/ News & Press / Video / How to choose the right capacitor type for a circuit! _ Film vs. Ceramic vs. Electrolytic
How to choose the right capacitor type for a circuit! _ Film vs. Ceramic vs. Electrolytic
WEBVTT Kind: captions Language: en
00:00:00.820 --> 00:00:02.960 Recently, I've been playing around 00:00:02.960 --> 00:00:05.280 with some high power LEDs. 00:00:05.280 --> 00:00:07.320 To efficiently dim the brightness, 00:00:07.320 --> 00:00:09.920 I built up this simple test circuit, 00:00:09.920 --> 00:00:11.900 which features a function generator 00:00:11.900 --> 00:00:15.280 to create an adjustable PWM signal, 00:00:15.280 --> 00:00:18.720 and n-channel MOSFET in series to the LED, 00:00:18.720 --> 00:00:21.820 to actually turn it on and off rapidly, 00:00:21.820 --> 00:00:25.620 and a TC4420 MOSFET driver IC 00:00:25.620 --> 00:00:29.080 to charge / discharge the power MOSFETs gates 00:00:29.080 --> 00:00:31.460 as quickly as possible. 00:00:31.460 --> 00:00:33.540 Now in the low frequency range 00:00:33.540 --> 00:00:36.600 This circuits dims the LED perfectly fine 00:00:36.600 --> 00:00:40.780 by changing the duty cycle of the PWM signal. 00:00:40.780 --> 00:00:45.000 But while for example using a frequency of 100 kHz 00:00:45.000 --> 00:00:47.400 and a duty cycle of 1%, 00:00:47.400 --> 00:00:50.120 The circuit works for a couple of minutes, 00:00:50.120 --> 00:00:52.360 but then randomly stops working. 00:00:52.360 --> 00:00:54.520 Because the MOSFET driver IC 00:00:54.520 --> 00:00:56.500 apparently destroyed itself. 00:00:57.020 --> 00:01:01.160 After replacing it. This circuit worked fine once again 00:01:01.160 --> 00:01:05.240 but this time I examined the pins voltages of the IC 00:01:05.240 --> 00:01:06.640 with my oscilloscope, 00:01:06.640 --> 00:01:09.440 to determine the culprits. 00:01:09.440 --> 00:01:13.220 And while probing the supply voltage pin of the IC, 00:01:13.220 --> 00:01:17.520 I noticed that there occurred 100 kHz oscillations 00:01:17.520 --> 00:01:21.220 with peak voltages of 28V and 2V 00:01:21.220 --> 00:01:23.400 Since that is partly beyond the ICs 00:01:23.400 --> 00:01:25.220 maximum supply voltage. 00:01:25.220 --> 00:01:29.420 It is no wonder that it self-destructs after a while. 00:01:29.420 --> 00:01:31.060 To solve this problem, 00:01:31.060 --> 00:01:34.800 The Würth Elektronik eiSos Group recently sent me 00:01:34.800 --> 00:01:37.280 three of their capacitor design kits. 00:01:37.280 --> 00:01:41.420 The General-Purpose DC Film Capacitors Design Kit 00:01:41.420 --> 00:01:45.100 The Multi-Layer Ceramic Chip Capacitors Design Kit 00:01:45.100 --> 00:01:49.420 and The Aluminum Electrolytic Capacitors Design Kit. 00:01:49.420 --> 00:01:50.880 So in this video, 00:01:50.880 --> 00:01:54.360 let's solve this mysterious IC supply voltage problem, 00:01:54.360 --> 00:01:55.560 and learn the difference 00:01:55.560 --> 00:01:57.960 between those three capacitor types, 00:01:57.960 --> 00:02:01.520 to find out which one you should use for which circuit. 00:02:01.960 --> 00:02:02.460 LET’S 00:02:02.460 --> 00:02:03.400 GET STARTED! 00:02:15.820 --> 00:02:17.500 This video is sponsored 00:02:17.500 --> 00:02:19.780 by The Würth Elektronik eiSos Group. 00:02:21.660 --> 00:02:26.020 Let's start off with our MOSFET driver IC problem. 00:02:26.020 --> 00:02:28.120 The supply voltage breaks down 00:02:28.120 --> 00:02:31.320 and afterwards an oscillation occurs. 00:02:31.320 --> 00:02:35.180 This happens with a frequency of 100 kHz. 00:02:35.180 --> 00:02:36.980 Which not coincidentally 00:02:36.980 --> 00:02:39.545 is the exact moment the power MOSFET gates 00:02:39.545 --> 00:02:41.520 get charged up. 00:02:41.520 --> 00:02:43.360 So, if we break it down 00:02:43.360 --> 00:02:45.600 the input signal (Vᴅᴅ) gets put high, 00:02:45.600 --> 00:02:47.480 which ultimately connects the gate 00:02:47.480 --> 00:02:50.940 of our power MOSFETs to the supply voltage. 00:02:50.940 --> 00:02:54.340 This action requires current for our IC. 00:02:54.340 --> 00:02:57.040 In order to power its own components, 00:02:57.040 --> 00:02:59.980 and ultimately charge up its own MOSFET gates 00:02:59.980 --> 00:03:02.180 00:03:02.180 --> 00:03:06.160 While observing this IC current through 1 ohm shunt. 00:03:06.160 --> 00:03:10.540 I noticed that it reached its first peak value of around 2A 00:03:10.540 --> 00:03:13.580 within only 15 nanoseconds. 00:03:13.580 --> 00:03:16.440 The only problem is that my power supply, 00:03:16.440 --> 00:03:18.700 due to its internal construction, 00:03:18.700 --> 00:03:21.900 is not the fastest acting energy source. 00:03:21.900 --> 00:03:25.100 That is why we can model its output impedance 00:03:25.100 --> 00:03:29.300 as a small resistor in series with an inductor. 00:03:29.300 --> 00:03:33.020 Now if the IC would require a constant 1A, 00:03:33.020 --> 00:03:34.980 we would only get a small voltage drop 00:03:34.980 --> 00:03:36.380 across the resistor, 00:03:36.380 --> 00:03:38.340 but no other problems! 00:03:38.340 --> 00:03:40.200 Since an inductor voltage drop 00:03:40.200 --> 00:03:43.420 only exists with a change in current flow. 00:03:43.420 --> 00:03:49.000 But since our IC wants to have 2A in a time of only 50ns, 00:03:49.000 --> 00:03:52.340 Our inductor now features a big voltage drop, 00:03:52.340 --> 00:03:54.280 which means we got a breakdown 00:03:54.280 --> 00:03:56.780 in the supply voltage of our IC. 00:03:57.300 --> 00:03:59.900 Combine that with a breadboard construction 00:03:59.900 --> 00:04:03.580 which comes with noticeable parasitic capacitances, 00:04:03.580 --> 00:04:06.000 and we got ourself a small oscillator 00:04:06.000 --> 00:04:07.580 on the supply voltage pin, 00:04:07.580 --> 00:04:09.920 that leads to problems. 00:04:09.920 --> 00:04:12.860 To solve that, we can add a capacitor 00:04:12.860 --> 00:04:15.460 in parallel to the supply voltage pin. 00:04:15.460 --> 00:04:18.700 Which is then often referred to as a bypass 00:04:18.700 --> 00:04:20.860 or decoupling capacitor. 00:04:20.860 --> 00:04:23.220 It's job is to basically provide 00:04:23.220 --> 00:04:25.620 the high current search for the IC, 00:04:25.620 --> 00:04:28.020 which the mains power supply can not offer 00:04:28.020 --> 00:04:29.920 because it is too slow. 00:04:29.920 --> 00:04:32.460 And thus it also suppresses noise 00:04:32.460 --> 00:04:35.240 for other ICs in the circuits. 00:04:35.240 --> 00:04:36.560 The only question is: 00:04:36.560 --> 00:04:40.320 “What capacitor type is best suited for this job?” 00:04:40.320 --> 00:04:43.440 The two main ratings, you usually see on them 00:04:43.440 --> 00:04:47.080 is their capacitance and their withstand voltage. 00:04:47.080 --> 00:04:49.840 Now since all of my capacitor voltage ratings 00:04:49.840 --> 00:04:52.280 are higher than the 12V I'm using, 00:04:52.280 --> 00:04:55.420 we should go for the highest capacitance rating. 00:04:55.420 --> 00:04:56.920 Right? 00:04:56.920 --> 00:05:00.480 I mean, since the capacitance rating is proportional 00:05:00.480 --> 00:05:02.720 to the stored energy of the capacitor, 00:05:02.720 --> 00:05:04.280 we should definitely be able 00:05:04.280 --> 00:05:07.140 to provide enough current with it. 00:05:07.140 --> 00:05:11.880 So I connected the 15,000μF electrolytic capacitor 00:05:11.880 --> 00:05:14.220 in parallel to the IC. 00:05:14.220 --> 00:05:16.560 And asserted that the oscillation peaks 00:05:16.560 --> 00:05:19.260 decreased to 16V and 8V. 00:05:19.260 --> 00:05:21.260 Seems decent. 00:05:21.260 --> 00:05:22.920 Out of curiosity though. 00:05:22.920 --> 00:05:27.760 I also tried out a small 150µF film capacitor 00:05:27.760 --> 00:05:29.680 as a decoupling capacitor, 00:05:29.680 --> 00:05:31.420 which worked even better! 00:05:31.420 --> 00:05:35.740 By decreasing the peaks to 13V and 10V 00:05:35.740 --> 00:05:38.460 But, why does such a puny small film capacitor 00:05:38.460 --> 00:05:41.880 whose capacity is 100,000 times smaller 00:05:41.880 --> 00:05:45.900 than the beefy electrolytic capacitor works better? 00:05:45.900 --> 00:05:49.140 Well, the reason is that while all capacitors 00:05:49.140 --> 00:05:51.220 share the same basic structure 00:05:51.220 --> 00:05:53.920 which means they got two metal electrodes, 00:05:53.920 --> 00:05:57.000 which are separated by a non conductive material 00:05:57.000 --> 00:05:58.820 called the Dielectric, 00:05:58.820 --> 00:06:01.660 in order to create an electric fields 00:06:01.660 --> 00:06:04.700 and the store energy when a voltage is applied, 00:06:04.700 --> 00:06:07.320 their materials all differ. 00:06:07.320 --> 00:06:10.300 My electrolytic capacitors for example, 00:06:10.300 --> 00:06:14.300 use aluminum foil in combination with an electrolytes. 00:06:14.300 --> 00:06:17.860 While my film capacitors use polypropylene 00:06:17.860 --> 00:06:21.540 and my ceramic capacitors use ... like the name implies 00:06:21.540 --> 00:06:23.240 Ceramic. 00:06:23.240 --> 00:06:27.000 This material choice influences electrical properties 00:06:27.000 --> 00:06:29.500 like the voltage or capacitance. 00:06:29.500 --> 00:06:33.080 But also other properties like for example, 00:06:33.080 --> 00:06:34.580 The expected lifetime 00:06:34.580 --> 00:06:37.700 or whether a capacitor is flammable 00:06:37.700 --> 00:06:39.740 But there are more hidden properties 00:06:39.740 --> 00:06:42.940 which we can discover by examining the capacitors 00:06:42.940 --> 00:06:43.680 with an LCR meter. 00:06:43.680 --> 00:06:45.280 (L: Inductance C: Capacitance R: Resistance) 00:06:45.280 --> 00:06:48.720 Sadly though the 15,000µF one 00:06:48.720 --> 00:06:50.880 overloaded the meter. 00:06:50.880 --> 00:06:52.480 But as a replacement, 00:06:52.480 --> 00:06:54.500 I used a 10µF one 00:06:54.500 --> 00:06:58.540 which works similarly as a decoupling capacitor. 00:06:58.540 --> 00:07:00.200 The first thing we notice is that 00:07:00.200 --> 00:07:03.720 the capacitor not only features a capacitance 00:07:03.720 --> 00:07:07.260 but also a resistance and inductance 00:07:07.260 --> 00:07:11.020 Those are called Equivalent Series Resistance 00:07:11.020 --> 00:07:12.380 (ESR) 00:07:12.380 --> 00:07:15.280 and Equivalent Series Inductance. 00:07:15.280 --> 00:07:16.400 (ESL) 00:07:16.400 --> 00:07:19.280 And they do exist in a practical capacitor 00:07:19.280 --> 00:07:21.740 due to its internal structure. 00:07:21.740 --> 00:07:24.200 The big problem with that though is that 00:07:24.200 --> 00:07:27.520 the parasitic resistance creates a power loss. 00:07:27.520 --> 00:07:31.440 As an example, we can use the 100 Hz measurement 00:07:31.440 --> 00:07:33.980 of the LCR meter to determine 00:07:33.980 --> 00:07:38.300 a dissipation factor of 0.097. 00:07:38.300 --> 00:07:41.140 The dissipation factor describes the relation 00:07:41.140 --> 00:07:43.820 between the ESR and the capacitive 00:07:43.820 --> 00:07:45.560 and inductive reactance. 00:07:45.560 --> 00:07:49.260 But let's neglect the inductive one for now. 00:07:49.260 --> 00:07:53.000 That means the overall impedance of our capacitor 00:07:53.000 --> 00:07:56.220 acts around 92% like a capacitor 00:07:56.220 --> 00:07:58.840 and 8% like a resistor. 00:07:58.840 --> 00:08:01.860 Which on the other hand means we waste energy 00:08:01.860 --> 00:08:06.020 that goes in and out of the capacitor as heat 00:08:06.020 --> 00:08:09.000 If we increase the frequency to one kilohertz 00:08:09.000 --> 00:08:12.060 We can see how the dissipation factor increases 00:08:12.060 --> 00:08:14.240 to 0.220 00:08:14.240 --> 00:08:16.840 which means the capacitor now features 00:08:16.840 --> 00:08:19.840 an even bigger resistive components. 00:08:19.840 --> 00:08:23.980 With rising frequency this DF value increases 00:08:23.980 --> 00:08:27.340 because the dielectric ohmic value increases 00:08:27.340 --> 00:08:29.920 while the capacitive reactance decreases 00:08:29.920 --> 00:08:32.200 with rising frequency. 00:08:32.200 --> 00:08:34.060 It gets especially interesting 00:08:34.060 --> 00:08:37.920 when the capacitive reactance = the inductive reactance 00:08:37.920 --> 00:08:39.460 of the ESL 00:08:39.460 --> 00:08:42.360 which happens at the self resonant frequency 00:08:42.360 --> 00:08:44.260 of the capacitor. 00:08:44.260 --> 00:08:46.060 Above this frequency, 00:08:46.060 --> 00:08:48.560 the capacitor acts more like an inductor 00:08:48.560 --> 00:08:50.240 than a capacitor. 00:08:50.240 --> 00:08:52.240 And thus, It’s not interesting for us 00:08:52.240 --> 00:08:54.680 when it comes to decoupling. 00:08:54.680 --> 00:08:58.000 Even the data sheets of the electrolytic capacitor 00:08:58.000 --> 00:09:03.500 gives us a dissipation factor of 16% at 120 Hz 00:09:03.500 --> 00:09:06.260 which means such electrolytic capacitors 00:09:06.260 --> 00:09:09.300 are better suited for ULF applications 00:09:09.300 --> 00:09:10.440 (ULF: Ultra Low Frequency) are better suited for ULF applications. 00:09:10.440 --> 00:09:14.880 But if we insert the 150µF film capacitor 00:09:14.880 --> 00:09:16.740 into the LCR meter. 00:09:16.740 --> 00:09:20.660 We can see that its dissipation factor is pretty much 0 00:09:20.660 --> 00:09:23.280 at 100 Hz and 1 kHz and 00:09:23.280 --> 00:09:27.140 Only goes up to around 0.001 00:09:27.140 --> 00:09:31.560 So 0.1% at 10 kHz 00:09:31.560 --> 00:09:33.600 The datasheet of the capacitor 00:09:33.600 --> 00:09:36.200 pretty much confirms those values. 00:09:36.200 --> 00:09:41.840 By giving a DF of only 0.26% at 100 kHz 00:09:41.840 --> 00:09:43.840 Meaning such film capacitors 00:09:43.840 --> 00:09:47.300 have a very low ESL and ESR rating 00:09:47.300 --> 00:09:50.260 and thus a high self resonant frequency. 00:09:50.260 --> 00:09:52.180 Which makes them suitable for LF & MF applications 00:09:52.180 --> 00:09:52.700 (LF: Low Frequency) 00:09:52.700 --> 00:09:54.200 (LF: Low Frequency MF: Medium Frequency) 00:09:54.200 --> 00:09:56.800 like our decoupling task. 00:09:56.800 --> 00:10:00.400 But we should not forget about our super tiny ceramic 00:10:00.400 --> 00:10:02.060 SMD capacitors. 00:10:02.060 --> 00:10:05.660 For which there apparently exists different classes 00:10:05.660 --> 00:10:09.520 like NP0 and X7R 00:10:09.520 --> 00:10:12.500 In a nutshell those two kinds feature a different 00:10:12.500 --> 00:10:13.880 base material. 00:10:13.880 --> 00:10:17.660 Which has the effect that class want ceramic capacitors 00:10:17.660 --> 00:10:20.440 like the NP0 are very stable 00:10:20.440 --> 00:10:22.580 over a wide temperature range 00:10:22.580 --> 00:10:26.800 while class two ceramic capacitors like the X7R 00:10:26.800 --> 00:10:29.940 are not as stable over a wide temperature range 00:10:29.940 --> 00:10:34.520 but feature way higher voltage dependent capacitances. 00:10:34.520 --> 00:10:37.540 That makes class 1 ceramic capacitors perfect 00:10:37.540 --> 00:10:39.620 for something like oscillators 00:10:39.620 --> 00:10:42.780 while class two ones could be used for decoupling. 00:10:42.780 --> 00:10:44.400 00:10:44.400 --> 00:10:48.120 To find that outs, I grabbed the 10µF one 00:10:48.120 --> 00:10:51.100 and checked it with my LCR meter. 00:10:51.100 --> 00:10:55.900 At 1 kHz, We got a dissipation factor of around 3% 00:10:55.900 --> 00:10:59.360 and at 10 kHz around 15% . 00:10:59.360 --> 00:11:02.140 So not as low as the film capacitor. 00:11:02.140 --> 00:11:05.700 But after soldering it to a THT breakout boards 00:11:05.700 --> 00:11:08.700 and connecting it to my MOSFET driver IC. 00:11:08.700 --> 00:11:11.880 It reduced the oscillation to better values 00:11:11.880 --> 00:11:15.360 than what the electrolytic capacitor offered. 00:11:15.360 --> 00:11:18.180 Now, of course a capacitor datasheet 00:11:18.180 --> 00:11:19.720 depending on its type 00:11:19.720 --> 00:11:22.180 can give us even more information 00:11:22.180 --> 00:11:24.380 like the insulation resistance 00:11:24.380 --> 00:11:26.360 which basically sits in parallel 00:11:26.360 --> 00:11:30.060 to the actual capacitance or the leakage current. 00:11:30.060 --> 00:11:33.460 Whose name pretty much speaks for itself. 00:11:33.460 --> 00:11:35.380 But you should now understand that 00:11:35.380 --> 00:11:37.180 while electrolytic capacitors 00:11:37.180 --> 00:11:39.300 can be used for buffering energy 00:11:39.300 --> 00:11:42.600 which is why you see them often in power supplies 00:11:42.600 --> 00:11:44.880 they are generally not well suited 00:11:44.880 --> 00:11:47.820 for higher frequency filters or decoupling. 00:11:48.519 --> 00:11:50.460 And if you want more information 00:11:50.460 --> 00:11:53.180 about other applications of capacitors 00:11:53.180 --> 00:11:56.720 and the usage of different capacitor types in general, 00:11:56.720 --> 00:11:59.000 then I highly recommend having a look 00:11:59.000 --> 00:12:02.640 at the webinar of The Würth Electronik eiSos Group 00:12:02.640 --> 00:12:06.080 which you can find in the video description. 00:12:06.080 --> 00:12:08.140 As always, thanks for watching 00:12:08.140 --> 00:12:10.620 Don't forget to like share and subscribe. 00:12:11.240 --> 00:12:11.740 STAY 00:12:11.740 --> 00:12:12.700 CREATIVE 00:12:12.700 --> 00:12:13.500 AND I’LL 00:12:13.500 --> 00:12:14.280 SEE YOU 00:12:14.280 --> 00:12:15.640 NEXT TIME! 00:12:15.640 --> 00:12:17.640 (As alway, Subtitle by PolaX3)
Office location
Engineering company LOTUS®
Russia, Ekaterinburg, Lunacharskogo street, 240/12