/ News & Press / Video / SDG #066 MLCC Capacitors Values Not What You Think! Tests with JLCPCB
SDG #066 MLCC Capacitors Values Not What You Think! Tests with JLCPCB
WEBVTT Kind: captions Language: en
00:00:00.290 hi in this video we're going to be 00:00:02.60000:00:02.610 taking a look at ceramic capacitors and 00:00:04.70000:00:04.710 why once you've deployed them in your 00:00:06.44000:00:06.450 circuit the capacitance value of those 00:00:09.02000:00:09.030 capacitors might not be what was 00:00:10.66900:00:10.679 specified in the datasheet so this might 00:00:13.40000:00:13.410 be already common knowledge to some of 00:00:15.28900:00:15.299 you and obviously some of you may have 00:00:16.97000:00:16.980 never come across this phenomenon but 00:00:18.74000:00:18.750 basically what happens is once you've 00:00:20.81000:00:20.820 applied a DC voltage to a Class two 00:00:23.12000:00:23.130 ceramic capacitor the capacitance value 00:00:25.79000:00:25.800 will be less than what was originally 00:00:27.10900:00:27.119 specified in the datasheet so this is 00:00:30.08000:00:30.090 particularly class 2 capacitors so 00:00:32.47900:00:32.489 things like x7r y5e those type of 00:00:35.84000:00:35.850 dielectric material ceramic capacitors 00:00:38.18000:00:38.190 and what I've done is I've built up a 00:00:40.52000:00:40.530 little circuit so that we can have a 00:00:42.04900:00:42.059 look at the effect and how different 00:00:43.97000:00:43.980 ceramic capacitors are affected in a 00:00:46.43000:00:46.440 different way right so here's a 00:00:48.95000:00:48.960 schematic for what I've implemented on 00:00:50.66000:00:50.670 the PCB and basically we've got an RC 00:00:52.91000:00:52.920 oscillator based around an op-amp so we 00:00:56.81000:00:56.820 set our reference voltage 0.7 volts with 00:00:59.09000:00:59.100 this diode here and then we've got an 00:01:01.22000:01:01.230 op-amp setup as a comparator and as the 00:01:04.67000:01:04.680 output increases to 5 volts then we 00:01:06.50000:01:06.510 start charging the capacitor and then 00:01:08.66000:01:08.670 once we reach a threshold then we set 00:01:10.82000:01:10.830 the output to zero volts and that starts 00:01:13.07000:01:13.080 discharging the capacitor so we see a 00:01:14.71900:01:14.729 waveform of charging and discharging the 00:01:16.91000:01:16.920 capacitor and the output looks like a 00:01:19.31000:01:19.320 square wave pulse train then we've got a 00:01:22.91000:01:22.920 little bit of circuitry here which gives 00:01:24.35000:01:24.360 us a sawtooth and then a filter here 00:01:27.83000:01:27.840 which gives us an output that's 00:01:29.03000:01:29.040 proportional to the oscillation 00:01:30.74000:01:30.750 frequency and that's what I've 00:01:32.48000:01:32.490 implemented on this PCB and this PCB is 00:01:35.71900:01:35.729 one made by a jail C PCB and this is the 00:01:38.87000:01:38.880 first red PCB that I've had made and 00:01:40.88000:01:40.890 actually the solder mask gives really 00:01:42.85900:01:42.869 good coverage on this particular board 00:01:44.56900:01:44.579 but yeah here's our bias voltage that we 00:01:46.99900:01:47.009 apply here these are active pastors 00:01:49.34000:01:49.350 under test obviously we only test one at 00:01:51.23000:01:51.240 a time but I laid out a few different 00:01:52.81900:01:52.829 footprints and then here's our DC output 00:01:55.55000:01:55.560 which gives the proportional voltage 00:01:57.52000:01:57.530 based on what the capacitance value is 00:02:01.09000:02:01.100 all right so here's our circuit all 00:02:03.13900:02:03.149 connected open what we want to do is 00:02:04.55000:02:04.560 tweak the potentiometer so it reads one 00:02:06.64900:02:06.659 ball exactly and so with zero bias we're 00:02:11.18000:02:11.190 reading a value of one so that means our 00:02:13.01000:02:13.020 term doesn't win 00:02:13.79000:02:13.800 nanofied caster is at 220 nanofarads now 00:02:18.32000:02:18.330 if we increase the bias to 5 volts you 00:02:22.18000:02:22.190 can see that this has now dropped to 00:02:24.71000:02:24.720 about 0.7 4 volts so that means our 220 00:02:29.75000:02:29.760 nanofied capacitor is now 162 nano cards 00:02:35.84000:02:35.850 and if we increase that to 10 volts its 00:02:43.54000:02:43.550 0.45 so 220 times 0.45 we're at 99 nano 00:02:49.34000:02:49.350 farad's and if we do it up to its 00:02:51.41000:02:51.420 maximum work in voltage of 16 volts 00:02:59.38000:02:59.390 we're at 0.3 so to 20 times 0.3 out of 00:03:05.09000:03:05.100 220 nano farad capacitor is now only 66 00:03:08.21000:03:08.220 nanofarads so you can see that if you 00:03:10.55000:03:10.560 were to use this in your circuit and you 00:03:12.20000:03:12.210 were relying on that value specifically 00:03:14.59000:03:14.600 then it wouldn't be behaving exactly as 00:03:18.19900:03:18.209 you intended so what I've done is I've 00:03:20.57000:03:20.580 tested a whole range of different 00:03:21.77000:03:21.780 capacitors and then logged all of the 00:03:24.05000:03:24.060 whistles first of all we've got the y 5v 00:03:27.16000:03:27.170 dielectric capacitors they're all 50 00:03:29.42000:03:29.430 volt 100 nano farad capacitors the only 00:03:32.48000:03:32.490 difference here is the case sizes but 00:03:34.82000:03:34.830 you can see here they all start off at 00:03:36.23000:03:36.240 their specified capacitance but by the 00:03:38.66000:03:38.670 time they've reached their maximum 00:03:39.86000:03:39.870 working voltage they're only at 20% of 00:03:42.41000:03:42.420 their original capacitance value and one 00:03:44.81000:03:44.820 thing that you can note here is that the 00:03:46.31000:03:46.320 larger 12:06 00:03:47.54000:03:47.550 capacitor does drop very slightly less 00:03:50.21000:03:50.220 rapidly than the smaller o 6:03 00:03:53.00000:03:53.010 capacitor and then we've got the x7r 00:03:56.09000:03:56.100 dielectric capacitors and I've tested a 00:03:58.07000:03:58.080 few parameters here so different case 00:04:00.26000:04:00.270 sizes different working voltages and 00:04:02.54000:04:02.550 then different capacitance values so 00:04:04.76000:04:04.770 these were all 100 nanofarads except for 00:04:07.52000:04:07.530 these two 1 micro farad capacitors and 00:04:09.50000:04:09.510 these both declined more rapidly than 00:04:12.65000:04:12.660 the lower capacitance capacitors and 00:04:15.55000:04:15.560 also the lower working voltage pasteur 00:04:18.68000:04:18.690 declined more rapidly than the high 00:04:20.39000:04:20.400 working voltage and that's illustrated 00:04:22.79000:04:22.800 more promptly here by the 200 volt 00:04:25.21900:04:25.229 capacitor which is declining 00:04:26.36000:04:26.370 significantly 00:04:27.59000:04:27.600 than all of the others then we've got 00:04:30.11000:04:30.120 basically the same story as before so as 00:04:32.15000:04:32.160 the capacitor size physical dimensions 00:04:34.46000:04:34.470 increase the decline in capacitance 00:04:37.34000:04:37.350 value is slower and I did test that 200 00:04:40.64000:04:40.650 volt capacitor all the way up to 150 00:04:43.22000:04:43.230 volts which is about as much as I could 00:04:44.54000:04:44.550 take it to and you can see it follows 00:04:46.70000:04:46.710 exactly the same shape and it's sort of 00:04:48.92000:04:48.930 asymptotes towards the same sort of 00:04:52.03000:04:52.040 nominal value at the end of its working 00:04:54.26000:04:54.270 voltage but if you did want to use your 00:04:56.81000:04:56.820 capacitor at 50 volts it might make more 00:04:59.12000:04:59.130 sense to use a 200 volt capacitor as 00:05:01.76000:05:01.770 opposed to something rated at 50 volts 00:05:04.87000:05:04.880 and then just for completeness I also 00:05:07.76000:05:07.770 tested a few different technologies of 00:05:09.71000:05:09.720 capacitors so I've kept on this graph 00:05:11.78000:05:11.790 one of the y5v capacitors that's this 00:05:14.21000:05:14.220 one with the steepest slope then we've 00:05:16.46000:05:16.470 got the x7r they've got an electrolytic 00:05:18.92000:05:18.930 in orange here and that actually 00:05:20.48000:05:20.490 increased in capacitance with voltage 00:05:22.79000:05:22.800 and then we've got three overlapping 00:05:25.04000:05:25.050 lines here which are very stable across 00:05:27.32000:05:27.330 the band and that is the class one 00:05:29.36000:05:29.370 ceramics or a C 0 G or an MP 0 ceramic 00:05:32.66000:05:32.670 capacitor those are excellent but 00:05:34.64000:05:34.650 they're generally not available in 00:05:35.84000:05:35.850 higher capacitance values then we've got 00:05:38.33000:05:38.340 a Weiner MKS 2 metalized polyester 00:05:40.58000:05:40.590 capacitor and was very stable and then 00:05:43.76000:05:43.770 we've got a Panasonic polyester 00:05:46.10000:05:46.110 capacitor and that Panasonic one was 00:05:48.65000:05:48.660 stable right up until its highest 00:05:50.54000:05:50.550 voltages here so at 100 volts it had 00:05:53.72000:05:53.730 just started to tail off there but the 00:05:55.55000:05:55.560 the wiemer MKS 2 was stable all the way 00:05:58.28000:05:58.290 up to its maximum working voltage so 00:06:01.52000:06:01.530 what is it about the ceramic capacitors 00:06:03.35000:06:03.360 that causes the capacitance to decrease 00:06:06.29000:06:06.300 with DC bias applied to the capacitor 00:06:09.05000:06:09.060 plates 00:06:09.53000:06:09.540 well the dielectric material within 00:06:11.36000:06:11.370 ml/cc capacitors is derived from barium 00:06:13.76000:06:13.770 titanate and as the voltage on these 00:06:16.01000:06:16.020 plates has increased the molecular shape 00:06:18.47000:06:18.480 of the barium titanate molecule shifts 00:06:20.51000:06:20.520 resulting in a clarity of the dipoles in 00:06:23.03000:06:23.040 the capacitor structure so with no DC 00:06:25.58000:06:25.590 bias on these plates 00:06:26.96000:06:26.970 all of these dipoles are free to rotate 00:06:29.45000:06:29.460 and that gives us the highest 00:06:31.99000:06:32.000 capacitance because the dielectric 00:06:33.95000:06:33.960 constant is a is maximum once we apply a 00:06:37.61000:06:37.620 DC bias to these plates some of the 00:06:40.46000:06:40.470 dipoles start 00:06:41.42000:06:41.430 become locked in position and they're no 00:06:43.43000:06:43.440 longer free to move that means that 00:06:45.92000:06:45.930 we've got a lower dielectric constant 00:06:48.26000:06:48.270 and therefore the capacitance is 00:06:49.79000:06:49.800 decreased and as the DC bias voltage 00:06:52.96900:06:52.979 increases more and more of these dipoles 00:06:54.98000:06:54.990 start to lock in place resulting in a 00:06:57.56000:06:57.570 lower capacitance this gives us some of 00:07:00.08000:07:00.090 the reasoning behind why we end up with 00:07:01.79000:07:01.800 these curves here so first of all for 00:07:04.55000:07:04.560 the high-voltage capacitors like this 00:07:06.43900:07:06.449 200 volt capacitor here the dielectric 00:07:08.84000:07:08.850 material is much much thicker than on 00:07:10.96900:07:10.979 the low voltage capacitors so what we 00:07:13.70000:07:13.710 actually see is that the electric field 00:07:16.10000:07:16.110 between the two plates is lower because 00:07:17.96000:07:17.970 there's a greater distance and therefore 00:07:19.37000:07:19.380 less dipoles are affected when the bias 00:07:22.37000:07:22.380 voltage increases that's also a similar 00:07:25.43000:07:25.440 effect to using smaller physical size 00:07:27.77000:07:27.780 capacitors so when the physical size is 00:07:30.08000:07:30.090 smaller for a given capacitance and 00:07:32.02900:07:32.039 voltage the actual dielectric layers are 00:07:34.58000:07:34.590 much thinner and therefore there's a 00:07:36.71000:07:36.720 much higher electric field so the 00:07:38.30000:07:38.310 dipoles become locked in place much 00:07:40.43000:07:40.440 lower DC bias voltage and similarly for 00:07:44.12000:07:44.130 the higher value capacitors so these 1 00:07:46.15900:07:46.169 microfarad capacitors in order to fit 00:07:47.99000:07:48.000 them into an Ohio v component size the 00:07:51.14000:07:51.150 layers are way thinner than they are for 00:07:53.51000:07:53.520 a 100 nano farad capacitor and therefore 00:07:56.27000:07:56.280 the electric field is even greater in 00:07:57.92000:07:57.930 those meaning that the dielectric 00:08:00.98000:08:00.990 constant decreases much more rapidly as 00:08:04.04000:08:04.050 a result of the DC bias across the two 00:08:06.02000:08:06.030 plates so there's a few things that you 00:08:08.48000:08:08.490 might want to consider when you're 00:08:09.50000:08:09.510 designing your circuit the first is that 00:08:11.45000:08:11.460 it may make more sense to use multiple 00:08:13.73000:08:13.740 smaller value capacitors than one large 00:08:16.37000:08:16.380 one because you'll be further up a slope 00:08:18.08000:08:18.090 here with your multiple smaller values 00:08:21.08000:08:21.090 in parallel the next is that you should 00:08:24.83000:08:24.840 consider looking at using a higher 00:08:26.54000:08:26.550 voltage capacitor where possible you 00:08:30.14000:08:30.150 know so even if your circuit is 00:08:31.52000:08:31.530 operating at 25 volts it may make more 00:08:34.04000:08:34.050 sense to use a 200 volt capacitor and 00:08:36.50000:08:36.510 just because it may get you further up 00:08:38.69000:08:38.700 the curve and then finally potentially 00:08:42.38000:08:42.390 the less desirable is that a physically 00:08:44.63000:08:44.640 larger capacitor will hold its value 00:08:46.79000:08:46.800 higher and a higher bias voltage the 00:08:50.69000:08:50.700 downside to that is that very small 00:08:53.69000:08:53.700 ceramic capacitors are excellent for 00:08:55.85000:08:55.860 decoupling because they have such a low 00:08:57.59000:08:57.600 inductance so by moving to a higher and 00:09:00.85000:09:00.860 physical size then the inductance will 00:09:04.01000:09:04.020 be increased and therefore you may have 00:09:05.90000:09:05.910 other effects but there's a few 00:09:07.43000:09:07.440 trade-offs to be made there and I would 00:09:09.65000:09:09.660 recommend that you potentially build up 00:09:11.24000:09:11.250 a little circuit like this they don't 00:09:12.74000:09:12.750 always publish the data in the data 00:09:15.17000:09:15.180 sheets I think the manufacturer TDK do 00:09:18.53000:09:18.540 have some very basic slopes in the 00:09:21.02000:09:21.030 datasheet for the ceramic capacitors but 00:09:23.57000:09:23.580 obviously they don't really want people 00:09:24.71000:09:24.720 seeing that their capacitors not going 00:09:27.38000:09:27.390 to perform as expected so if you design 00:09:29.93000:09:29.940 a little board like this and then you 00:09:32.18000:09:32.190 can quickly test different types of 00:09:33.95000:09:33.960 capacitors and see what kind of result 00:09:35.54000:09:35.550 they're going to have if your circuit is 00:09:37.34000:09:37.350 that critical so this would be 00:09:39.32000:09:39.330 particularly useful for things like 00:09:40.61000:09:40.620 switch mode power suppliers where you're 00:09:42.26000:09:42.270 really relying on the capacitance value 00:09:44.33000:09:44.340 to keep your circuit stable and say you 00:09:47.12000:09:47.130 might want to design something like this 00:09:48.41000:09:48.420 and then give a few different types of 00:09:50.69000:09:50.700 capacitor a test so I hope you found 00:09:53.03000:09:53.040 that useful leaving a comments or 00:09:55.13000:09:55.140 thoughts or your experiences down in the 00:09:57.02000:09:57.030 comments down below but until next time 00:09:59.00000:09:59.010 thanks for watching 00:10:03.63000:10:03.640 you 00:10:13.82000:10:13.830
Office location
Engineering company LOTUS®
Russia, Ekaterinburg, Lunacharskogo street, 240/12