/ News & Press / Video / Replacing MLCCs with Polymer Capacitors - The Learning Circuit

Replacing MLCCs with Polymer Capacitors - The Learning Circuit

WEBVTT
Kind: captions
Language: en

00:00:00.260
the following program is brought to you
00:00:02.690 00:00:02.700 by element14 the electronics community
00:00:05.360 00:00:05.370 where you can connect and collaborate
00:00:06.619 00:00:06.629 with top engineers from around the world
00:00:08.780 00:00:08.790 join now at element14.com hello my name
00:00:13.970 00:00:13.980 is James and welcome back to the
00:00:15.620 00:00:15.630 learning circuit as you can tell I'm not
00:00:18.200 00:00:18.210 Karen this video is part two of my
00:00:20.870 00:00:20.880 series on polymer capacitors if you are
00:00:24.200 00:00:24.210 totally new to capacitors you will want
00:00:26.570 00:00:26.580 to check out Karen's video on them she
00:00:28.759 00:00:28.769 did a great job covering the basics
00:00:30.290 00:00:30.300 you'll also want to take a look at my
00:00:32.179 00:00:32.189 previous video which introduced these
00:00:34.459 00:00:34.469 special capacitors called polymers now
00:00:37.520 00:00:37.530 in this video I am making measurements
00:00:40.280 00:00:40.290 in real circuits before and after
00:00:42.860 00:00:42.870 swapping out capacitor types there is a
00:00:45.529 00:00:45.539 ton I want to cover but I will give a
00:00:48.110 00:00:48.120 very quick review of polymers first even
00:00:58.069 00:00:58.079 though we call these capacitors polymers
00:00:59.840 00:00:59.850 that material is part of the cathode
00:01:02.060 00:01:02.070 layer there are two major types which
00:01:04.549 00:01:04.559 use either aluminum or talam as the
00:01:07.250 00:01:07.260 anode with their oxide as the dielectric
00:01:10.300 00:01:10.310 polymers or polymer electrolytic s--
00:01:12.679 00:01:12.689 offer low ESR high capacitance and long
00:01:15.469 00:01:15.479 operational life unlike ceramic
00:01:18.050 00:01:18.060 capacitors they do not change with
00:01:19.550 00:01:19.560 applied voltage or significantly drift
00:01:21.679 00:01:21.689 with temperature or time see I told you
00:01:25.700 00:01:25.710 it would be quick now I'll slow down so
00:01:28.219 00:01:28.229 that we can talk about how we will
00:01:29.330 00:01:29.340 measure them
00:01:30.640 00:01:30.650 [Music]
00:01:33.250 00:01:33.260 the first project addresses the idea of
00:01:36.230 00:01:36.240 replacing a ceramic or ml/cc with a
00:01:38.660 00:01:38.670 polymer capacitor at the time of this
00:01:41.390 00:01:41.400 video there's a shortage of high
00:01:43.460 00:01:43.470 capacitance ceramics so a common task is
00:01:47.180 00:01:47.190 trying to replace a ceramic with a
00:01:49.040 00:01:49.050 polymer in an existing design to
00:01:52.160 00:01:52.170 simulate such a task I am replacing the
00:01:54.620 00:01:54.630 output ceramic capacitor in an
00:01:56.840 00:01:56.850 evaluation board available from TI this
00:01:59.960 00:01:59.970 board has a TPS six to zero nine seven
00:02:02.749 00:02:02.759 step-down converter the input can take
00:02:05.600 00:02:05.610 up to six volts and it is configured for
00:02:07.520 00:02:07.530 1.2 volts out for my testing I'll be
00:02:10.699 00:02:10.709 using 2.5 volts in and a load of one ohm
00:02:13.430 00:02:13.440 which will drive about 1 amp through the
00:02:15.979 00:02:15.989 board for the second measurement
00:02:18.440 00:02:18.450 comparison I am going back to one of my
00:02:21.410 00:02:21.420 favorite 8-bit computers the Commodore
00:02:23.810 00:02:23.820 64 it has some traditional aluminum
00:02:26.270 00:02:26.280 electrolytic s-- that I'm going to
00:02:27.650 00:02:27.660 replace with polymers to see if it
00:02:30.199 00:02:30.209 changes how the on board power supplies
00:02:32.060 00:02:32.070 behave in both cases I am measuring a DC
00:02:35.690 00:02:35.700 or rail voltage or more correctly the
00:02:38.660 00:02:38.670 peak-to-peak noise on top of that rail
00:02:40.990 00:02:41.000 next I will explain how to make this
00:02:43.460 00:02:43.470 measurement with an oscilloscope an
00:02:49.150 00:02:49.160 oscilloscope is great for measuring the
00:02:51.949 00:02:51.959 peak-to-peak noise or the AC component
00:02:54.620 00:02:54.630 of a DC voltage rail for the DC to DC
00:02:57.620 00:02:57.630 converter I soldered this pigtail coax
00:02:59.840 00:02:59.850 to my board when using a coax cable like
00:03:02.390 00:03:02.400 this one you must be careful not to
00:03:05.120 00:03:05.130 exceed the Scopes maximum input voltage
00:03:07.729 00:03:07.739 since it is a one to one attenuation
00:03:10.479 00:03:10.489 some scope companies even make a special
00:03:13.310 00:03:13.320 power rail probe specifically for this
00:03:15.830 00:03:15.840 measurement which will solve that
00:03:17.660 00:03:17.670 voltage problem as well as a few others
00:03:19.699 00:03:19.709 the cable then connects to the special
00:03:22.280 00:03:22.290 probe when we get to measuring the c64 s
00:03:25.310 00:03:25.320 rail voltages I will use a browser
00:03:27.020 00:03:27.030 version of this special probe
00:03:30.180 00:03:30.190 00:03:32.540 00:03:32.550 the TI board comes with a 22 microfarad
00:03:36.270 00:03:36.280 ceramic capacitor on the output of the
00:03:38.520 00:03:38.530 regulator we need to measure its
00:03:40.559 00:03:40.569 performance for a baseline I hook up the
00:03:43.170 00:03:43.180 board turn on the input voltage and look
00:03:46.140 00:03:46.150 at the scope here we see the 1.2 volts
00:03:48.630 00:03:48.640 out and that there is a little bit of
00:03:50.580 00:03:50.590 noise on the rail the peak-to-peak
00:03:52.410 00:03:52.420 measurement shows about 40 millivolts a
00:03:54.960 00:03:54.970 noise which sounds high to me
00:03:57.360 00:03:57.370 I want more resolution for that peak to
00:03:59.880 00:03:59.890 peak measurement so I offset the signal
00:04:02.550 00:04:02.560 to the center of the screen and crank up
00:04:04.830 00:04:04.840 the volts per division now there is a
00:04:07.860 00:04:07.870 lot more detail to the peak to peak
00:04:09.660 00:04:09.670 noise also notice that the measurement
00:04:12.360 00:04:12.370 value dropped a bit this change is
00:04:14.610 00:04:14.620 because the scope is getting more
00:04:15.990 00:04:16.000 vertical resolution around the part of
00:04:18.539 00:04:18.549 the signal we care about let's see what
00:04:20.940 00:04:20.950 happens when the load is turned on Wow
00:04:24.710 00:04:24.720 check out all those spikes they force
00:04:27.690 00:04:27.700 the peak-to-peak voltage up to 157
00:04:30.240 00:04:30.250 millivolts on a 1.2 volt signal that's
00:04:33.000 00:04:33.010 like a 13 percent change in margin
00:04:36.290 00:04:36.300 before we criticize this board design
00:04:38.730 00:04:38.740 let's consider a couple of things first
00:04:40.830 00:04:40.840 this is an evaluation module it's meant
00:04:43.530 00:04:43.540 to allow somebody to see if this chip
00:04:45.510 00:04:45.520 will work in their application there are
00:04:47.550 00:04:47.560 multiple modes and other configurations
00:04:49.800 00:04:49.810 that we can be using which may change
00:04:51.360 00:04:51.370 its performance second the electronic
00:04:54.960 00:04:54.970 load that I'm using isn't perfectly
00:04:56.969 00:04:56.979 resistive and is being connected with
00:04:59.040 00:04:59.050 relatively long wires the purpose of
00:05:01.770 00:05:01.780 this comparison test is to simulate a
00:05:03.810 00:05:03.820 real-world case of trying to change
00:05:05.580 00:05:05.590 capacitors in an existing design with
00:05:08.969 00:05:08.979 the baseline performance established I
00:05:10.980 00:05:10.990 carefully remove the ceramic capacitor
00:05:13.230 00:05:13.240 and then replace it with a 22 micro
00:05:15.630 00:05:15.640 farad polymer talam by the way Taylan
00:05:18.840 00:05:18.850 capacitors marked our anode and aluminum
00:05:21.270 00:05:21.280 capacitors marked their cathode I guess
00:05:24.270 00:05:24.280 they just wanted to be different
00:05:26.170 00:05:26.180 with the Palmer talam the no-load
00:05:28.750 00:05:28.760 regular output has 31 millivolts of
00:05:31.060 00:05:31.070 ripple that is a big increase from the
00:05:34.090 00:05:34.100 ceramics 18 millivolts with no load what
00:05:36.910 00:05:36.920 do you think is going to happen when we
00:05:38.470 00:05:38.480 turn the load on let's go see the
00:05:42.970 00:05:42.980 transient spikes drive the peak to peak
00:05:44.500 00:05:44.510 noise up to 130 millivolts now remember
00:05:47.620 00:05:47.630 the ceramic capacitor was at 157
00:05:49.960 00:05:49.970 millivolts so the polymer reduced the
00:05:52.150 00:05:52.160 spikes by 17% the ESR of the polymer is
00:05:55.930 00:05:55.940 higher than the ceramic normally this
00:05:57.910 00:05:57.920 would result in a larger voltage drop
00:05:59.770 00:05:59.780 across the capacitor increasing the
00:06:01.480 00:06:01.490 ripple voltage I think what we are
00:06:03.610 00:06:03.620 seeing is that the polymers ESR is
00:06:05.620 00:06:05.630 dampening the switching transients as an
00:06:08.200 00:06:08.210 experiment I decided to try adding a 100
00:06:10.630 00:06:10.640 nano farad ceramic capacitor onto one of
00:06:12.730 00:06:12.740 the extra pads look at the performance
00:06:15.010 00:06:15.020 now the transients are significantly
00:06:17.740 00:06:17.750 reduced and the peak to peak voltage has
00:06:19.720 00:06:19.730 dropped to 19 millivolts I know what
00:06:25.090 00:06:25.100 you're thinking
00:06:25.660 00:06:25.670 the bald engineer just said in order to
00:06:28.180 00:06:28.190 replace a ceramic capacitor replace it
00:06:30.430 00:06:30.440 with a ceramic capacitor technically
00:06:32.680 00:06:32.690 that is correct which by the way is the
00:06:34.480 00:06:34.490 best kind of correct however what I am
00:06:36.760 00:06:36.770 suggesting is that you always consider a
00:06:38.950 00:06:38.960 small value ceramic for reducing
00:06:41.410 00:06:41.420 switching noise and in this case a 50
00:06:44.260 00:06:44.270 volt 100 nano farad Oh 805 ceramic are
00:06:47.290 00:06:47.300 still easy to come by because they are
00:06:49.240 00:06:49.250 by far the most popular capacitor
00:06:51.850 00:06:51.860 produced oh and before moving on I did
00:06:55.690 00:06:55.700 go back and put the 20 2 micro farad
00:06:57.310 00:06:57.320 ceramic into the board with the 100 nano
00:06:59.830 00:06:59.840 farad to see what it looked like and we
00:07:02.260 00:07:02.270 can see that it's performance
00:07:03.370 00:07:03.380 dramatically improved as well so what
00:07:07.840 00:07:07.850 have we learned about polymers in a
00:07:09.370 00:07:09.380 switching DC to DC converter first even
00:07:13.240 00:07:13.250 if you can get the same case size
00:07:14.740 00:07:14.750 voltage and capacitance value these are
00:07:17.140 00:07:17.150 not drop-in replacements for a ceramic a
00:07:19.890 00:07:19.900 switching converter will operate
00:07:21.910 00:07:21.920 differently so you need to verify its
00:07:24.340 00:07:24.350 operation second it is also possible
00:07:27.700 00:07:27.710 that a polymer alone could improve the
00:07:29.890 00:07:29.900 performance of a circuit and third
00:07:32.410 00:07:32.420 always consider a low value filter
00:07:34.510 00:07:34.520 capacitors when designing your boards
00:07:37.159 00:07:37.169 just in case my friends Eric and Steve
00:07:39.379 00:07:39.389 watch this video I do need to mention
00:07:41.300 00:07:41.310 that measuring the peak-to-peak voltage
00:07:42.860 00:07:42.870 of a regulators output is only one of
00:07:45.409 00:07:45.419 many measurements that need to be made
00:07:47.239 00:07:47.249 to truly evaluate a capacitors
00:07:50.269 00:07:50.279 performance in a converter you would
00:07:52.159 00:07:52.169 need to measure things like the power
00:07:53.899 00:07:53.909 supply rejection ratio or its control
00:07:56.420 00:07:56.430 loop stability or its parameters across
00:07:59.149 00:07:59.159 things like temperature but I'm gonna
00:08:02.029 00:08:02.039 have to save those measurements for
00:08:03.649 00:08:03.659 another video let's move on to see how
00:08:06.589 00:08:06.599 aluminum polymers function when
00:08:08.089 00:08:08.099 replacing traditional aluminum
00:08:09.860 00:08:09.870 electrolytic Sande a design that uses
00:08:11.959 00:08:11.969 linear regulators here is one of my many
00:08:18.559 00:08:18.569 Commodore 64 motherboards on it are
00:08:21.379 00:08:21.389 three huge traditional aluminum
00:08:23.480 00:08:23.490 electrolytic capacitors back on a
00:08:25.790 00:08:25.800 workbench Wednesday's episode I replace
00:08:28.339 00:08:28.349 those capacitors with polymer aluminum's
00:08:30.439 00:08:30.449 using some cool soldering tools the
00:08:33.259 00:08:33.269 leads for this capacitor are on the same
00:08:35.480 00:08:35.490 axis so they are called an axial
00:08:38.420 00:08:38.430 capacitor while these are called a
00:08:40.939 00:08:40.949 radial since the leads I don't know
00:08:44.720 00:08:44.730 something about the radius of a circle
00:08:46.639 00:08:46.649 yeah I don't know why they're called
00:08:48.530 00:08:48.540 radials I couldn't find a polymer with
00:08:50.900 00:08:50.910 an axial configuration so that meant I
00:08:53.389 00:08:53.399 had to use radials and get creative with
00:08:55.160 00:08:55.170 my soldering which you'll see later but
00:08:57.680 00:08:57.690 before we get to that I want to talk
00:08:59.150 00:08:59.160 about three measurement points they are
00:09:01.519 00:09:01.529 the five and 12 volt outputs of the
00:09:03.500 00:09:03.510 linear regulators as well as the 12 volt
00:09:06.319 00:09:06.329 regulators input oh and make sure you
00:09:09.350 00:09:09.360 check the show notes for a post I put
00:09:11.150 00:09:11.160 together on C 90 it's an electrolytic
00:09:13.519 00:09:13.529 that looks like AC is being directly
00:09:15.380 00:09:15.390 applied to it to make measurements
00:09:17.569 00:09:17.579 easier I'm going to measure the 12 volt
00:09:19.939 00:09:19.949 and 5 volt rails using their C 102 and C
00:09:22.939 00:09:22.949 57 s on the board while I am measuring
00:09:25.730 00:09:25.740 the voltages to give the c64 something
00:09:28.100 00:09:28.110 to do it is running a dead cart test the
00:09:31.310 00:09:31.320 computer is actually fine I just need a
00:09:33.170 00:09:33.180 distract it with something like poke at
00:09:35.030 00:09:35.040 it first up is the 5 volt rail even
00:09:39.050 00:09:39.060 though this is a linear regulator look
00:09:41.060 00:09:41.070 at the pattern in its ripple voltage it
00:09:42.829 00:09:42.839 is clear that as the c64 performs
00:09:45.769 00:09:45.779 different operations it causes changes
00:09:48.050 00:09:48.060 in the contour
00:09:48.960 00:09:48.970 which creates the overall ripple which
00:09:52.080 00:09:52.090 with the original capacitors is around
00:09:54.300 00:09:54.310 24 millivolts peak to peak I probably
00:09:57.060 00:09:57.070 should point out that I'm not using the
00:09:58.620 00:09:58.630 original Commodore 64 supply instead I'm
00:10:01.740 00:10:01.750 using a modern replacement but I don't
00:10:04.200 00:10:04.210 expect that to change these results the
00:10:06.600 00:10:06.610 12 volt supply has a similar looking
00:10:08.310 00:10:08.320 pattern with a peak to peak voltage of
00:10:09.930 00:10:09.940 about 30 millivolts what is more
00:10:12.120 00:10:12.130 interesting is to go and look at the
00:10:13.740 00:10:13.750 input side of the 12 volt supply that's
00:10:16.950 00:10:16.960 60 Hertz isn't it amazing that the
00:10:19.860 00:10:19.870 linear regulator is almost entirely
00:10:21.570 00:10:21.580 eliminating that signal from the output
00:10:24.350 00:10:24.360 granite the peak to peak change is only
00:10:26.520 00:10:26.530 about 1.1 8 volts but it still means
00:10:29.940 00:10:29.950 that the regulator is working hard now
00:10:33.120 00:10:33.130 it is time for me to do a little bit of
00:10:34.410 00:10:34.420 soldering to replace the original
00:10:35.490 00:10:35.500 capacitors once again I'll check the
00:10:38.970 00:10:38.980 file output first and it's about the
00:10:42.810 00:10:42.820 same as before that's weird
00:10:46.220 00:10:46.230 let's go check the 12 volt supply okay
00:10:51.090 00:10:51.100 it's actually a little bit more Wow
00:10:53.790 00:10:53.800 that's not something I expected or is it
00:10:58.110 00:10:58.120 switching the polymers did change the
00:11:00.330 00:11:00.340 output ripple a little bit but that's
00:11:02.370 00:11:02.380 not the big impact these capacitors are
00:11:05.010 00:11:05.020 on the input of the regulators so let's
00:11:07.890 00:11:07.900 go take another look at that 12 volt
00:11:09.630 00:11:09.640 input again and now we see it dropped by
00:11:12.480 00:11:12.490 about 100 millivolts you might say that
00:11:15.360 00:11:15.370 a 10% reduction isn't very much but any
00:11:18.690 00:11:18.700 voltage ripple we remove on the input
00:11:21.360 00:11:21.370 means that the regulator does not have
00:11:23.790 00:11:23.800 to work as hard on the output speaking
00:11:27.810 00:11:27.820 of output let's go change the
00:11:29.640 00:11:29.650 electrolytic capacitors on the linear
00:11:31.500 00:11:31.510 regulators output side and check out
00:11:34.440 00:11:34.450 these numbers the 5 volt rail dropped to
00:11:38.280 00:11:38.290 about 13 millivolts which is a serious
00:11:41.400 00:11:41.410 drop less ripple on the 5 volt rail
00:11:44.160 00:11:44.170 means a more stable system to be honest
00:11:46.890 00:11:46.900 this is the kind of change I was hoping
00:11:48.840 00:11:48.850 to see a 30% improvement just by
00:11:51.570 00:11:51.580 switching to a polymer now you might be
00:11:54.900 00:11:54.910 eager to point out that these caps are
00:11:56.790 00:11:56.800 40 years old so maybe some of the
00:11:59.820 00:11:59.830 improvement is just from them being new
00:12:01.820 00:12:01.830 capacitors let's go check using my D mm
00:12:05.480 00:12:05.490 I measure the capacitance is ten point
00:12:07.820 00:12:07.830 four seven micro farad's for a 10 micro
00:12:09.830 00:12:09.840 farad capacitor so that's a good sign
00:12:11.480 00:12:11.490 but the other measurement we need to
00:12:13.520 00:12:13.530 consider is the leakage current that
00:12:16.340 00:12:16.350 measurement tells us the state of the
00:12:17.990 00:12:18.000 dielectric and the like left in the
00:12:19.850 00:12:19.860 electrolyte it should be less than 500
00:12:22.790 00:12:22.800 nano amps to measure that I applied five
00:12:25.310 00:12:25.320 volts DC to the capacitor with a 10
00:12:27.560 00:12:27.570 milli amp limit then I waited one minute
00:12:30.290 00:12:30.300 to see that the leakage was about 140
00:12:32.660 00:12:32.670 nano amps these two measurements tells
00:12:35.810 00:12:35.820 me that the dielectric and the
00:12:37.460 00:12:37.470 electrolyte are still in workable shape
00:12:39.470 00:12:39.480 so I think it's safe to say that most of
00:12:42.530 00:12:42.540 the improvement came from the fact that
00:12:44.240 00:12:44.250 we use a polymer capacitor type just
00:12:50.630 00:12:50.640 like with my previous episode on polymer
00:12:52.370 00:12:52.380 capacitors if you'd like to ask
00:12:54.020 00:12:54.030 questions please head over to element14
00:12:56.180 00:12:56.190 there is a link in the description and
00:12:58.400 00:12:58.410 we'll put one on screen right now at
00:13:01.430 00:13:01.440 that link I will also include a ton of
00:13:03.440 00:13:03.450 scope screenshots and details from the
00:13:05.450 00:13:05.460 measurements that I showed there are a
00:13:07.700 00:13:07.710 few bits that I left out for time so you
00:13:09.740 00:13:09.750 may be interested to check that out in
00:13:11.720 00:13:11.730 summary the biggest challenge with using
00:13:13.700 00:13:13.710 polymers and existing hardware is their
00:13:15.980 00:13:15.990 physical form factors from an electrical
00:13:18.530 00:13:18.540 perspective we saw that the impedance of
00:13:20.420 00:13:20.430 a polymer will have an effect on your
00:13:22.250 00:13:22.260 circuit so you'll need to do some
00:13:24.140 00:13:24.150 testing to see how well they work
00:13:25.220 00:13:25.230 however it's been my experience that if
00:13:28.430 00:13:28.440 they're designed in from the beginning
00:13:29.750 00:13:29.760 there will not be many issues I hope you
00:13:32.480 00:13:32.490 enjoyed the measurements and learn
00:13:33.650 00:13:33.660 something about polymers along the way
00:13:35.120 00:13:35.130 thank you for watching again if you have
00:13:37.460 00:13:37.470 questions please come find me on the
00:13:38.960 00:13:38.970 element14 community
00:13:41.720 00:13:41.730

Sales phone