the overlap, 00:13:14.490 --> 00:13:15.960 align:middle line:90% but I burn a bunch of energy. 00:13:15.960 --> 00:13:18.540 align:middle line:84% And likewise, putting the inductor in series 00:13:18.540 --> 00:13:22.510 align:middle line:84% reduces the overlap, but I burn a bunch of energy. 00:13:22.510 --> 00:13:26.230 align:middle line:84% Could I do that without burning a bunch of energy? 00:13:26.230 --> 00:13:30.300 align:middle line:84% Could I somehow get that energy back? 00:13:30.300 --> 00:13:33.130 align:middle line:84% And what would be the benefit of that? 00:13:33.130 --> 00:13:37.650 align:middle line:84% This loss, if I only thought about the overlaps, 00:13:37.650 --> 00:13:40.600 align:middle line:84% this loss would be greatly reduced. 00:13:40.600 --> 00:13:43.620 align:middle line:84% And that would either let me, A, get rid of that loss 00:13:43.620 --> 00:13:46.050 align:middle line:84% and get a very high efficiency converter, 00:13:46.050 --> 00:13:49.470 align:middle line:84% or, B, crank up the switching frequency 00:13:49.470 --> 00:13:53.723 align:middle line:84% and make my component smaller or some combination thereof. 00:13:53.723 --> 00:13:55.140 align:middle line:84% As you make things smaller, it has 00:13:55.140 --> 00:13:58.450 align:middle line:84% to get more efficient in order to meet thermal requirements. 00:13:58.450 --> 00:14:01.380 align:middle line:84% But nonetheless, what that would do 00:14:01.380 --> 00:14:06.540 align:middle line:84% is let me either get to higher efficiencies or higher switching 00:14:06.540 --> 00:14:07.600 align:middle line:90% frequencies. 00:14:07.600 --> 00:14:09.224 align:middle line:90% OK. 00:14:09.224 --> 00:14:10.990 align:middle line:90% And people do this. 00:14:10.990 --> 00:14:14.580 align:middle line:84% They come up with circuits, either changes 00:14:14.580 --> 00:14:17.910 align:middle line:84% in the circuit design or changes in the control, 00:14:17.910 --> 00:14:24.240 align:middle line:84% that will let people get this so-called soft switching. 00:14:24.240 --> 00:14:28.360 align:middle line:84% This pink and yellow transition we would consider hard-switched. 00:14:28.360 --> 00:14:30.880 align:middle line:84% A soft switch transition is one in which 00:14:30.880 --> 00:14:33.910 align:middle line:84% you reduce the voltage and current overlap at the switching 00:14:33.910 --> 00:14:37.900 align:middle line:84% transition and get much lower switching loss. 00:14:37.900 --> 00:14:42.220 align:middle line:84% Now I say that many, many very high performance converters 00:14:42.220 --> 00:14:42.860 align:middle line:90% do this. 00:14:42.860 --> 00:14:45.970 align:middle line:84% This is a very common thing at the high performance 00:14:45.970 --> 00:14:46.820 align:middle line:90% end of the spectrum. 00:14:46.820 --> 00:14:49.990 align:middle line:84% In fact, if you looked at the power converter that's 00:14:49.990 --> 00:14:53.200 align:middle line:84% in this little miniaturized laptop adapter, which 00:14:53.200 --> 00:14:56.710 align:middle line:84% is, A, very small, and, B, exceeds 00:14:56.710 --> 00:14:59.320 align:middle line:84% 95% end to end efficiency so they 00:14:59.320 --> 00:15:01.090 align:middle line:84% can make it small without it getting hot, 00:15:01.090 --> 00:15:05.050 align:middle line:84% that design uses soft switching techniques 00:15:05.050 --> 00:15:09.040 align:middle line:84% and turns up the frequency to make it small 00:15:09.040 --> 00:15:12.500 align:middle line:84% and uses part of that benefit to make it efficient. 00:15:12.500 --> 00:15:15.005 align:middle line:90% 00:15:15.005 --> 00:15:17.550 align:middle line:84% So what I want to do is show you the kind 00:15:17.550 --> 00:15:20.430 align:middle line:84% of category of techniques that you might use there. 00:15:20.430 --> 00:15:22.800 align:middle line:84% I will say, though, that while this is very, very 00:15:22.800 --> 00:15:25.860 align:middle line:84% common in the high performance end of the spectrum, 00:15:25.860 --> 00:15:28.410 align:middle line:84% where you're trying for great miniaturization or really 00:15:28.410 --> 00:15:31.710 align:middle line:84% fast response, something like that, perhaps the majority 00:15:31.710 --> 00:15:35.020 align:middle line:84% of power converters designed in the world don't do this. 00:15:35.020 --> 00:15:37.720 align:middle line:90% They just get an optimum. 00:15:37.720 --> 00:15:38.220 align:middle line:90% They live. 00:15:38.220 --> 00:15:42.090 align:middle line:84% They try to get good devices, live with the switching loss, 00:15:42.090 --> 00:15:43.570 align:middle line:90% make the design trade-offs. 00:15:43.570 --> 00:15:46.860 align:middle line:84% All the stuff I showed you so far in the term 00:15:46.860 --> 00:15:50.940 align:middle line:84% is the majority of things, that's the way they're designed. 00:15:50.940 --> 00:15:53.520 align:middle line:84% Nonetheless, when you're really trying to push the boundaries 00:15:53.520 --> 00:15:55.230 align:middle line:84% or, for whatever reason, you want 00:15:55.230 --> 00:15:58.350 align:middle line:84% to go to very high frequencies, suddenly you 00:15:58.350 --> 00:16:00.700 align:middle line:84% have to use these other kinds of techniques. 00:16:00.700 --> 00:16:01.200 align:middle line:90% OK. 00:16:01.200 --> 00:16:02.345 align:middle line:90% Any questions about that? 00:16:02.345 --> 00:16:04.910 align:middle line:90% 00:16:04.910 --> 00:16:06.140 align:middle line:90% OK. 00:16:06.140 --> 00:16:08.810 align:middle line:84% So let's think about what we might 00:16:08.810 --> 00:16:12.806 align:middle line:84% do to build so-called soft switch converters. 00:16:12.806 --> 00:16:19.930 align:middle line:90% 00:16:19.930 --> 00:16:22.070 align:middle line:90% And how soft is soft? 00:16:22.070 --> 00:16:25.670 align:middle line:84% Well, there's all degrees of softness. 00:16:25.670 --> 00:16:27.580 align:middle line:90% But we'll talk about it. 00:16:27.580 --> 00:16:29.860 align:middle line:90% Now there's two flavors. 00:16:29.860 --> 00:16:35.650 align:middle line:84% There's one that I'll call zero voltage switching, ZVS, and one 00:16:35.650 --> 00:16:39.280 align:middle line:84% that I'll call zero current switching, ZCS. 00:16:39.280 --> 00:16:43.120 align:middle line:84% So you'll see these acronyms used everywhere. 00:16:43.120 --> 00:16:47.020 align:middle line:84% And essentially, I'll describe what they mean. 00:16:47.020 --> 00:16:51.940 align:middle line:84% You also have any-- if I think about a given switch, 00:16:51.940 --> 00:16:54.610 align:middle line:84% like here I'll think about the active switch here, 00:16:54.610 --> 00:16:56.090 align:middle line:90% he has two transitions. 00:16:56.090 --> 00:16:58.790 align:middle line:84% He has a turn on transition and a turn off transition. 00:16:58.790 --> 00:17:02.020 align:middle line:84% I got to think about both of those because both of those 00:17:02.020 --> 00:17:03.910 align:middle line:90% give me some loss. 00:17:03.910 --> 00:17:06.295 align:middle line:90% So I might think about turn off. 00:17:06.295 --> 00:17:11.560 align:middle line:90% 00:17:11.560 --> 00:17:15.339 align:middle line:90% And I might think about turn on. 00:17:15.339 --> 00:17:15.970 align:middle line:90% OK. 00:17:15.970 --> 00:17:20.000 align:middle line:84% So let's think about these four possibilities. 00:17:20.000 --> 00:17:21.710 align:middle line:90% OK. 00:17:21.710 --> 00:17:24.230 align:middle line:90% What is ZVS turn on? 00:17:24.230 --> 00:17:28.670 align:middle line:84% ZVS turn off, if I thought about, say, the turn off 00:17:28.670 --> 00:17:34.670 align:middle line:84% I had here, I had something where the current was 00:17:34.670 --> 00:17:40.730 align:middle line:90% falling like this at turn off. 00:17:40.730 --> 00:17:45.410 align:middle line:84% And under a hard switch transition-- 00:17:45.410 --> 00:17:48.170 align:middle line:90% so this is I switch. 00:17:48.170 --> 00:17:50.870 align:middle line:84% Under a hard switch transition, we expect kind of V 00:17:50.870 --> 00:17:56.330 align:middle line:84% switch to rise before the current falls like this. 00:17:56.330 --> 00:17:58.740 align:middle line:84% And that would be a hard switch transition. 00:17:58.740 --> 00:18:01.430 align:middle line:84% What we do in a zero voltage turn 00:18:01.430 --> 00:18:08.220 align:middle line:84% off is we typically do something to delay 00:18:08.220 --> 00:18:10.560 align:middle line:90% the voltage rise on the switch. 00:18:10.560 --> 00:18:13.830 align:middle line:84% Maybe it does something like that. 00:18:13.830 --> 00:18:16.035 align:middle line:84% Well, that ought to look pretty familiar to you. 00:18:16.035 --> 00:18:16.680 align:middle line:90% Right? 00:18:16.680 --> 00:18:19.590 align:middle line:84% That's what the snubber did over here. 00:18:19.590 --> 00:18:26.070 align:middle line:84% So I might get this kind of behavior 00:18:26.070 --> 00:18:29.630 align:middle line:84% by taking my switch that I'm going 00:18:29.630 --> 00:18:34.370 align:middle line:84% to try to turn off and just put a capacitor across them. 00:18:34.370 --> 00:18:38.510 align:middle line:84% And so if I put an external capacitor across the switch, 00:18:38.510 --> 00:18:40.940 align:middle line:84% when it turns off, it'll delay the rise of the switch 00:18:40.940 --> 00:18:42.920 align:middle line:90% and I'll reduce the overlap. 00:18:42.920 --> 00:18:46.550 align:middle line:84% So it's no different than what the snubber does, but now 00:18:46.550 --> 00:18:49.040 align:middle line:84% that I've put energy in this capacitor, 00:18:49.040 --> 00:18:51.470 align:middle line:84% we've got to figure out a way to get that energy back 00:18:51.470 --> 00:18:53.080 align:middle line:90% without dissipating it. 00:18:53.080 --> 00:18:55.220 align:middle line:84% So I can't just turn on the switch 00:18:55.220 --> 00:18:58.000 align:middle line:84% because I dumped that energy and lose it. 00:18:58.000 --> 00:19:00.620 align:middle line:90% OK. 00:19:00.620 --> 00:19:07.310 align:middle line:90% What about zero voltage turn on? 00:19:07.310 --> 00:19:07.810 align:middle line:90% OK. 00:19:07.810 --> 00:19:09.560 align:middle line:84% What would zero voltage turn on look like? 00:19:09.560 --> 00:19:14.410 align:middle line:84% Well, we said, OK, what we expect normally at turn 00:19:14.410 --> 00:19:18.910 align:middle line:84% on is the current in the switch is going to rise 00:19:18.910 --> 00:19:21.361 align:middle line:90% at the turn on transition. 00:19:21.361 --> 00:19:21.861 align:middle line:90% Right. 00:19:21.861 --> 00:19:24.807 align:middle line:90% 00:19:24.807 --> 00:19:31.030 align:middle line:84% And usually, after the current rises-- 00:19:31.030 --> 00:19:33.400 align:middle line:90% here's the switch voltage-- 00:19:33.400 --> 00:19:35.240 align:middle line:90% the switch voltage falls. 00:19:35.240 --> 00:19:39.050 align:middle line:84% So that's what I would get and I'd have this overlap period. 00:19:39.050 --> 00:19:43.990 align:middle line:84% Well, what if I could be somehow super clever in my circuit 00:19:43.990 --> 00:19:47.170 align:middle line:84% through magical means and just happen 00:19:47.170 --> 00:19:49.660 align:middle line:84% to work it so that instead of the voltage 00:19:49.660 --> 00:19:51.970 align:middle line:84% falling because I'm turning the switch on, 00:19:51.970 --> 00:19:54.700 align:middle line:84% I do something else in the circuit 00:19:54.700 --> 00:19:57.160 align:middle line:84% to make the voltage go to zero before I 00:19:57.160 --> 00:19:58.750 align:middle line:90% try to turn the switch on? 00:19:58.750 --> 00:20:05.240 align:middle line:84% Maybe I have a circuit where somehow through natural action 00:20:05.240 --> 00:20:06.950 align:middle line:84% of the circuit, the voltage goes to zero 00:20:06.950 --> 00:20:09.408 align:middle line:84% and then I can turn the switch on with no voltage across it 00:20:09.408 --> 00:20:11.690 align:middle line:90% and then I get no overlap? 00:20:11.690 --> 00:20:16.190 align:middle line:84% So zero voltage switching, either on or off, 00:20:16.190 --> 00:20:20.180 align:middle line:84% is arranging the switching operation such 00:20:20.180 --> 00:20:24.740 align:middle line:84% that I move the voltage around in a way that gives me 00:20:24.740 --> 00:20:29.800 align:middle line:84% little overlap between voltage and current at the transition. 00:20:29.800 --> 00:20:32.020 align:middle line:84% And we'll talk about how you might do that, 00:20:32.020 --> 00:20:33.230 align:middle line:90% but that's the idea. 00:20:33.230 --> 00:20:34.250 align:middle line:90% Any questions? 00:20:34.250 --> 00:20:37.830 align:middle line:90% 00:20:37.830 --> 00:20:45.110 align:middle line:84% So what would a zero current turn on look like? 00:20:45.110 --> 00:20:51.920 align:middle line:84% Well, we said, OK, for turn on, usually, 00:20:51.920 --> 00:20:54.980 align:middle line:84% what happens is I have my switch current 00:20:54.980 --> 00:20:58.460 align:middle line:84% and he's rising because this is turn on so here's I switch. 00:20:58.460 --> 00:21:00.020 align:middle line:84% The current is going to go from zero 00:21:00.020 --> 00:21:00.830 align:middle line:84% and then I'm going to turn him on. 00:21:00.830 --> 00:21:02.180 align:middle line:90% It's going to be something. 00:21:02.180 --> 00:21:11.090 align:middle line:84% And normally, what happens is the voltage 00:21:11.090 --> 00:21:14.523 align:middle line:90% falls after the current rises. 00:21:14.523 --> 00:21:17.220 align:middle line:90% 00:21:17.220 --> 00:21:19.625 align:middle line:90% Right? 00:21:19.625 --> 00:21:25.680 align:middle line:90% So what if I could-- 00:21:25.680 --> 00:21:27.290 align:middle line:90% what if I could do this? 00:21:27.290 --> 00:21:32.048 align:middle line:90% 00:21:32.048 --> 00:21:32.715 align:middle line:90% What am I doing? 00:21:32.715 --> 00:21:40.020 align:middle line:90% 00:21:40.020 --> 00:21:42.040 align:middle line:90% What if I could do this? 00:21:42.040 --> 00:21:46.080 align:middle line:84% What if I could say, all right, instead of having the switch 00:21:46.080 --> 00:21:51.290 align:middle line:84% current rise early, I did some magic game such 00:21:51.290 --> 00:21:53.545 align:middle line:90% that he didn't rise until later? 00:21:53.545 --> 00:21:56.830 align:middle line:90% 00:21:56.830 --> 00:22:00.010 align:middle line:84% I somehow delay the switch current rise 00:22:00.010 --> 00:22:02.160 align:middle line:84% until after the voltage has fallen. 00:22:02.160 --> 00:22:05.110 align:middle line:90% 00:22:05.110 --> 00:22:09.780 align:middle line:84% I would then have little overlap and that would be good. 00:22:09.780 --> 00:22:12.540 align:middle line:84% Well, this is very similar to the notion 00:22:12.540 --> 00:22:16.290 align:middle line:84% of if I had my switch, you know, maybe I 00:22:16.290 --> 00:22:21.670 align:middle line:84% could go put an inductor in series with the switch. 00:22:21.670 --> 00:22:24.690 align:middle line:84% So even after I turn the switch on, 00:22:24.690 --> 00:22:28.350 align:middle line:84% the voltage doesn't appear across the switch. 00:22:28.350 --> 00:22:30.630 align:middle line:90% It could-- this is v switch. 00:22:30.630 --> 00:22:32.460 align:middle line:84% It could appear across this inductor. 00:22:32.460 --> 00:22:35.105 align:middle line:84% And the current in the inductor will be delayed in its rise. 00:22:35.105 --> 00:22:39.330 align:middle line:84% So in some sense, this snubber helps 00:22:39.330 --> 00:22:41.280 align:middle line:84% give me zero current turn on except 00:22:41.280 --> 00:22:43.150 align:middle line:90% that I'm dumping the energy. 00:22:43.150 --> 00:22:46.450 align:middle line:84% So I can play this same game except that then I 00:22:46.450 --> 00:22:52.740 align:middle line:84% got to figure out how to recover the energy in this inductor. 00:22:52.740 --> 00:22:59.550 align:middle line:84% And you can see that these two zero voltage turn off 00:22:59.550 --> 00:23:03.360 align:middle line:84% and zero current turn on are kind of duals of one another. 00:23:03.360 --> 00:23:06.690 align:middle line:84% I'm just swapping voltage and current waveforms. 00:23:06.690 --> 00:23:11.010 align:middle line:84% And I'm swapping the role of the capacitor and zero voltage 00:23:11.010 --> 00:23:14.400 align:middle line:84% turn off and the inductor and zero current turn on. 00:23:14.400 --> 00:23:18.690 align:middle line:84% There are other ways to get ZVS turn off or ZCS turn on, 00:23:18.690 --> 00:23:22.790 align:middle line:84% but I'm showing you a common way to do it. 00:23:22.790 --> 00:23:24.640 align:middle line:90% What's the last one? 00:23:24.640 --> 00:23:31.100 align:middle line:84% Well, we said I want to do zero current turn off. 00:23:31.100 --> 00:23:36.410 align:middle line:84% And we said what usually happens is that my switch voltage 00:23:36.410 --> 00:23:46.200 align:middle line:84% rises and my switch current can fall only after that happens. 00:23:46.200 --> 00:23:50.970 align:middle line:84% Well, what if I were really clever and said, 00:23:50.970 --> 00:23:53.280 align:middle line:90% OK, here's I switch? 00:23:53.280 --> 00:23:56.040 align:middle line:84% What if I were really clever and somehow I 00:23:56.040 --> 00:24:00.570 align:middle line:84% could make the current go to zero before the time I 00:24:00.570 --> 00:24:02.553 align:middle line:90% wanted to turn off? 00:24:02.553 --> 00:24:03.970 align:middle line:84% If the current were naturally zero 00:24:03.970 --> 00:24:06.730 align:middle line:84% when I was going to turn off, I can turn off the switch 00:24:06.730 --> 00:24:08.450 align:middle line:84% with no overlap in voltage and current. 00:24:08.450 --> 00:24:10.190 align:middle line:90% I'm done. 00:24:10.190 --> 00:24:10.940 align:middle line:90% All right. 00:24:10.940 --> 00:24:13.460 align:middle line:84% I've got to come up with some magical circuit means 00:24:13.460 --> 00:24:14.040 align:middle line:90% to do this. 00:24:14.040 --> 00:24:16.290 align:middle line:84% We've kind of got ideas for how we might do these two. 00:24:16.290 --> 00:24:18.582 align:middle line:84% We've got to figure out how we might do these two guys. 00:24:18.582 --> 00:24:20.130 align:middle line:90% But that's the basic concept. 00:24:20.130 --> 00:24:22.400 align:middle line:84% I'm either going to manipulate the voltage 00:24:22.400 --> 00:24:24.897 align:middle line:84% to make the voltage and current overlap smaller 00:24:24.897 --> 00:24:26.480 align:middle line:84% or I'm going to manipulate the current 00:24:26.480 --> 00:24:29.030 align:middle line:84% to make the voltage and current overlap small or both 00:24:29.030 --> 00:24:30.150 align:middle line:90% for that matter. 00:24:30.150 --> 00:24:35.090 align:middle line:84% But I'm trying to create this kind of magical opportunity 00:24:35.090 --> 00:24:38.900 align:middle line:84% to change my switch configurations at low loss. 00:24:38.900 --> 00:24:42.940 align:middle line:90% And that's not for free. 00:24:42.940 --> 00:24:44.820 align:middle line:84% If I just go to design a buck converter, 00:24:44.820 --> 00:24:47.790 align:middle line:90% I don't get that for free. 00:24:47.790 --> 00:24:50.310 align:middle line:84% But maybe I can do something more complicated 00:24:50.310 --> 00:24:53.320 align:middle line:90% that makes that work. 00:24:53.320 --> 00:24:54.820 align:middle line:84% So does everybody, first of all, get 00:24:54.820 --> 00:24:56.870 align:middle line:90% the premise of soft switching? 00:24:56.870 --> 00:24:59.710 align:middle line:90% 00:24:59.710 --> 00:25:01.210 align:middle line:90% OK. 00:25:01.210 --> 00:25:05.500 align:middle line:84% Let me show you a few examples of different ways people 00:25:05.500 --> 00:25:08.070 align:middle line:90% do this. 00:25:08.070 --> 00:25:11.250 align:middle line:84% And I'm going to show you examples in DC to DC converters 00:25:11.250 --> 00:25:11.850 align:middle line:90% today. 00:25:11.850 --> 00:25:14.340 align:middle line:84% We'll talk about more ideas of doing 00:25:14.340 --> 00:25:18.030 align:middle line:84% soft switching in other kind of converters next class. 00:25:18.030 --> 00:25:24.190 align:middle line:84% But here is the first example I'd like to show you. 00:25:24.190 --> 00:25:27.535 align:middle line:84% Now this circuit-- and I'm showing you an actual circuit 00:25:27.535 --> 00:25:30.160 align:middle line:84% because I wanted to show you the real circuit and the switching 00:25:30.160 --> 00:25:31.450 align:middle line:90% waveforms from it-- 00:25:31.450 --> 00:25:34.520 align:middle line:84% is a version of a buck converter. 00:25:34.520 --> 00:25:36.430 align:middle line:84% And you could do this same technique 00:25:36.430 --> 00:25:38.005 align:middle line:90% in a regular buck converter. 00:25:38.005 --> 00:25:40.630 align:middle line:84% It's just that this version of the circuit I'm showing you here 00:25:40.630 --> 00:25:42.370 align:middle line:84% is what we would call an inverted buck 00:25:42.370 --> 00:25:44.930 align:middle line:84% converter or a common positive buck converter. 00:25:44.930 --> 00:25:48.130 align:middle line:84% That is, the switch is ground referenced and the input 00:25:48.130 --> 00:25:50.477 align:middle line:84% and output share a common positive terminal 00:25:50.477 --> 00:25:52.060 align:middle line:84% instead of a common negative terminal, 00:25:52.060 --> 00:25:53.390 align:middle line:90% which they normally would. 00:25:53.390 --> 00:25:55.900 align:middle line:84% But you could see that this is still a buck converter. 00:25:55.900 --> 00:25:59.560 align:middle line:84% When the switch is on, I get V in minus V out 00:25:59.560 --> 00:26:01.000 align:middle line:90% across the inductor. 00:26:01.000 --> 00:26:03.160 align:middle line:84% And when the switch is off, I get minus V out 00:26:03.160 --> 00:26:04.070 align:middle line:90% across the inductor. 00:26:04.070 --> 00:26:06.220 align:middle line:90% It is a buck converter. 00:26:06.220 --> 00:26:09.620 align:middle line:84% But how are we going to run this buck converter? 00:26:09.620 --> 00:26:11.680 align:middle line:84% Well, let me tell you how we're going 00:26:11.680 --> 00:26:12.980 align:middle line:90% to run this buck converter. 00:26:12.980 --> 00:26:17.110 align:middle line:84% And the one thing you can't see in this picture-- 00:26:17.110 --> 00:26:18.772 align:middle line:84% first of all, is there any question 00:26:18.772 --> 00:26:20.480 align:middle line:84% as to why this thing is a buck converter? 00:26:20.480 --> 00:26:23.990 align:middle line:90% 00:26:23.990 --> 00:26:24.590 align:middle line:90% OK. 00:26:24.590 --> 00:26:26.545 align:middle line:90% So here's the circuit. 00:26:26.545 --> 00:26:29.330 align:middle line:90% 00:26:29.330 --> 00:26:32.760 align:middle line:90% And it looks like this. 00:26:32.760 --> 00:26:34.130 align:middle line:90% So this is my buck-- 00:26:34.130 --> 00:26:35.747 align:middle line:90% this is my basic buck converter. 00:26:35.747 --> 00:26:37.580 align:middle line:84% If I drew this for you and put it on an exam 00:26:37.580 --> 00:26:39.788 align:middle line:84% and say what kind of converter that is, you'd tell me 00:26:39.788 --> 00:26:42.410 align:middle line:90% it's a buck converter. 00:26:42.410 --> 00:26:43.530 align:middle line:90% Missing something here. 00:26:43.530 --> 00:26:44.780 align:middle line:90% I'm missing my output voltage. 00:26:44.780 --> 00:26:48.580 align:middle line:90% 00:26:48.580 --> 00:26:51.280 align:middle line:90% So here we go. 00:26:51.280 --> 00:26:55.570 align:middle line:84% Here's my output capacitor and my output resistor, for example. 00:26:55.570 --> 00:27:00.910 align:middle line:90% Here's V out and here's IL. 00:27:00.910 --> 00:27:01.750 align:middle line:90% OK. 00:27:01.750 --> 00:27:03.430 align:middle line:90% That's a buck converter. 00:27:03.430 --> 00:27:07.340 align:middle line:90% 00:27:07.340 --> 00:27:09.710 align:middle line:84% We're going to have to do some work 00:27:09.710 --> 00:27:13.230 align:middle line:84% to get soft switching in this converter. 00:27:13.230 --> 00:27:15.390 align:middle line:84% But let me tell you the one thing that's missing 00:27:15.390 --> 00:27:17.040 align:middle line:90% in the circuit diagram picture. 00:27:17.040 --> 00:27:21.510 align:middle line:84% And that is that there's a capacitance here, 00:27:21.510 --> 00:27:26.870 align:middle line:84% an effective capacitance here, across the device. 00:27:26.870 --> 00:27:29.340 align:middle line:84% Where does that capacitance come from? 00:27:29.340 --> 00:27:33.120 align:middle line:84% Well, it comes from the fact that this switch 00:27:33.120 --> 00:27:34.860 align:middle line:90% has capacitance. 00:27:34.860 --> 00:27:37.320 align:middle line:84% And so that's in parallel with the device. 00:27:37.320 --> 00:27:41.200 align:middle line:84% The diode also has some capacitance. 00:27:41.200 --> 00:27:44.780 align:middle line:84% And that's connected between this node and a fixed potential. 00:27:44.780 --> 00:27:47.080 align:middle line:84% So from an AC perspective, it also 00:27:47.080 --> 00:27:49.623 align:middle line:84% looks like capacitance sitting here. 00:27:49.623 --> 00:27:51.040 align:middle line:84% So what we're going to do is we're 00:27:51.040 --> 00:27:55.240 align:middle line:84% going to use the net capacitance at this node to help give us 00:27:55.240 --> 00:27:58.975 align:middle line:90% zero voltage switching. 00:27:58.975 --> 00:28:02.440 align:middle line:84% And I'll remind you that a capacitor across the device 00:28:02.440 --> 00:28:06.140 align:middle line:84% is just what we need to get zero voltage turn off. 00:28:06.140 --> 00:28:10.270 align:middle line:84% So we're going to do this trick because I've put 00:28:10.270 --> 00:28:11.600 align:middle line:90% a capacitor across the device. 00:28:11.600 --> 00:28:13.990 align:middle line:84% So if I see a design with a capacitor across the device, 00:28:13.990 --> 00:28:16.460 align:middle line:84% I think, yeah, I'm probably doing zero voltage switching. 00:28:16.460 --> 00:28:18.940 align:middle line:90% So how does this thing work? 00:28:18.940 --> 00:28:19.720 align:middle line:90% OK. 00:28:19.720 --> 00:28:20.920 align:middle line:90% Well, I'm going to turn-- 00:28:20.920 --> 00:28:24.120 align:middle line:90% let's assume the switch is on. 00:28:24.120 --> 00:28:27.130 align:middle line:84% I'm in my normal state when the switch is on. 00:28:27.130 --> 00:28:29.930 align:middle line:84% The inductor current's going to rise. 00:28:29.930 --> 00:28:30.665 align:middle line:90% OK. 00:28:30.665 --> 00:28:32.540 align:middle line:84% We're going to let the inductor current rise. 00:28:32.540 --> 00:28:34.748 align:middle line:84% And we're actually going to have a large ripple ratio 00:28:34.748 --> 00:28:36.410 align:middle line:90% in this inductor current. 00:28:36.410 --> 00:28:37.970 align:middle line:84% And eventually, the inductor current 00:28:37.970 --> 00:28:39.553 align:middle line:84% is going to get up to some level where 00:28:39.553 --> 00:28:41.600 align:middle line:90% I tell the switch to turn off. 00:28:41.600 --> 00:28:42.100 align:middle line:90% OK. 00:28:42.100 --> 00:28:53.230 align:middle line:84% Now what happens when this inductor current IL is big 00:28:53.230 --> 00:28:54.430 align:middle line:90% and I turn off? 00:28:54.430 --> 00:28:57.700 align:middle line:84% Well, if this capacitor's sitting across the switch, 00:28:57.700 --> 00:28:59.890 align:middle line:84% even though IL is big and coming here, 00:28:59.890 --> 00:29:05.670 align:middle line:84% this capacitor delays the rise in voltage on the switch. 00:29:05.670 --> 00:29:09.390 align:middle line:84% And I can turn off the switch under approximately zero voltage 00:29:09.390 --> 00:29:11.670 align:middle line:84% because of the capacitance of this node. 00:29:11.670 --> 00:29:12.300 align:middle line:90% All right. 00:29:12.300 --> 00:29:17.340 align:middle line:84% So that's exactly this picture here. 00:29:17.340 --> 00:29:18.270 align:middle line:90% OK. 00:29:18.270 --> 00:29:22.950 align:middle line:84% So what happens, eventually, this inductor 00:29:22.950 --> 00:29:27.320 align:middle line:84% charges up this capacitor until the diode turns on. 00:29:27.320 --> 00:29:29.660 align:middle line:84% So I get a zero voltage transition. 00:29:29.660 --> 00:29:32.330 align:middle line:84% You can see that in this operating diagram here. 00:29:32.330 --> 00:29:36.790 align:middle line:84% The voltage across the switch goes up 00:29:36.790 --> 00:29:41.230 align:middle line:84% until it hits VN, which is exactly 00:29:41.230 --> 00:29:43.150 align:middle line:84% the condition in which the diode is on 00:29:43.150 --> 00:29:45.550 align:middle line:84% and the inductor is going to circulate 00:29:45.550 --> 00:29:47.120 align:middle line:90% its current into the output. 00:29:47.120 --> 00:29:49.900 align:middle line:84% Does that make sense to everybody? 00:29:49.900 --> 00:29:51.850 align:middle line:84% Now what I have to do in this circuit 00:29:51.850 --> 00:29:57.730 align:middle line:84% is somehow ensure that this voltage across the switch 00:29:57.730 --> 00:30:00.490 align:middle line:84% be switch, which I guess I should have kept 00:30:00.490 --> 00:30:02.720 align:middle line:90% my color scheme consistent. 00:30:02.720 --> 00:30:08.282 align:middle line:84% V switch is going to get back down to 0. 00:30:08.282 --> 00:30:09.740 align:middle line:84% How does that work in this circuit? 00:30:09.740 --> 00:30:11.440 align:middle line:84% What magic do I do in the circuit? 00:30:11.440 --> 00:30:13.390 align:middle line:84% Well, in this circuit, what we're going to do 00:30:13.390 --> 00:30:15.582 align:middle line:84% is we're going to let the diode conduct. 00:30:15.582 --> 00:30:18.040 align:middle line:84% And because there's a negative voltage across the inductor, 00:30:18.040 --> 00:30:21.360 align:middle line:84% the inductor current is going to ramp down. 00:30:21.360 --> 00:30:25.000 align:middle line:84% Until the inductor current hits 0. 00:30:25.000 --> 00:30:31.210 align:middle line:84% OK once the inductor current hits zero, the switch is off. 00:30:31.210 --> 00:30:33.010 align:middle line:90% The diode is off. 00:30:33.010 --> 00:30:35.710 align:middle line:84% And all I essentially have is a resonance. 00:30:35.710 --> 00:30:38.100 align:middle line:84% If I think about V out as being constant, 00:30:38.100 --> 00:30:41.530 align:middle line:84% I have a resonance between the buck conductor 00:30:41.530 --> 00:30:45.250 align:middle line:84% and the capacitance at the switch. 00:30:45.250 --> 00:30:48.620 align:middle line:90% So both devices are now off. 00:30:48.620 --> 00:30:51.770 align:middle line:84% By the way, this diode is turned off at zero current, 00:30:51.770 --> 00:30:54.770 align:middle line:84% naturally, because the current's ramped down to zero. 00:30:54.770 --> 00:30:56.270 align:middle line:90% OK. 00:30:56.270 --> 00:31:01.250 align:middle line:84% But what's going to happen then, the inductor and the capacitor 00:31:01.250 --> 00:31:02.750 align:middle line:84% are going to ring with each other 00:31:02.750 --> 00:31:04.470 align:middle line:84% and the voltage across the transistor 00:31:04.470 --> 00:31:06.500 align:middle line:90% is going to ring down. 00:31:06.500 --> 00:31:08.040 align:middle line:84% How far is it going to ring down? 00:31:08.040 --> 00:31:09.980 align:middle line:84% That makes for a great homework problem. 00:31:09.980 --> 00:31:13.890 align:middle line:84% But it's going to ring down to V in minus 2V out. 00:31:13.890 --> 00:31:18.770 align:middle line:84% So if the output voltage V out was exactly half the input 00:31:18.770 --> 00:31:23.270 align:middle line:84% voltage, this node would ring exactly down to zero. 00:31:23.270 --> 00:31:26.433 align:middle line:84% If it was a little bit bigger, it wouldn't quite get to zero. 00:31:26.433 --> 00:31:29.100 align:middle line:84% If it was a little bit less, it would ring all the way into zero 00:31:29.100 --> 00:31:33.410 align:middle line:84% and the internal body diode of this MOSFET 00:31:33.410 --> 00:31:36.470 align:middle line:84% would turn on because don't forget there's 00:31:36.470 --> 00:31:39.910 align:middle line:90% a free diode right there. 00:31:39.910 --> 00:31:41.703 align:middle line:84% So what that means is this circuit, 00:31:41.703 --> 00:31:44.120 align:middle line:84% as long as I let the inductor current get all the way down 00:31:44.120 --> 00:31:46.160 align:middle line:84% to zero and then the diode turns off, 00:31:46.160 --> 00:31:49.130 align:middle line:84% this voltage can ring all the way down to zero 00:31:49.130 --> 00:31:53.120 align:middle line:84% and I get the opportunity to turn on the switch 00:31:53.120 --> 00:31:55.850 align:middle line:84% when he already has zero voltage across them. 00:31:55.850 --> 00:31:57.920 align:middle line:84% The resonance between the buck inductor 00:31:57.920 --> 00:32:03.440 align:middle line:84% and the capacitor at that node give me a free zero voltage 00:32:03.440 --> 00:32:05.710 align:middle line:90% switching opportunity. 00:32:05.710 --> 00:32:07.790 align:middle line:84% The current in the device is zero. 00:32:07.790 --> 00:32:09.940 align:middle line:84% The voltages naturally run down to zero. 00:32:09.940 --> 00:32:11.900 align:middle line:90% And I can turn them on for free. 00:32:11.900 --> 00:32:12.400 align:middle line:90% OK. 00:32:12.400 --> 00:32:15.940 align:middle line:84% So I've got zero voltage turn off at the switch, zero 00:32:15.940 --> 00:32:18.550 align:middle line:90% voltage turn on at the switch. 00:32:18.550 --> 00:32:19.755 align:middle line:90% Any questions about that? 00:32:19.755 --> 00:32:22.350 align:middle line:90% 00:32:22.350 --> 00:32:24.540 align:middle line:84% These are actually real waveforms. 00:32:24.540 --> 00:32:26.670 align:middle line:84% The green is the inductor current ringing 00:32:26.670 --> 00:32:28.890 align:middle line:84% up and down with this high ringing. 00:32:28.890 --> 00:32:34.260 align:middle line:84% The blue waveform is actually the drain voltage 00:32:34.260 --> 00:32:35.320 align:middle line:90% of the transistor. 00:32:35.320 --> 00:32:37.230 align:middle line:90% So you can see where it's on. 00:32:37.230 --> 00:32:38.940 align:middle line:84% This is the gate drive turning off. 00:32:38.940 --> 00:32:40.320 align:middle line:90% The voltage goes up. 00:32:40.320 --> 00:32:42.480 align:middle line:84% Then it eventually rings down so I can turn them 00:32:42.480 --> 00:32:45.456 align:middle line:90% on again for free. 00:32:45.456 --> 00:32:48.490 align:middle line:84% In this example, this was coming from 100 volts 00:32:48.490 --> 00:32:51.920 align:middle line:84% in to 35 volts out in this snapshot. 00:32:51.920 --> 00:32:54.610 align:middle line:84% And it was running at about 8 megahertz. 00:32:54.610 --> 00:32:58.832 align:middle line:84% So if you do your calculations for 100 volts input and running 00:32:58.832 --> 00:33:00.790 align:middle line:84% at 8 megahertz, you try to do that hard switch, 00:33:00.790 --> 00:33:02.123 align:middle line:90% you're going to be very unhappy. 00:33:02.123 --> 00:33:04.870 align:middle line:90% 00:33:04.870 --> 00:33:08.440 align:middle line:84% So one of the things is, by soft switching this thing, 00:33:08.440 --> 00:33:11.110 align:middle line:84% we to get to very high frequencies 00:33:11.110 --> 00:33:13.610 align:middle line:84% while maintaining high efficiency. 00:33:13.610 --> 00:33:16.990 align:middle line:84% The efficiency of that converter at that point is about 97%. 00:33:16.990 --> 00:33:18.160 align:middle line:90% Yeah. 00:33:18.160 --> 00:33:20.860 align:middle line:84% STUDENT: About how large is the parasitic capacitance 00:33:20.860 --> 00:33:22.070 align:middle line:90% across the transistor? 00:33:22.070 --> 00:33:24.452 align:middle line:84% And do you choose the transistor based 00:33:24.452 --> 00:33:25.660 align:middle line:90% on the parasitic capacitance? 00:33:25.660 --> 00:33:29.860 align:middle line:84% PROFESSOR: That's an excellent question. 00:33:29.860 --> 00:33:33.940 align:middle line:90% In this design, you have to-- 00:33:33.940 --> 00:33:35.710 align:middle line:84% the switching frequency is going to be 00:33:35.710 --> 00:33:38.470 align:middle line:84% determined by the inductance value 00:33:38.470 --> 00:33:40.670 align:middle line:90% and by the net capacitance. 00:33:40.670 --> 00:33:42.910 align:middle line:84% If all of the net capacitance is provided 00:33:42.910 --> 00:33:46.070 align:middle line:84% by the switch in the diode, then you very much care. 00:33:46.070 --> 00:33:47.810 align:middle line:84% You can't throw a random switch in there. 00:33:47.810 --> 00:33:49.210 align:middle line:84% You need one with the right capacitance 00:33:49.210 --> 00:33:51.752 align:middle line:84% that will get you at the right frequency and everything else. 00:33:51.752 --> 00:33:54.630 align:middle line:84% If you added-- you could, and people often 00:33:54.630 --> 00:33:57.938 align:middle line:84% do add external capacitance, so that then that's 00:33:57.938 --> 00:34:00.480 align:middle line:84% the controlled capacitance and the device capacitance doesn't 00:34:00.480 --> 00:34:01.590 align:middle line:90% matter as much. 00:34:01.590 --> 00:34:03.540 align:middle line:84% But when you're doing this game, yes, you 00:34:03.540 --> 00:34:05.640 align:middle line:84% care much more about what the parasitics look 00:34:05.640 --> 00:34:08.570 align:middle line:84% like because you're actually using the parasitics here. 00:34:08.570 --> 00:34:10.670 align:middle line:84% And I should say, by the way, I'm 00:34:10.670 --> 00:34:12.440 align:middle line:84% pretending that the device capacitance 00:34:12.440 --> 00:34:14.152 align:middle line:90% is a lossless capacitance. 00:34:14.152 --> 00:34:15.860 align:middle line:84% Were that it were so but you actually get 00:34:15.860 --> 00:34:18.540 align:middle line:90% some loss in that capacitance. 00:34:18.540 --> 00:34:20.420 align:middle line:84% So sometimes you would like to use 00:34:20.420 --> 00:34:24.260 align:middle line:84% a high quality external capacitor 00:34:24.260 --> 00:34:28.070 align:middle line:84% so that that ringdown is less lossy than it would otherwise 00:34:28.070 --> 00:34:28.730 align:middle line:90% be. 00:34:28.730 --> 00:34:31.429 align:middle line:84% But in this case, we pushed up the switching frequency so far 00:34:31.429 --> 00:34:34.420 align:middle line:84% that we could just do it with the device capacitance. 00:34:34.420 --> 00:34:34.920 align:middle line:90% OK. 00:34:34.920 --> 00:34:38.860 align:middle line:90% 00:34:38.860 --> 00:34:39.760 align:middle line:90% Now. 00:34:39.760 --> 00:34:42.357 align:middle line:84% The fact that this inductor is resonating with the capacitor 00:34:42.357 --> 00:34:44.440 align:middle line:84% also means you've got to be very careful about how 00:34:44.440 --> 00:34:45.710 align:middle line:90% you size that inductor. 00:34:45.710 --> 00:34:47.210 align:middle line:84% You can't pick any inductance value. 00:34:47.210 --> 00:34:48.230 align:middle line:84% You've got to pick the right one. 00:34:48.230 --> 00:34:50.350 align:middle line:84% But it turns out to be an extremely small inductor 00:34:50.350 --> 00:34:52.320 align:middle line:84% because it has a lot of ripple in it. 00:34:52.320 --> 00:34:54.040 align:middle line:90% OK. 00:34:54.040 --> 00:34:56.600 align:middle line:84% How do we control power in this converter? 00:34:56.600 --> 00:34:59.170 align:middle line:84% Well, basically, we can turn on the switch 00:34:59.170 --> 00:35:01.470 align:middle line:84% until the inductor current hits the current I want. 00:35:01.470 --> 00:35:04.090 align:middle line:84% It's like peak current control, if you 00:35:04.090 --> 00:35:06.210 align:middle line:90% will, but with high ripple. 00:35:06.210 --> 00:35:08.473 align:middle line:84% So this is a variable frequency converter. 00:35:08.473 --> 00:35:09.890 align:middle line:84% In fact, over its operating range, 00:35:09.890 --> 00:35:13.010 align:middle line:84% this converter varies between 5 and 10 megahertz. 00:35:13.010 --> 00:35:15.010 align:middle line:90% So it's not fixed frequency. 00:35:15.010 --> 00:35:17.920 align:middle line:84% So if you're going to do this game in this design, 00:35:17.920 --> 00:35:21.070 align:middle line:84% you've had to design the whole converter and its operation, 00:35:21.070 --> 00:35:24.010 align:middle line:84% its control all around getting soft switching. 00:35:24.010 --> 00:35:26.170 align:middle line:84% So it's not like I can say I want my switching 00:35:26.170 --> 00:35:29.470 align:middle line:84% frequency to be this and just go design it with PWM. 00:35:29.470 --> 00:35:30.970 align:middle line:90% No. 00:35:30.970 --> 00:35:33.460 align:middle line:84% The whole thing is designed around the soft switch 00:35:33.460 --> 00:35:36.240 align:middle line:90% operation. 00:35:36.240 --> 00:35:38.990 align:middle line:90% Any questions about that? 00:35:38.990 --> 00:35:40.110 align:middle line:90% Where might you do this? 00:35:40.110 --> 00:35:41.660 align:middle line:90% I'll show you one more example. 00:35:41.660 --> 00:35:44.190 align:middle line:84% I should say it has a limited zero voltage switching range. 00:35:44.190 --> 00:35:46.730 align:middle line:84% Technically, we need the output voltage 00:35:46.730 --> 00:35:50.710 align:middle line:84% to be greater than half the input voltage which 00:35:50.710 --> 00:35:53.393 align:middle line:84% is, you know, limits your buck conversion ratio that you 00:35:53.393 --> 00:35:54.560 align:middle line:90% can get with this converter. 00:35:54.560 --> 00:35:57.190 align:middle line:84% Maybe you can get down to a third of the input voltage 00:35:57.190 --> 00:36:00.260 align:middle line:84% before your switching losses become too big. 00:36:00.260 --> 00:36:02.690 align:middle line:84% It won't be quite soft switched but it'll be close. 00:36:02.690 --> 00:36:05.187 align:middle line:84% So there are a lot of constraints around this. 00:36:05.187 --> 00:36:06.770 align:middle line:84% What's your switching frequency range? 00:36:06.770 --> 00:36:08.645 align:middle line:84% What's your allowed voltage conversion range? 00:36:08.645 --> 00:36:10.513 align:middle line:90% What's the power range? 00:36:10.513 --> 00:36:12.430 align:middle line:84% You've put yourself under a lot of constraints 00:36:12.430 --> 00:36:15.190 align:middle line:84% that may be a normal PWM converter doesn't have, 00:36:15.190 --> 00:36:17.110 align:middle line:84% but you get to go to a high frequency 00:36:17.110 --> 00:36:21.250 align:middle line:84% and make everything small and react fast. 00:36:21.250 --> 00:36:22.820 align:middle line:90% This is another example. 00:36:22.820 --> 00:36:26.750 align:middle line:90% This is a different converter. 00:36:26.750 --> 00:36:28.320 align:middle line:90% This is a higher voltage. 00:36:28.320 --> 00:36:31.920 align:middle line:84% This is for a wide input voltage range, higher power. 00:36:31.920 --> 00:36:33.930 align:middle line:90% This is 1 to 4 megahertz. 00:36:33.930 --> 00:36:35.540 align:middle line:84% You can see that we're still getting 00:36:35.540 --> 00:36:38.660 align:middle line:84% very high efficiencies across a wide power 00:36:38.660 --> 00:36:42.920 align:middle line:90% range, so in this example 98%. 00:36:42.920 --> 00:36:44.940 align:middle line:84% This is the converter it went in. 00:36:44.940 --> 00:36:48.770 align:middle line:84% It's actually a computer power supply. 00:36:48.770 --> 00:36:53.040 align:middle line:84% So it has a rectifier and then a pair of these buck converters. 00:36:53.040 --> 00:36:54.260 align:middle line:90% You can see them here. 00:36:54.260 --> 00:36:57.880 align:middle line:84% And then there's another converter to do isolation. 00:36:57.880 --> 00:37:01.780 align:middle line:84% And these are the twice line frequency energy caps. 00:37:01.780 --> 00:37:03.430 align:middle line:84% But the beauty of doing this, getting 00:37:03.430 --> 00:37:05.110 align:middle line:84% to this very high frequency, this 00:37:05.110 --> 00:37:07.065 align:middle line:84% is one kind of soft switch converter, 00:37:07.065 --> 00:37:08.690 align:middle line:84% the resonant transition buck converter. 00:37:08.690 --> 00:37:11.530 align:middle line:84% There's another kind of soft switch converter based on what's 00:37:11.530 --> 00:37:13.510 align:middle line:90% known as a dual active bridge. 00:37:13.510 --> 00:37:17.740 align:middle line:84% But this whole converter end to end, AC in to DC out, 00:37:17.740 --> 00:37:21.080 align:middle line:90% is like almost 96% efficient. 00:37:21.080 --> 00:37:24.520 align:middle line:84% It's 80 plus platinum rated efficiency. 00:37:24.520 --> 00:37:26.980 align:middle line:84% And it's way out there in the combination 00:37:26.980 --> 00:37:29.780 align:middle line:84% of what you can get in terms of efficiency and power density. 00:37:29.780 --> 00:37:33.160 align:middle line:84% In fact, if you look at these designs that are way out here, 00:37:33.160 --> 00:37:36.010 align:middle line:84% giving you high combinations of efficiency and power density, 00:37:36.010 --> 00:37:39.580 align:middle line:84% they tend to all be soft switch designs. 00:37:39.580 --> 00:37:41.200 align:middle line:84% If you don't care about power density, 00:37:41.200 --> 00:37:42.700 align:middle line:84% you can get high efficiency but make 00:37:42.700 --> 00:37:45.630 align:middle line:84% it big by using a low switching frequency. 00:37:45.630 --> 00:37:49.420 align:middle line:84% But if you really want to make something small and efficient, 00:37:49.420 --> 00:37:51.740 align:middle line:90% this is the way to go. 00:37:51.740 --> 00:37:53.830 align:middle line:90% OK. 00:37:53.830 --> 00:37:55.930 align:middle line:84% That's the zero voltage switching example. 00:37:55.930 --> 00:37:57.430 align:middle line:84% And you can see, by the way, there's 00:37:57.430 --> 00:38:04.060 align:middle line:84% a duel between ZVS turn off and ZCS turn on and ZCS turn on 00:38:04.060 --> 00:38:06.680 align:middle line:90% and ZVS-- 00:38:06.680 --> 00:38:10.400 align:middle line:84% I'm sorry, ZCS turn off and ZVS turn on. 00:38:10.400 --> 00:38:10.900 align:middle line:90% OK. 00:38:10.900 --> 00:38:14.290 align:middle line:84% So you can do any combination of these techniques that you want. 00:38:14.290 --> 00:38:17.030 align:middle line:84% They're not all created equal, however. 00:38:17.030 --> 00:38:17.530 align:middle line:90% OK. 00:38:17.530 --> 00:38:18.030 align:middle line:90% Why? 00:38:18.030 --> 00:38:20.590 align:middle line:90% 00:38:20.590 --> 00:38:24.910 align:middle line:84% All devices have some internal parasitic capacitance. 00:38:24.910 --> 00:38:29.620 align:middle line:84% So even if I turn this switch on with zero current coming 00:38:29.620 --> 00:38:32.980 align:middle line:84% from here, if he has voltage across them, 00:38:32.980 --> 00:38:35.840 align:middle line:84% I still dissipate energy in this switch. 00:38:35.840 --> 00:38:41.170 align:middle line:84% So if I want to think about the energy associated 00:38:41.170 --> 00:38:45.000 align:middle line:84% with the switch capacitance, I really 00:38:45.000 --> 00:38:47.350 align:middle line:90% need zero voltage switching. 00:38:47.350 --> 00:38:50.110 align:middle line:84% On the other hand, if I had a switch that was in a package 00:38:50.110 --> 00:38:55.240 align:middle line:84% with a huge inductance, turning this on and off 00:38:55.240 --> 00:38:57.850 align:middle line:84% the voltage will not help me if I have energy stored 00:38:57.850 --> 00:39:00.850 align:middle line:90% in this parasitic inductance. 00:39:00.850 --> 00:39:05.000 align:middle line:84% So depending upon whether I had a lot of capacitance 00:39:05.000 --> 00:39:06.590 align:middle line:84% or I had a lot of inductance, I might 00:39:06.590 --> 00:39:10.040 align:middle line:84% think about doing one or the other parasitically. 00:39:10.040 --> 00:39:13.820 align:middle line:84% To go to really high frequencies where many people in my group 00:39:13.820 --> 00:39:17.420 align:middle line:84% design converters, we tend to focus on zero voltage switching 00:39:17.420 --> 00:39:19.340 align:middle line:84% because that lets you push up the switching 00:39:19.340 --> 00:39:21.410 align:middle line:84% frequency in the face of the device capacitance 00:39:21.410 --> 00:39:22.710 align:middle line:90% as being very important. 00:39:22.710 --> 00:39:24.920 align:middle line:90% But that's a design decision. 00:39:24.920 --> 00:39:26.540 align:middle line:84% And in fact, that's why there are 00:39:26.540 --> 00:39:27.840 align:middle line:90% so many different techniques. 00:39:27.840 --> 00:39:30.390 align:middle line:84% I'm only showing you examples of these techniques. 00:39:30.390 --> 00:39:33.980 align:middle line:84% There's a lot of different ones because different devices, 00:39:33.980 --> 00:39:35.720 align:middle line:84% different applications, and operating 00:39:35.720 --> 00:39:38.570 align:middle line:84% range all might drive you towards a different circuit that 00:39:38.570 --> 00:39:40.650 align:middle line:84% would have advantage or disadvantage. 00:39:40.650 --> 00:39:43.550 align:middle line:84% And that's part of the art of designing a high performance 00:39:43.550 --> 00:39:46.250 align:middle line:84% converter is knowing which technique to pick 00:39:46.250 --> 00:39:48.500 align:middle line:84% in what application and what topology 00:39:48.500 --> 00:39:49.900 align:middle line:90% and what control mode to use. 00:39:49.900 --> 00:39:54.790 align:middle line:90% 00:39:54.790 --> 00:39:56.300 align:middle line:90% Let's talk about this one. 00:39:56.300 --> 00:40:01.510 align:middle line:84% This is a zero current switch, quasi-resonant buck converter. 00:40:01.510 --> 00:40:05.200 align:middle line:84% This is part of a particular flavor of soft switch 00:40:05.200 --> 00:40:05.718 align:middle line:90% converters. 00:40:05.718 --> 00:40:08.260 align:middle line:84% And this paper reference here, this is actually a very famous 00:40:08.260 --> 00:40:11.530 align:middle line:90% paper from the late 1980s. 00:40:11.530 --> 00:40:13.420 align:middle line:84% And you will see commercial converters 00:40:13.420 --> 00:40:16.150 align:middle line:84% using the quasi-resonant technique, either ZVS 00:40:16.150 --> 00:40:18.500 align:middle line:84% quasi-resonant or ZCS quasi-resonant. 00:40:18.500 --> 00:40:20.290 align:middle line:84% It's not as popular as it used to be, 00:40:20.290 --> 00:40:23.860 align:middle line:84% but I thought I'd illustrate a zero current switch converter. 00:40:23.860 --> 00:40:26.860 align:middle line:84% So if I look at this converter, what do I see? 00:40:26.860 --> 00:40:30.370 align:middle line:84% If I took away Cr and Lr and I just took them out, 00:40:30.370 --> 00:40:33.450 align:middle line:84% what kind of converter would this be? 00:40:33.450 --> 00:40:36.350 align:middle line:84% It'd be a straight up buck converter, right? 00:40:36.350 --> 00:40:41.440 align:middle line:84% What we are doing is we're putting a resonant inductor 00:40:41.440 --> 00:40:43.300 align:middle line:90% in series with the switch. 00:40:43.300 --> 00:40:45.640 align:middle line:84% So if you told me nothing else about this converter 00:40:45.640 --> 00:40:47.710 align:middle line:84% and I looked at that, I'd say, boy, they're 00:40:47.710 --> 00:40:49.312 align:middle line:84% probably using zero current switching 00:40:49.312 --> 00:40:51.520 align:middle line:84% because I got this inductor in series with the switch 00:40:51.520 --> 00:40:54.220 align:middle line:90% and I don't want to-- 00:40:54.220 --> 00:40:56.320 align:middle line:84% that will let me turn on the switch 00:40:56.320 --> 00:40:58.780 align:middle line:90% and hold the current low in it. 00:40:58.780 --> 00:41:05.340 align:middle line:84% That would be ZCS turn on, this trick. 00:41:05.340 --> 00:41:10.450 align:middle line:84% And then that probably means I need ZCS turn off, too, 00:41:10.450 --> 00:41:12.390 align:middle line:84% because if I turn off the switch when 00:41:12.390 --> 00:41:16.290 align:middle line:84% the current's in that inductor, I'm going to be unhappy. 00:41:16.290 --> 00:41:17.010 align:middle line:90% OK. 00:41:17.010 --> 00:41:19.900 align:middle line:90% So how does this thing work? 00:41:19.900 --> 00:41:22.790 align:middle line:90% Suppose the diode is conducting. 00:41:22.790 --> 00:41:25.220 align:middle line:84% So this is acting like a buck converter. 00:41:25.220 --> 00:41:28.190 align:middle line:84% I'm going to commutate between the switch on and the diode on. 00:41:28.190 --> 00:41:30.130 align:middle line:84% But suppose the diode's conducting. 00:41:30.130 --> 00:41:33.680 align:middle line:84% This inductor here is just a giant inductor. 00:41:33.680 --> 00:41:36.350 align:middle line:84% So think of him as having a constant current. 00:41:36.350 --> 00:41:44.050 align:middle line:84% When I want to switch, I want to turn the switch on, fine, 00:41:44.050 --> 00:41:45.710 align:middle line:90% I just turn the switch on. 00:41:45.710 --> 00:41:48.070 align:middle line:84% He turns on under zero current because he's 00:41:48.070 --> 00:41:50.400 align:middle line:90% in series with this inductor. 00:41:50.400 --> 00:41:50.900 align:middle line:90% OK. 00:41:50.900 --> 00:41:53.030 align:middle line:84% So I get this kind of switching transition. 00:41:53.030 --> 00:41:56.030 align:middle line:84% He can turn on at low loss, at least 00:41:56.030 --> 00:42:00.330 align:middle line:90% ignoring his switch capacitance. 00:42:00.330 --> 00:42:01.090 align:middle line:90% All right. 00:42:01.090 --> 00:42:07.652 align:middle line:84% Once the switch is on, now I have V in across the inductor 00:42:07.652 --> 00:42:09.360 align:middle line:84% and the diode's conducting so I basically 00:42:09.360 --> 00:42:11.370 align:middle line:84% have V in across this resonant inductor. 00:42:11.370 --> 00:42:14.380 align:middle line:84% And the current in the inductor is going to ramp up linearly. 00:42:14.380 --> 00:42:16.450 align:middle line:90% That's this picture here. 00:42:16.450 --> 00:42:19.080 align:middle line:84% The current in this inductor is going to ramp up linearly. 00:42:19.080 --> 00:42:20.778 align:middle line:84% Eventually, the current in the inductor 00:42:20.778 --> 00:42:22.320 align:middle line:84% is going to exceed the output current 00:42:22.320 --> 00:42:26.100 align:middle line:84% and the diode is going to turn off. 00:42:26.100 --> 00:42:31.110 align:middle line:84% Once the diode turns off, basically, now Lr, 00:42:31.110 --> 00:42:33.250 align:middle line:84% though it's still at zero voltage, 00:42:33.250 --> 00:42:37.270 align:middle line:84% Lr is going to resonate with Cr here. 00:42:37.270 --> 00:42:40.035 align:middle line:84% And I'm going to have a current pulse I 00:42:40.035 --> 00:42:41.230 align:middle line:90% Lr that's going to do this. 00:42:41.230 --> 00:42:42.970 align:middle line:90% It's going to ring up. 00:42:42.970 --> 00:42:45.400 align:middle line:84% It was linearly charging and then it rings up. 00:42:45.400 --> 00:42:47.740 align:middle line:84% And then this voltage is going to get bigger 00:42:47.740 --> 00:42:50.200 align:middle line:84% than the input voltage and it's going to ring down. 00:42:50.200 --> 00:42:55.810 align:middle line:84% And eventually, the current is going to reverse and charge back 00:42:55.810 --> 00:42:58.210 align:middle line:90% this way. 00:42:58.210 --> 00:43:00.160 align:middle line:84% Once the current is charging back this way, 00:43:00.160 --> 00:43:02.590 align:middle line:84% that means it's going through this diode. 00:43:02.590 --> 00:43:05.950 align:middle line:90% I can turn the FET off. 00:43:05.950 --> 00:43:10.000 align:middle line:84% When the FET's off, once I Lr goes to zero again, 00:43:10.000 --> 00:43:12.350 align:middle line:90% everything just stays off. 00:43:12.350 --> 00:43:15.590 align:middle line:84% And during that time period, this diode 00:43:15.590 --> 00:43:17.910 align:middle line:90% is going to be conducting again. 00:43:17.910 --> 00:43:18.430 align:middle line:90% All right. 00:43:18.430 --> 00:43:20.472 align:middle line:84% So we're going to basically take a buck converter 00:43:20.472 --> 00:43:24.240 align:middle line:84% and put this Lr and Cr in it so that, basically, 00:43:24.240 --> 00:43:27.570 align:middle line:84% when I turn on the transistor, I get a resonant pulse of current 00:43:27.570 --> 00:43:30.840 align:middle line:84% and return to zero and then the diode is going to conduct again, 00:43:30.840 --> 00:43:34.140 align:middle line:90% essentially, because when this-- 00:43:34.140 --> 00:43:36.120 align:middle line:90% I should have been clear. 00:43:36.120 --> 00:43:39.330 align:middle line:84% This rings up, VCR peaks, and then VCR rings 00:43:39.330 --> 00:43:41.530 align:middle line:84% down to zero and the diode picks up the current. 00:43:41.530 --> 00:43:44.790 align:middle line:84% So what happens is I have a situation where this diode is 00:43:44.790 --> 00:43:48.540 align:middle line:84% carrying the output current except for pulses 00:43:48.540 --> 00:43:51.200 align:middle line:90% when I trip this device on. 00:43:51.200 --> 00:43:55.190 align:middle line:84% And depending upon how often I trip this device on, 00:43:55.190 --> 00:43:59.510 align:middle line:84% I get more and more pulses from the input. 00:43:59.510 --> 00:44:01.650 align:middle line:90% So this is another technique. 00:44:01.650 --> 00:44:03.800 align:middle line:84% We're going to use resonant elements introduced 00:44:03.800 --> 00:44:04.990 align:middle line:90% into the circuit. 00:44:04.990 --> 00:44:07.350 align:middle line:84% It doesn't look anything like a PWM converter. 00:44:07.350 --> 00:44:10.820 align:middle line:84% I'm getting these resonant pulses from the input. 00:44:10.820 --> 00:44:15.050 align:middle line:84% And it turns out that the average voltage at this node, 00:44:15.050 --> 00:44:17.180 align:middle line:84% which determines the average output voltage, 00:44:17.180 --> 00:44:21.020 align:middle line:84% is precisely determined by how often I 00:44:21.020 --> 00:44:24.830 align:middle line:84% trip this switch on relative to the resonant frequency. 00:44:24.830 --> 00:44:28.600 align:middle line:84% So it's actually really a frequency controlled converter. 00:44:28.600 --> 00:44:29.270 align:middle line:90% OK. 00:44:29.270 --> 00:44:31.850 align:middle line:84% If I trip the switch on a lot, I get a higher output voltage. 00:44:31.850 --> 00:44:33.392 align:middle line:84% If I trip the switch on occasionally, 00:44:33.392 --> 00:44:36.090 align:middle line:90% I get a lower output voltage. 00:44:36.090 --> 00:44:40.650 align:middle line:84% So why might I do this technique? 00:44:40.650 --> 00:44:42.990 align:middle line:84% Well, notice that the current in the switch 00:44:42.990 --> 00:44:47.670 align:middle line:90% is very smooth, very low EMI. 00:44:47.670 --> 00:44:50.250 align:middle line:90% That's kind of nice. 00:44:50.250 --> 00:44:53.700 align:middle line:84% The overlap switching loss is gone because of the zero 00:44:53.700 --> 00:44:55.650 align:middle line:84% current switching, although I still 00:44:55.650 --> 00:45:00.750 align:middle line:84% have the capacitive discharge loss in the switch. 00:45:00.750 --> 00:45:03.420 align:middle line:90% So it has some advantages. 00:45:03.420 --> 00:45:05.370 align:middle line:84% On the other hand, I can still get 00:45:05.370 --> 00:45:07.080 align:middle line:84% ringing between the device capacitance 00:45:07.080 --> 00:45:08.350 align:middle line:90% and the resonant inductance. 00:45:08.350 --> 00:45:10.510 align:middle line:84% I can still have problems with ringing. 00:45:10.510 --> 00:45:13.080 align:middle line:84% So this kind of technique is good to moderate frequencies 00:45:13.080 --> 00:45:14.620 align:middle line:90% but not super high frequencies. 00:45:14.620 --> 00:45:16.500 align:middle line:84% It might be good for a device type that 00:45:16.500 --> 00:45:19.650 align:middle line:84% likes to turn on and off with low current in it. 00:45:19.650 --> 00:45:20.150 align:middle line:90% OK. 00:45:20.150 --> 00:45:25.190 align:middle line:90% 00:45:25.190 --> 00:45:29.120 align:middle line:84% Other limits-- because I have these pulsey kind of waveforms, 00:45:29.120 --> 00:45:33.070 align:middle line:84% I tend to have a lot of reactive power transfer in this circuit. 00:45:33.070 --> 00:45:35.570 align:middle line:84% It's not like I've got square wave voltages and currents 00:45:35.570 --> 00:45:39.920 align:middle line:84% and I'm kind of optimizing how power flows through the circuit. 00:45:39.920 --> 00:45:42.188 align:middle line:84% And I also tend to have increased device stress. 00:45:42.188 --> 00:45:43.730 align:middle line:84% So this diode, instead of being rated 00:45:43.730 --> 00:45:46.010 align:middle line:84% for the input voltage like a normal buck converter, 00:45:46.010 --> 00:45:48.335 align:middle line:84% might be rated for twice the input voltage. 00:45:48.335 --> 00:45:49.460 align:middle line:90% So I've gotten some things. 00:45:49.460 --> 00:45:51.320 align:middle line:84% I've gotten some reduced switching loss, 00:45:51.320 --> 00:45:54.050 align:middle line:84% but I've had to entirely compromise everything 00:45:54.050 --> 00:45:57.858 align:middle line:90% about my converter to get it. 00:45:57.858 --> 00:45:58.900 align:middle line:90% Any questions about that? 00:45:58.900 --> 00:46:02.310 align:middle line:90% 00:46:02.310 --> 00:46:07.230 align:middle line:84% Now I would characterize that as one flavor of soft switch 00:46:07.230 --> 00:46:10.350 align:middle line:84% converters where you design the whole converter 00:46:10.350 --> 00:46:14.880 align:middle line:84% around the switching regime to get you the nice soft switched 00:46:14.880 --> 00:46:15.880 align:middle line:90% operation. 00:46:15.880 --> 00:46:18.510 align:middle line:84% So how you control it, how you pick the components, 00:46:18.510 --> 00:46:21.510 align:middle line:84% all of that stuff is 100% determined 00:46:21.510 --> 00:46:25.320 align:middle line:84% by getting soft switching and keeping soft switching where 00:46:25.320 --> 00:46:26.310 align:middle line:90% you want it. 00:46:26.310 --> 00:46:31.590 align:middle line:84% So it's just like there's all kinds of classes of soft switch 00:46:31.590 --> 00:46:33.310 align:middle line:84% converter that you can talk about. 00:46:33.310 --> 00:46:36.540 align:middle line:84% But a big category is circuits that you design the circuit 00:46:36.540 --> 00:46:41.550 align:middle line:84% around getting the magic operating waveform somehow. 00:46:41.550 --> 00:46:46.740 align:middle line:84% Another class is how can I augment a standard PWM converter 00:46:46.740 --> 00:46:49.020 align:middle line:84% with some extra circuitry that will 00:46:49.020 --> 00:46:51.720 align:middle line:84% let me run it just like a normal PWM converter, everything 00:46:51.720 --> 00:46:53.790 align:middle line:84% we've been doing all semester, but still 00:46:53.790 --> 00:46:56.070 align:middle line:84% soft switch the transistors, recover 00:46:56.070 --> 00:46:59.130 align:middle line:84% the energy in any snubbing, and still run it 00:46:59.130 --> 00:47:00.570 align:middle line:90% like a normal converter? 00:47:00.570 --> 00:47:08.020 align:middle line:84% And I'm almost out of time, but I'll point to one example. 00:47:08.020 --> 00:47:10.280 align:middle line:84% And I'll just give you a notion of this. 00:47:10.280 --> 00:47:13.450 align:middle line:84% And in fact, this circuit or a close version 00:47:13.450 --> 00:47:19.420 align:middle line:84% is described in KPVS, the snubbers and soft switching 00:47:19.420 --> 00:47:19.920 align:middle line:90% chapter. 00:47:19.920 --> 00:47:23.940 align:middle line:90% 00:47:23.940 --> 00:47:25.590 align:middle line:90% Here is the idea. 00:47:25.590 --> 00:47:34.590 align:middle line:84% If I ignored this Lr, D1, and S1, what would this circuit-- 00:47:34.590 --> 00:47:35.490 align:middle line:90% and Cr-- 00:47:35.490 --> 00:47:38.910 align:middle line:84% if I ignore these elements, what would this circuit be? 00:47:38.910 --> 00:47:40.500 align:middle line:84% Just be a standard boost converter. 00:47:40.500 --> 00:47:42.120 align:middle line:90% Right? 00:47:42.120 --> 00:47:47.400 align:middle line:84% And what we want to do is think about designing a boost 00:47:47.400 --> 00:47:48.340 align:middle line:90% converter. 00:47:48.340 --> 00:47:51.180 align:middle line:90% 00:47:51.180 --> 00:47:53.460 align:middle line:84% So here's my standard boost converter. 00:47:53.460 --> 00:47:56.850 align:middle line:84% If I worked out what the switching losses of this thing 00:47:56.850 --> 00:48:06.240 align:middle line:84% were, I have, basically, the fact 00:48:06.240 --> 00:48:10.260 align:middle line:84% that he's going to turn on and turn off loss of this guy 00:48:10.260 --> 00:48:14.460 align:middle line:84% and then diode reverse recovery is also often a big deal. 00:48:14.460 --> 00:48:15.300 align:middle line:90% OK. 00:48:15.300 --> 00:48:17.460 align:middle line:84% Well, what happens if I came along and said, 00:48:17.460 --> 00:48:21.840 align:middle line:84% look, I want to help so I'm going 00:48:21.840 --> 00:48:25.920 align:middle line:84% to go put a snubber capacitor across my device 00:48:25.920 --> 00:48:29.234 align:middle line:84% and give myself zero voltage turn off? 00:48:29.234 --> 00:48:32.550 align:middle line:84% This would still work just like a boost converter. 00:48:32.550 --> 00:48:34.410 align:middle line:90% I can turn them off. 00:48:34.410 --> 00:48:35.700 align:middle line:90% He captures the energy. 00:48:35.700 --> 00:48:38.052 align:middle line:84% And the only problem I have now is 00:48:38.052 --> 00:48:39.510 align:middle line:84% that when I turn the switch on, I'm 00:48:39.510 --> 00:48:41.954 align:middle line:84% going to dump that capacitor energy. 00:48:41.954 --> 00:48:44.270 align:middle line:84% So maybe I can come up with a circuit 00:48:44.270 --> 00:48:47.600 align:middle line:84% that just when I want to go turn the switch back on 00:48:47.600 --> 00:48:50.040 align:middle line:84% will suck the energy off that capacitor 00:48:50.040 --> 00:48:52.850 align:middle line:84% and let me turn on the switch with zero voltage. 00:48:52.850 --> 00:48:54.380 align:middle line:90% And the way they do that-- 00:48:54.380 --> 00:48:57.920 align:middle line:84% very clever design-- is they put in another little baby boost 00:48:57.920 --> 00:49:01.220 align:middle line:90% circuit like this. 00:49:01.220 --> 00:49:04.310 align:middle line:84% And what it'll do is the orange circuit doesn't do anything. 00:49:04.310 --> 00:49:06.380 align:middle line:90% It's normally off. 00:49:06.380 --> 00:49:09.590 align:middle line:84% When I turn off the switch, the capacitor 00:49:09.590 --> 00:49:10.940 align:middle line:90% gives me zero voltage switching. 00:49:10.940 --> 00:49:12.357 align:middle line:84% But other than that, it looks just 00:49:12.357 --> 00:49:13.860 align:middle line:90% like a normal PWM converter. 00:49:13.860 --> 00:49:15.650 align:middle line:84% When I get to the end of the cycle 00:49:15.650 --> 00:49:19.370 align:middle line:84% and I want to turn the switch back on, what I do is I quickly 00:49:19.370 --> 00:49:24.050 align:middle line:84% trip on the orange circuit and basically use this inductor 00:49:24.050 --> 00:49:27.220 align:middle line:84% to suck all the charge off of this capacitor 00:49:27.220 --> 00:49:31.130 align:middle line:84% and recover the energy on this capacitor, at which point 00:49:31.130 --> 00:49:35.343 align:middle line:84% I can turn on this converter with zero voltage switching-- 00:49:35.343 --> 00:49:37.510 align:middle line:84% turn on this transistor with zero voltage switching. 00:49:37.510 --> 00:49:39.180 align:middle line:84% And then the orange circuit, what 00:49:39.180 --> 00:49:42.340 align:middle line:84% I do is I turn this thing off and this inductor 00:49:42.340 --> 00:49:45.350 align:middle line:84% current goes to the output and gets delivered. 00:49:45.350 --> 00:49:45.850 align:middle line:90% OK. 00:49:45.850 --> 00:49:49.370 align:middle line:84% So the orange circuit is, essentially, his entire job, 00:49:49.370 --> 00:49:52.930 align:middle line:84% he only runs once in a little for a very short time 00:49:52.930 --> 00:49:55.910 align:middle line:84% to basically recover the energy that's on the capacitor 00:49:55.910 --> 00:49:59.770 align:middle line:84% so I can get zero voltage turn off and zero voltage turn on. 00:49:59.770 --> 00:50:04.330 align:middle line:84% This is called a zero voltage transition converter. 00:50:04.330 --> 00:50:07.540 align:middle line:84% The main converter in white is soft switched. 00:50:07.540 --> 00:50:14.830 align:middle line:84% This converter is got zero current switching. 00:50:14.830 --> 00:50:19.570 align:middle line:84% But still, this guy's still turning on and off 00:50:19.570 --> 00:50:20.360 align:middle line:90% at zero voltage. 00:50:20.360 --> 00:50:22.750 align:middle line:84% He's got very high peak pulse power. 00:50:22.750 --> 00:50:27.070 align:middle line:84% But nonetheless, as long as I use this orange circuit 00:50:27.070 --> 00:50:29.590 align:middle line:84% at the opportune time, I can run the rest of the converter 00:50:29.590 --> 00:50:31.900 align:middle line:84% just like it's a normal PWM converter. 00:50:31.900 --> 00:50:34.750 align:middle line:84% So I'm adding an auxiliary circuit 00:50:34.750 --> 00:50:36.610 align:middle line:84% comprising the green and orange circuit 00:50:36.610 --> 00:50:38.980 align:middle line:84% to let me operate the regular circuit just 00:50:38.980 --> 00:50:40.990 align:middle line:90% like a normal PWM converter. 00:50:40.990 --> 00:50:44.630 align:middle line:84% And that's a whole category of switch capacitor converters. 00:50:44.630 --> 00:50:47.190 align:middle line:84% How do I augment my normal PWM design 00:50:47.190 --> 00:50:49.600 align:middle line:84% so that it also gives me soft switching 00:50:49.600 --> 00:50:54.830 align:middle line:84% and I can get either a higher efficiency or higher frequency? 00:50:54.830 --> 00:50:57.180 align:middle line:90% Any final questions for today? 00:50:57.180 --> 00:50:57.680 align:middle line:90% Yeah. 00:50:57.680 --> 00:50:58.452 align:middle line:90% Laurie. 00:50:58.452 --> 00:51:00.785 align:middle line:84% STUDENT: What would be the losses on the orange circuit? 00:51:00.785 --> 00:51:02.300 align:middle line:90% Is that the same thing? 00:51:02.300 --> 00:51:03.690 align:middle line:90% PROFESSOR: Yes, it is. 00:51:03.690 --> 00:51:07.580 align:middle line:84% I mean, he is taking the full peak current 00:51:07.580 --> 00:51:10.950 align:middle line:84% and he's operating into the full high output voltage. 00:51:10.950 --> 00:51:15.150 align:middle line:90% And so it's not negligible. 00:51:15.150 --> 00:51:17.000 align:middle line:84% But if you do it right, it's a lot less 00:51:17.000 --> 00:51:19.080 align:middle line:84% than your original high power circuit was. 00:51:19.080 --> 00:51:21.432 align:middle line:90% So he's got high peak power. 00:51:21.432 --> 00:51:23.390 align:middle line:84% You can optimize the orange circuit differently 00:51:23.390 --> 00:51:25.070 align:middle line:90% than the white circuit. 00:51:25.070 --> 00:51:29.760 align:middle line:84% And it will overall be better but it's not negligible loss. 00:51:29.760 --> 00:51:32.280 align:middle line:84% You still have to think about it. 00:51:32.280 --> 00:51:32.780 align:middle line:90% OK. 00:51:32.780 --> 00:51:35.980 align:middle line:90% We will take this up next class. 00:51:35.980 --> 00:51:42.000 align:middle line:90%