1 00:00:00,000 --> 00:00:19,290 *36C3 preroll music* 2 00:00:19,810 --> 00:00:26,710 Herald: Okay, that was fast two minutes. The next talk is going to be on DC to DC 3 00:00:26,710 --> 00:00:39,640 converters by Zoé Bőle. She's an enthusiast for open hardware and fan of 4 00:00:39,640 --> 00:00:50,010 DIY and has been working on the topic of DC to DC converters for a long time and I 5 00:00:50,010 --> 00:00:58,900 have to keep on talking now because it seems that her computer is not really 6 00:00:58,900 --> 00:01:07,420 communicating with the presentation device. We do have a picture but we don't 7 00:01:07,420 --> 00:01:29,490 get it moving. *troubleshooting whispers in background* 8 00:01:29,490 --> 00:01:32,759 Herald: While they're still having some 9 00:01:32,759 --> 00:01:38,950 issues up here I might remind you that it is very helpful if you take your trash 10 00:01:38,950 --> 00:01:44,969 with you and now please welcome Zoë and we are ready to *inaudible*. 11 00:01:44,969 --> 00:01:47,589 Give her a warm hand! 12 00:01:54,503 --> 00:01:59,740 Zoé Bőle: Hi, my name is Zoë and this is 13 00:01:59,740 --> 00:02:05,520 the first time I'm standing here in a chaos stage so I'm a little bit, like, 14 00:02:05,520 --> 00:02:13,030 anxious but I'm here to talk to you… *applause* 15 00:02:13,030 --> 00:02:23,110 Zoé: I'm here to talk about these DC converters and the talk's called "DC/DC 16 00:02:23,110 --> 00:02:28,100 converters and everything you wanted to know about them" but it's unlikely I can 17 00:02:28,100 --> 00:02:36,260 fit everything into a 50 minute talk, so it's like not everything, but my goal is 18 00:02:36,260 --> 00:02:42,540 to provide you some starting points and give you an overview and hopefully if you 19 00:02:42,540 --> 00:02:50,150 already worked with DC/DC's then you're also not gonna be annoyed and not gonna be 20 00:02:50,150 --> 00:02:57,840 bored. Before I start with the DC/DC topic, I would like ask you to be 21 00:02:57,840 --> 00:03:04,159 excellent to each other and this is not related to my talk but I hear people 22 00:03:04,159 --> 00:03:08,939 starting clapping when someone broke the bottle accidentally and I think it's super 23 00:03:08,939 --> 00:03:18,829 not cool. Yesterday I saw someone breaking down in tears because they just broke a 24 00:03:18,829 --> 00:03:23,700 bottle and everybody was clapping and paying attention to them 25 00:03:23,700 --> 00:03:31,959 and that was like harassment. So, please don't be the one who starts clapping. But 26 00:03:31,959 --> 00:03:42,019 also I'm not here to forbid you to clap and… just know what's happening. So, 27 00:03:42,019 --> 00:03:49,010 brief introduction to DC/DC's and why: quite often you need different voltages 28 00:03:49,010 --> 00:03:56,459 than what you have available. For example you have a microcontroller or you have an 29 00:03:56,459 --> 00:04:02,260 FPGA and you work with the battery then you need to provide a different 30 00:04:02,260 --> 00:04:14,950 voltage for that circuit and the trivial solution is to just use two resistors and 31 00:04:14,950 --> 00:04:21,160 make a voltage divider, but this is totally unsuited for power delivery 32 00:04:21,160 --> 00:04:27,600 because as you start loading the output, the output voltage starts dropping. Also, 33 00:04:27,600 --> 00:04:36,930 this circuit dissipates power even if there is no useful load on the output. So 34 00:04:36,930 --> 00:04:44,400 this is only useful for signals and to have some kind of feedback and regulate 35 00:04:44,400 --> 00:04:52,389 the output to a desired level you can use an LDO, which is the same thing but you 36 00:04:52,389 --> 00:05:01,470 control, one resistor – very simplified – to always keep a desired output voltage. 37 00:05:01,470 --> 00:05:08,780 Of course this can only go lower than your input and your efficiency is limited to 38 00:05:08,780 --> 00:05:17,383 the ratio of the output voltage and the input voltage, and this is even in ideal 39 00:05:17,383 --> 00:05:26,400 situation. So, instead of burning up power in your converter you can just use 40 00:05:26,400 --> 00:05:31,740 switches and this is the idea behind switching supplies that you use a switch 41 00:05:31,740 --> 00:05:36,200 element which is either fully on or fully off. 42 00:05:36,200 --> 00:05:41,340 And if it's fully on then there is no loss on a switch and if it's 43 00:05:41,340 --> 00:05:46,300 fully off there is no current flowing through it so there is no loss either. 44 00:05:46,300 --> 00:05:51,560 There are some practical problems with this approach but 45 00:05:51,560 --> 00:06:01,639 but this works for LEDs and heaters if your switching frequency is high enough. 46 00:06:01,639 --> 00:06:08,930 To think of a DC/DC converter is a box with four terminals. It has an input 47 00:06:08,930 --> 00:06:15,750 side and an output side. Right now I'm talking about buck step-down DC/DC 48 00:06:15,750 --> 00:06:21,711 converters which are non isolated. This means the ground in the input side and the 49 00:06:21,711 --> 00:06:31,719 ground on the output side are connected together inside and this limits certain uses. 50 00:06:31,719 --> 00:06:37,940 Also you should not connect these DC/DC converters in series, so if you have 51 00:06:37,940 --> 00:06:44,070 a block like this and you think "oh, I could use two or three of them and just 52 00:06:44,070 --> 00:06:50,080 connect them in series to give it higher input voltages," that's gonna blow up very 53 00:06:50,080 --> 00:07:03,340 quickly. A block looks like this on a screen and might look like this in reality. 54 00:07:03,340 --> 00:07:13,296 Let's take a look inside. So all of these DC/DC converters consist of a power stage, 55 00:07:13,296 --> 00:07:23,620 a control system, and the feedback. The feedback is there to provide a 56 00:07:23,620 --> 00:07:31,169 regulated output regardless of the operating conditions. So what's inside a 57 00:07:31,169 --> 00:07:38,460 power stage? To have a deeper look inside we can consider this asynchronous buck 58 00:07:38,460 --> 00:07:46,099 converter where a switching element – a MOSFET – is controlled by an analog and 59 00:07:46,099 --> 00:07:53,490 digital circuitry. Feedback is provided from the output voltage and we see a diode 60 00:07:53,490 --> 00:07:59,490 in the middle which I'm going to talk about soon. You also see two capacitors on 61 00:07:59,490 --> 00:08:04,430 the input side and on the output side, which are also very important. 62 00:08:04,430 --> 00:08:07,740 More about them later. 63 00:08:07,740 --> 00:08:14,790 Let's consider the first situation: the switch is on – this is 64 00:08:14,790 --> 00:08:26,289 so-called "the on state" – and this forms a loop from the input to the output. The 65 00:08:26,289 --> 00:08:36,780 input capacitor we can neglect and in an ideal situation the output capacitor is, 66 00:08:36,780 --> 00:08:46,410 um, I will talk more about more about the output later. 67 00:08:46,410 --> 00:09:07,910 *pause* 68 00:09:07,910 --> 00:09:14,443 All right, I don't wanna make this into a lecture and everybody is sleeping in 69 00:09:14,443 --> 00:09:19,530 and the fun part will start very soon. 70 00:09:19,530 --> 00:09:26,580 So, this DC/DC has two states: either the switch is on or switch is off. Right now 71 00:09:26,580 --> 00:09:30,880 the switch is on and you see that the current can flow from the input through 72 00:09:30,880 --> 00:09:41,430 this inductor to the output. The inductor resists the change of current. It's like 73 00:09:41,430 --> 00:09:49,339 pushing a heavy mass and once it starts moving it wants to keep moving. That's why 74 00:09:49,339 --> 00:09:57,610 in this "on" state the input current flows through the inductor and starts to 75 00:09:57,610 --> 00:10:06,370 increase while it's also flowing to the output. Then the converter turns the 76 00:10:06,370 --> 00:10:14,860 switch off, which comes to the off state, and now the diode comes into play, which 77 00:10:14,860 --> 00:10:22,940 will keep the current recirculating. In this "off" state there is no current from 78 00:10:22,940 --> 00:10:29,390 the power from the source to the output, but the output is still powered from this 79 00:10:29,390 --> 00:10:40,720 decaying magnetic field through the inductor. Sometimes you hear about 80 00:10:40,720 --> 00:10:48,779 synchronous DC/DC converters where this diode is replaced by another switch. 81 00:10:48,779 --> 00:10:58,000 In that case efficiency's increased since the voltage drop across MOSFET is lower 82 00:10:58,000 --> 00:11:04,506 than the forward voltage of the diode. In this case, as you can see, current is still 83 00:11:04,506 --> 00:11:10,184 being delivered to the output. And this is the big advantage of the buck converter 84 00:11:10,184 --> 00:11:16,760 that in both an "on" and an "off" state the output is sourced with current. 85 00:11:16,760 --> 00:11:25,149 What the output capacitor does there is it provides the difference 86 00:11:25,149 --> 00:11:33,070 between the inductor current. On the lower end you can see the inductor current. 87 00:11:33,070 --> 00:11:39,320 As the switch is on it ramps up and as switch is off it ramps down, and in the middle 88 00:11:39,320 --> 00:11:48,930 you see this line which is the output current. So you see these triangles and 89 00:11:48,930 --> 00:11:57,040 this is what's provided by the output capacitor. Alright, so this is an actual 90 00:11:57,040 --> 00:12:01,621 part without the simplifications and I would like to talk a bit about the 91 00:12:01,621 --> 00:12:08,010 reference voltage and how that works. So this device creates an internal 0.7V 92 00:12:08,010 --> 00:12:18,149 reference and you can program the output voltage by choosing R1 and R2 on the 93 00:12:18,149 --> 00:12:27,360 left side so at your desired output exactly 0.7V will be at this voltage 94 00:12:27,360 --> 00:12:36,726 divider and this converter will keep regulating to reach the state. 95 00:12:45,190 --> 00:12:53,351 If you're looking for DC/DC converter to your next project then you might see 96 00:12:53,351 --> 00:12:58,152 a bunch of parameters and I'm gonna talk about those. 97 00:12:58,152 --> 00:13:04,243 So first you see a 3.3 volt 2 amp converter. What does it mean? 98 00:13:07,065 --> 00:13:12,389 This depends on how and who specifies that output 99 00:13:12,389 --> 00:13:19,110 because someone says it's two amps if it can provide two amps for a second and 100 00:13:19,110 --> 00:13:25,220 someone says it's two amps if it can continuously provide the two amps even in 101 00:13:25,220 --> 00:13:32,670 a warm environment, so it's important to talk about if it's a peak or continuous 102 00:13:32,670 --> 00:13:40,399 current rating. Then there is this so called "output ripple." You saw that 103 00:13:40,399 --> 00:13:48,760 switching action going on and off and that will create a ripple on the output voltage 104 00:13:48,760 --> 00:13:58,009 so it won't be 3.3V it will be oscillating around that. This can be as low as a few 105 00:13:58,009 --> 00:14:07,740 microvolts and as high as a few volts, depending on the parameters. Also there is 106 00:14:07,740 --> 00:14:18,040 a voltage accuracy: maybe it's labeled as 3.3V but actually it's 3.5 or 3.0. 107 00:14:18,040 --> 00:14:30,420 Load regulation: it's maybe 3.3V when it's unloaded and as you increase the output it 108 00:14:30,420 --> 00:14:37,889 starts changing the output voltage. There is the line regulation which means the 109 00:14:37,889 --> 00:14:46,570 input voltage has influence over the output, which is undesired. Then there is 110 00:14:46,570 --> 00:14:52,800 this maximum input voltage rating. Let's say this converter can tolerate seven 111 00:14:52,800 --> 00:15:00,260 volts on its input so you think "oh let's just hook it up to USB, that's 5V, right?" 112 00:15:00,260 --> 00:15:12,399 Yes, but no, because when you use cables and non-ideal conditions, you can create 113 00:15:12,399 --> 00:15:24,190 transients which overshoot the voltage possibly way above this maximum rating and 114 00:15:24,190 --> 00:15:33,180 this can lead to very nasty surprises. Because sometimes they fail short, which 115 00:15:33,180 --> 00:15:38,330 means they connect their input directly to their output. 116 00:15:38,330 --> 00:15:45,360 In this case the device you connected to the converter might also go up in flames. 117 00:15:45,360 --> 00:15:50,886 So mind the transients and always have some margin between 118 00:15:50,886 --> 00:15:57,824 your desired input voltage and the maximum the converter can tolerate. 119 00:15:57,824 --> 00:16:06,800 Then you might see 95% efficiency and that's also question at 120 00:16:06,800 --> 00:16:18,380 which load because at maximum specified load it will be lower, and at lower/less load it 121 00:16:18,380 --> 00:16:23,810 will be also lower, so there is this efficiency peak. That marketing people love 122 00:16:23,810 --> 00:16:29,170 to specify. There's also this so called "quiescent current" which means your 123 00:16:29,170 --> 00:16:38,720 converter draws current from your input even when there is nothing on its output 124 00:16:38,720 --> 00:16:45,831 and if it runs from a battery this can drain your battery in days or weeks, so 125 00:16:45,831 --> 00:16:53,910 you must pay attention to this. And there is this other factor called "switching 126 00:16:53,910 --> 00:17:00,600 frequency" so how fast, how often the internal switch changes state, but this 127 00:17:00,600 --> 00:17:06,290 might not be a constant value, especially with the previously mentioned quiescent 128 00:17:06,290 --> 00:17:16,960 current feature, the converters that excel at having a low quiescent current don't 129 00:17:16,960 --> 00:17:23,640 have fixed switching frequency, so you might have noise at different frequency 130 00:17:23,640 --> 00:17:35,040 bands and disturb your circuits or radio noise. Let's talk about a few features, 131 00:17:35,040 --> 00:17:43,130 you might want to look for. "Enable": enable functionality. This is very useful to 132 00:17:43,130 --> 00:17:50,600 easily disable your DC/DC converter and without having to interrupt either the 133 00:17:50,600 --> 00:18:01,900 input side or the output side. Let's say you have a 20 amp output converter – you 134 00:18:01,900 --> 00:18:09,080 really don't want to switch the 20 amp with a mechanical big switch. Instead 135 00:18:09,080 --> 00:18:15,650 of that you have a logic input to your DC/DC converter with which you can 136 00:18:15,650 --> 00:18:23,440 turn this completely off. Then there is so called "undervoltage lockout": you might 137 00:18:23,440 --> 00:18:31,750 want to prevent it from running below a certain input voltage to prevent draining 138 00:18:31,750 --> 00:18:40,180 your battery too deep and turning it completely off. There's "power good" that 139 00:18:40,180 --> 00:18:45,800 can provide information to your processor that the output voltage is in regulation 140 00:18:45,800 --> 00:18:51,470 and stabilized. So if you hook up the "power good" output to let's say a reset 141 00:18:51,470 --> 00:18:58,680 line or "enable", then you can be sure that the output voltage is always stable 142 00:18:58,680 --> 00:19:07,270 and your processors are not going to go into glitch. Overtemperature shutdown is 143 00:19:07,270 --> 00:19:16,050 very common these days and that makes these tiny converters almost 144 00:19:16,050 --> 00:19:22,460 indestructible because if they get too hot they just turn off completely before they 145 00:19:22,460 --> 00:19:30,770 get permanently damaged. Efficient standby: this is the so called low 146 00:19:30,770 --> 00:19:37,760 quiescent current option. That means if your output is off, your processor is 147 00:19:37,760 --> 00:19:45,750 sleeping, then it willl reduce switching action to reduce switching losses and 148 00:19:45,750 --> 00:19:51,900 might only draw a few micro amps or even nano amps. Very important for battery 149 00:19:51,900 --> 00:19:59,130 powered applications. Then you might see overcurrent protection which makes the 150 00:19:59,130 --> 00:20:04,800 output very robust. You can even make a short circuit and the overcurrent protection 151 00:20:04,800 --> 00:20:17,420 will limit the output current to this value and this prevents damaging of the 152 00:20:17,420 --> 00:20:24,440 converter and also damage to the cables and switches if they are rated to 153 00:20:24,440 --> 00:20:33,000 withstand the overcurrent protection limit. Now let's talk about noise. The 154 00:20:33,000 --> 00:20:43,740 output ripple is not exactly noise. Output ripple is there because the output 155 00:20:43,740 --> 00:20:55,160 capacitor is non-ideal and usually this this is very low on a properly designed 156 00:20:55,160 --> 00:21:02,440 converter but if you measure the output you might see spikes on the output and 157 00:21:02,440 --> 00:21:14,870 that's not ripple. That's conducted EMI because on that inductor the windings are 158 00:21:14,870 --> 00:21:21,780 coupled very closely, there is some capacitive coupling between the wires, so 159 00:21:21,780 --> 00:21:28,800 the digital on-off action from the switches will propagate to some extent to 160 00:21:28,800 --> 00:21:35,250 the output. This is attenuated by the capacitors but they cannot be completely 161 00:21:35,250 --> 00:21:42,030 filtered off and you will see the switching frequency and even upper 162 00:21:42,030 --> 00:21:50,790 harmonics of it but this can also be filtered. There is also radiated EMI, 163 00:21:50,790 --> 00:21:56,980 which comes mostly from the switching node and capacitive coupling to the ground 164 00:21:56,980 --> 00:22:05,920 plane, and also the inductor – if it's not shielded then a magnetic field can also 165 00:22:05,920 --> 00:22:15,960 radiate out and cause interference. On this picture what you see is that gray 166 00:22:15,960 --> 00:22:27,280 block, that's a shielded inductor, and the two blue connectors at the end of this PCB 167 00:22:27,280 --> 00:22:34,750 are screw terminals. I personally advise against using this style of screw 168 00:22:34,750 --> 00:22:41,700 terminals because the wires can easily slip out, make a short, or you don't 169 00:22:41,700 --> 00:22:52,900 notice that they are not connected, so I prefer a different style of connectors. 170 00:22:52,900 --> 00:23:00,350 It's good to know about non-ideal components. The capacitors that are used 171 00:23:00,350 --> 00:23:10,120 have a so called DC bias. These multi- layer ceramic capacitors are very 172 00:23:10,120 --> 00:23:17,720 sensitive to the DC voltage across the terminals and if they are rated, let's say 173 00:23:17,720 --> 00:23:24,830 20 microfarads, at the rated voltage they might lose up to 90 percent of their 174 00:23:24,830 --> 00:23:30,661 capacity. So you always have to pick a capacitor that's rated to a higher voltage 175 00:23:30,661 --> 00:23:38,330 than what your output is to compensate for this effect, and you also need to put more 176 00:23:38,330 --> 00:23:47,770 capacitors at your output than what you would think in an ideal situation. Mind 177 00:23:47,770 --> 00:23:55,001 the transients! As I said, if you plan to hotplug, connect to live wires to your 178 00:23:55,001 --> 00:24:01,290 converter, you have to keep in mind the inrush current. Those capacitors, when 179 00:24:01,290 --> 00:24:11,750 they are fully discharged and you connect that to the input, then they will try to 180 00:24:11,750 --> 00:24:21,490 charge to the input voltage as fast as the cabling lets that happen, and the 181 00:24:21,490 --> 00:24:28,980 cables have inductance which will store energy and overshoot the input voltage. 182 00:24:28,980 --> 00:24:39,350 When fiddling with MOSFETs, don't forget the ESD protection. MOSFETs are very 183 00:24:39,350 --> 00:24:49,790 sensitive at their gate, because the oxide layer is so thin that even 20V voltage is 184 00:24:49,790 --> 00:24:57,450 enough to break it down, and a 20V ESD strike is something you probably don't 185 00:24:57,450 --> 00:25:08,920 even notice, but it can damage the MOSFETs. And ever avoid the 7800 series 186 00:25:08,920 --> 00:25:17,481 LDO, because it's a very old part and I still see it in new designs, while there 187 00:25:17,481 --> 00:25:25,930 are much better ones with better regulation, less quiescent current, and 188 00:25:25,930 --> 00:25:39,990 it's also an LDO, so it's like just marginally related to DC/DCs. If 189 00:25:39,990 --> 00:25:46,300 you make your own DC/DC converters instead of buying one, you should read the 190 00:25:46,300 --> 00:25:57,670 datasheet and follow the instructions because the manufacturers give you a 191 00:25:57,670 --> 00:26:05,340 proven tested layout, which is typically good advice to follow and you should only 192 00:26:05,340 --> 00:26:25,900 deviate from that if you know what you're doing. *I'm sorry* 193 00:26:25,900 --> 00:26:48,300 *pause* Alright, that mostly concludes what I was trying 194 00:26:48,300 --> 00:26:56,810 to talk about and now it's time for your questions. 195 00:26:56,810 --> 00:27:02,073 *applause* 196 00:27:02,073 --> 00:27:09,030 Herald: Now there are two microphones one there and one over there, usually they 197 00:27:09,030 --> 00:27:18,720 are… ah, here comes the light. Are there any questions? How about the signal angel, 198 00:27:18,720 --> 00:27:25,190 does the internet have any questions? The internet doesn't have a question but 199 00:27:25,190 --> 00:27:29,110 here's one up front. Q: What would you recommend instead of 200 00:27:29,110 --> 00:27:35,650 screw terminals? Zoé: That's a very good question and that 201 00:27:35,650 --> 00:27:42,290 really depends on the application. You can have different kind of screw terminals, 202 00:27:42,290 --> 00:27:56,370 which use either crimped therminals on the cable so you have a cable shoe. 203 00:27:56,370 --> 00:28:01,640 Q: Like a ring or something? Zoé: Yes. Because then there is no way that it can 204 00:28:01,640 --> 00:28:09,610 slip out. For less current you can use dupont connectors, they can take like 205 00:28:09,610 --> 00:28:22,880 2 to 3A per contact. You know the standard pin header and that kind of thing. And 206 00:28:22,880 --> 00:28:29,800 there are also latching connectors from molex and other manufacturers. The problem 207 00:28:29,800 --> 00:28:39,310 is with that you need crimping tools and those can be very expensive. So it first 208 00:28:39,310 --> 00:28:48,540 makes sense to get those when you have a hackerspace or you can share it with other 209 00:28:48,540 --> 00:28:52,820 people. Herald: The next question, please? 210 00:28:52,820 --> 00:28:58,990 Q: Thank you for your talk. On your last slide last point you mentioned stability 211 00:28:58,990 --> 00:29:07,170 analysis. What is your experience with running such converters in parallel for 212 00:29:07,170 --> 00:29:13,650 redundancy and how would you do the analysis there? 213 00:29:13,650 --> 00:29:22,920 Zoé: Running current mode converters parallel is typically okay, but they won't 214 00:29:22,920 --> 00:29:30,250 do current sharing automatically. So this one converter has a certain output voltage 215 00:29:30,250 --> 00:29:37,720 set and the other one has little bit different voltage, and that will create a 216 00:29:37,720 --> 00:29:44,090 difference in their output currents, and there are topologies and there are 217 00:29:44,090 --> 00:29:50,620 converters which are prepared for parallel operation and they can provide current 218 00:29:50,620 --> 00:29:56,300 information to all of the parallel converters, and they can ultimately 219 00:29:56,300 --> 00:30:07,410 synchronize. For stability, that should not influence the stability of it. What I 220 00:30:07,410 --> 00:30:14,760 should have mentioned is stability analysis because we have a control loop. 221 00:30:14,760 --> 00:30:22,780 The control loop takes the output and creates a control signal that influences 222 00:30:22,780 --> 00:30:36,600 the output, but this loop has a delay, and because of this delay, basically you can 223 00:30:36,600 --> 00:30:47,230 make an oscillator of this, and to avoid that, you can use a network analyzer and 224 00:30:47,230 --> 00:30:56,860 inject a signal into the converter. Q: Thank you. 225 00:30:56,860 --> 00:31:09,170 Herald: Yeah, you go ahead, over there. Q: Hi, what would you say the choice is 226 00:31:09,170 --> 00:31:13,520 between a dis-synchronous mode or a forced-synchronous mode? 227 00:31:13,520 --> 00:31:33,010 Zoé: That's a very good question. All right so when I talked about this briefly 228 00:31:33,010 --> 00:31:39,970 and mentioned the synchronous converters, with forced synchronous converters you 229 00:31:39,970 --> 00:31:49,290 have a control switch and those have typically fixed switch frequency. If the 230 00:31:49,290 --> 00:31:55,030 output current is zero, then during half of the period current will flow backward 231 00:31:55,030 --> 00:32:01,750 from the output capacitor to the input side and then the next half period that 232 00:32:01,750 --> 00:32:06,430 current will flow back from the input to the output, so basically energy swings 233 00:32:06,430 --> 00:32:15,660 between input and output and this causes efficiency loss, but this also avoids 234 00:32:15,660 --> 00:32:30,770 operation in discontinuous mode, which reduces ripple and reduces EMI. So it 235 00:32:30,770 --> 00:32:34,850 depends on your application. Q: Thanks. 236 00:32:34,850 --> 00:32:38,700 Zoé: You're welcome. Herald: The next question? 237 00:32:38,700 --> 00:32:48,390 Q: Hi, Zoé, thank you for the talk! I have a question about: you mentioned linear 238 00:32:48,390 --> 00:32:54,220 regulators at the end, what are they used for in this context? 239 00:32:54,220 --> 00:33:02,010 Zoé: you mean 7800 series? Q: Yes. 240 00:33:02,010 --> 00:33:13,910 Q: Not, the one before, I think. Zoé: Those were very good regulators in 241 00:33:13,910 --> 00:33:22,130 the 70s and those are linear regulators, and the problem with the 7800 series is 242 00:33:22,130 --> 00:33:28,540 everybody knows about them because books are full of them but they have quite a few 243 00:33:28,540 --> 00:33:41,270 milli amps of quiescent current. They also have bad regulation against load and line 244 00:33:41,270 --> 00:33:48,140 transients and they are not cheaper than much of their alternatives, so there's 245 00:33:48,140 --> 00:33:56,020 really no reason to use those. You can use for example a DC/DC pre-regulator and then 246 00:33:56,020 --> 00:33:59,590 an LDO afterward to smooth out the voltage. 247 00:33:59,590 --> 00:34:07,204 Q: Okay, thank you. Herald: Go ahead 248 00:34:07,204 --> 00:34:13,490 Q: Thank you very much! My question is, you mentioned the noise coupled via the 249 00:34:13,490 --> 00:34:18,440 inductor to the output. Which sort of filter do you recommend: differential 250 00:34:18,440 --> 00:34:24,060 noise or common mode noise, and input or output? Which is most important from your 251 00:34:24,060 --> 00:34:32,040 perspective? Zoé: So lots of the noise goes actually 252 00:34:32,040 --> 00:34:40,460 back to the input supply and I said that in an ideal circuit the input capacitor is 253 00:34:40,460 --> 00:34:46,970 not necessary, but in a real circuit the input capacitor is critical because the 254 00:34:46,970 --> 00:35:03,520 input inductance is seen by by the switch. If you let me show you. On this chart you 255 00:35:03,520 --> 00:35:11,660 see the inductor current and the input current, it follows the inductor current 256 00:35:11,660 --> 00:35:17,720 only during the on phase which means after the end of the on phase and beginning of 257 00:35:17,720 --> 00:35:25,520 the off phase it falls from maximum value to zero and later on at the end of the off 258 00:35:25,520 --> 00:35:32,460 phase and the beginning of the on phase the current jumps from zero to the output 259 00:35:32,460 --> 00:35:42,380 current, and these jumps in the supply current create an awful lot of EMI if the 260 00:35:42,380 --> 00:35:51,610 input capacitor is not large enough so this is a very critical thing. I saw quite 261 00:35:51,610 --> 00:35:56,810 a few converters where the input capacitor is under dimensioned and when you run it 262 00:35:56,810 --> 00:36:03,370 over longer wires with more parasitic inductance, that can create a lot of EMI. 263 00:36:03,370 --> 00:36:12,090 For ways of reducing the the noise on the output, the best way is to have proper 264 00:36:12,090 --> 00:36:20,710 filtering capacitors. If you use ceramic capacitors and enough high enough value 265 00:36:20,710 --> 00:36:33,670 you can get rid of almost all of the noise. I made a design which had microvolt 266 00:36:33,670 --> 00:36:43,610 noise because I found a capacitor with its resonance frequency exactly at the 267 00:36:43,610 --> 00:36:48,610 switching frequency, so basically all that noise that was coming from switching 268 00:36:48,610 --> 00:36:55,260 action was reflected away and higher frequency ranges where it got filtered 269 00:36:55,260 --> 00:37:05,270 dissipated much faster. You can use PI filters at the output but mind 270 00:37:05,270 --> 00:37:16,890 that you worsen the transient behavior of your converters. So if your load suddenly 271 00:37:16,890 --> 00:37:23,060 needs a lot more power and starts drawing more current, then your converter will 272 00:37:23,060 --> 00:37:35,940 react slower because of the filter you just added. PI filters or RC 273 00:37:35,940 --> 00:37:38,890 filters if you don't need that much current. 274 00:37:38,890 --> 00:37:44,620 Q: Okay, thanks. Herald: Okay great I don't see any more 275 00:37:44,620 --> 00:37:52,540 questions, so everything seems to be fully explained. Thank you and give her a 276 00:37:52,540 --> 00:37:55,220 applause and good night. 277 00:37:55,220 --> 00:37:56,670 *applause* 278 00:37:56,670 --> 00:38:02,960 Zoé: Thank you 279 00:38:02,960 --> 00:38:06,840 *postroll music* 280 00:38:06,840 --> 00:38:30,000 Subtitles created by c3subtitles.de in the year 2020. Join, and help us!