From a thread about recapping speakers.
Joe writes:
"Hi you guys, really interesting thread, I am new to AK, old to pro-audio. Retired pro audio designer, a lot of experience with caps, golden ear listening tests, and years of feedback to my designs in the field. I don’t usually express my opions in forums, a lot of bad experiences in the antique radio forums from a lot of lurkers just wanting to trash people trying to help, but this forum seems a lot better, I’ll give it a try. If any of you guys are from the recording industry, you would know my most successful design, the MCI JH-600 console. Or if you are into antique radios, I am sure you know me from Epay as radiosphonos. Anyways, I am old, and senile from 40 years of low level lead poisoning from solder, sinkers, and bullets, I’ll ramble, and forget in the middle what at the beginning I was decided to say in the end, and typing with two fingers is no help, this kinda talk is so much easier in person and with a white board. In no particular order, I believe I can give some of you guys a little more information to think about. Ok, electrolytics and several issues about them raised in this thread: if they are old they are not what they used to be, and by a lot. And for several reasons, this is a can of worms, yikes. Ok to understand this you need to consider the basics of capacitors in general and lytics in particular. A capacitor is simply conductors near each other, usually considered “plates” (planes) and for practical construction, parallel to each other (this is to make wanted capacitors, there are unwanted capacitors everywhere in our circuits like wires near each other, or lands on a circuit board, elements in a tube and on and on), but let’s discuss manufactured practical capacitors first. 2 Parallel plates or planes of conductors separated by an insulator. Farad(or somebody) discover way back when that when electricity was applied current flowed. This current would start high and quickly drop off and then stop .The capacitor to store energy is determined by the area of these plates, and the distance between them. Move em closer, capacity increases, double the area, double the capacity, Already with just this much info in a basic capacitor we are at the root of where a lot of the things this thread discusses originate. Without a white board, I’ll try to explain in writing. Ok, BUT, moving them closer to increase capacity (to make the capacitor smaller, DECREASES what voltage we can use this cap at. At some point they get close enough for voltage to jump (arc) the gap! This is the ‘breakdown voltage”. Different insulators have different properties and allow us to make our caps bigger or smaller depending on their breakdown voltages. The insulators can have another property, they can conduct small currents when their breakdown voltage isn’t exceeded, like say we used cardboard. Perfectly dry is has no leakage, but if it absorbs humidty from the air, water conducts electricity and it can become leaky, allowing a small current to flow, so our capacitor could have a leakage that varies from day to day dependent on humidity. Another important design issue here is what insulators do when they break down. Air insulator, after it broke down (arced) if the voltage is lowered, becomes an insulator again. Paper might have burned, and left carbon ash that conducts, and thus be leaky, more or less depending on how long the breakdown sustained. Tantakum capacitors, as great as they seem to be with their small size, low DCR and even lower Inductive reactance have the sad disadvantage that if their voltage rating is exceeded EVER, even for a pico-second, they blow shorted and never recover. That’s why you’ll never see em used for filter in good designs, they are relegated by knowing engineers (read that as ‘people who have burned down buildings with their designs”) to timing circuits and running Chinese made toy cars and helicopters. The first caps were air insulated, then glass, which was way better than air. Then cardboard, then oil immersed cardboard (or paper), because the oil didn’t conduct and didn’t allow humidity to enter our insulator. Electrolytics are planer conductors (two pieces of aluminum foil) wound into a cylinder, so our part is small instead of like a yardstick. separated by an electrolyte, which is a chemical paste that has the advantage of being a very good insulator with a very high breakdown voltage for the distance between the plates, allowing us to make very small parts with a very large capacitance Sideways now, we also can consider another factor, the ESR. For discussion’s sake lets make up a capacitor: suppose two pieces of one square foot area separated by an insulator yield a capacity of 12 microfarad. Great, it could be 12’ by 12”, but could also be 1 inch wide and 144 inches long (and could be rolled into a cylinder so it doesn’t stick twelve feet out of our product, more on this engineer’s nightmare later!). so lets leave it unrolled for now, and use 1 inch wide and 12 feet long. Lets connect to it at one end. Ok, assume that our aluminum foil has a dc resistance at 1’ wide of .1 ohms per foot. The capacity is 12 microfarad, BUT some of the capacity is in series with a dc resistance. Consider this as 12 1 foot long 1 mfd capacitors in parallel. The first one nearest our connection is 1mFd with no series resistance in parallel with another mFd in series with .2ohms, and the next is a mFd in series with .4 ohms, and on to the twelfth capacitor in series with 2.4 ohms! I ain’t into thinking too hard yet this morning and doing the math, but you are going to have a 10 mFd cap with an esr (equivalent series resistance). Now, suppose we hooked up our wires in the middle. Then thinking about this as twelve capacitors paralled, the ends are only 6 feet away with a DCR of 1.2 ohms, giving us a lower esr than if they are on one end. Now it gets worse, let’s roll it up. Great, it is small enough now to be practical but what have we done. We made a coil, coils exhibit a property known as inductance, got to do with current in a conductor creates a magnetic field, take away the current and the magnetic field remains for a bit, decays slowly compared to the disappearance of the current, and then a conductor in a magnetic field generates a current, so this back EMF creates a magnetic field, what it amounts to is unlike a capacitor which conducts better and better (theoretically) at higher and higher frequencies, inductors oppose current flow at high and high frequencies. We calculate or measure this flow in ohms, which is confusing, but it is called inductive reactance and is frequency dependant. Considering our model capacitor again with the wires on the ends again, we have 12 1 mFd capacitors in parallel but each further on has a DCR of 2 times .1ohm, and an XsubL (inductive reactance) in series, reactance being related to the number of turns and the frequency of operation. Now if you sketch this and crank the numbers you will find that if we hooked our connections at opposite ends, we would have the smallest inductive reactance and that is how electrolytics are made. Smallest DCR too. It is still big at high frequencies, originally we realized this back when and on decoupling electrolytcs, especially on integrated circuits that like to oscillate, we have been paralleling them with a small value mica or ceramic, to bypass the very high frequencies. On paralleling lytics with small value caps, I would have to see the circuit, it depends on the lytic ESR at the frequencies of operation, and the goal. I can say it doesn’t hurt anything, not ever, and if you aren’t an engineer, go ahead, it makes a big difference in some places, sonically and in active circuits, stability. Ok, long long ago, someone discovered while experimenting with insulators and capacitors, in the search for high breakdown voltage insulators in smaller thicknesses, with the goal of reducing cap size and more important reducing capacitor cost of manufacture, that if they saturated paper with an electrolyte paste (a chemical that conducts electricity) and then applied a slowly increasing voltage, the electrolyte would electrochemically change into an extremely efficient insulator (now, an electrolyte conducts electricity, but the electrolyte in electrolytic capacitor is electrochemically changed to an insulator, and applied voltages maintain this insulating condition, so I belive “electrolytic capacitor” is a wrong description, although it is used to make them, they should be called ‘insulatic capacitors” or something, as long as an elecrolytic is working, it has no electrolyte in it, it has an efficient insulator). Until the last 20 years or so no other capacitor type exhibited the amount of capacity per size, and they are cheap to make, no exotic materials, just aluminum foil, paper, and goo, so also capacity per dollar, always a big factor in manufacturing, often the biggest. (don’t get me started).
Lytics bluntly suck, but they do do filtering and sometimes decoupling well, in fact for the dollar, better than anything else. Small size per capacitance, low cost per capacitance, excellent self healing/recover after an overvoltage, years ago, only practical process to get high value capacitors, their was no other. This has lead not only to their almost exclusive use in power supply filtering and decoupling, which is a good engineering choice with the exception of their lifetime, but also to their use where they really suck, like in crossovers (suck for MANY reasons). Never in a blind AB golden ears listing test have they ever been chosen as the best choice for audio flowing through them, unless there was something else they were doing like leaking helping. If you can replace a crossover lytic with some other type of capacitor, you will always be better off. On the other hand, if no practical substitute is available because of size or cost, well, lytics have worked all this time and will keep working. BUT, lytics age, and more than one way. Older ones especially get dry, capacitance changes, breakdown voltage lowers, ESR increases, all bad things. Ideal maintenace of a lytic is in circuit, voltage applied and the voltage should be near it max operating voltage. Everyone knew that when you replace a lytic, to use the same voltage or higher. BAD. If you use say a 100 volt lytic in a 100 volt application, geat, the factory formed dielectric is maintained, the more it is used the longer it will live. If you replace it with a 450 volt part, yes, it works, today, the dielectric will reform at the new voltage, and it wasn’t designed for that, there is too much x or Y, and although what reformed to the new voltage is happily working, the extra X or Y has become electrolyte again (read that as “conductor”), leakage, changed capacity, etc. Not just well meaning techs, I have seen many engineers design a higher voltage part in because of availability, not understanding all the ramifications. Even a new lytic is a potential problem, look at the daycode, if it is never used but is more than 2 years old, throw it away. I was a great one for walking in the stockroom and announcing trueisms like that and being responsible for junking millions of parts, (like the time I announced that all audio ic’s should have tinned brass or copper legs, and if a magnet sticks to the legs, throw em away! But that is another story, although I will say, even in crossovers, all parts should be non-ferrous leads, terminal strips should be plated or tinned non-ferrous metal, same with binding post screws, etc.). Electrolytics (insulytics) maintain their insulating dielectric when voltage is applied to them. When they ain’t they slowly decay. It depends on the manufacturer, storage conditions, (temp) etc etc, but typical lytic infant mortality rises quickly with age. Say a particular group of caps are all good the day the were made, at 2 years storage and then use, 5% failure would be typical and at 5 years maybe 20 percent. Yes, reforming would greatly lower them numbers, but they would never meet new specs, lytics are cheap, why screw around. Ok, back to the thinking about this thread, lytics in crossovers, non-polar. We original used polarized lytics in non-polar applications by putting in series back to back. the capacity is half the value of on. The breakdown voltage is the same as either one. The diode trick sounds good in theory, but sounds terrible, especially as an audio waveform crosses the diode “knee” at .6 volts. The original non polar capacitors were two lytics, squished together round and put in the same can. Then someone got the bright idea of three pieces of aluminum foil wound together, connections to the outside pieces, cost of manufacturing goes down, size went down, great. BUT they are still lytics with lytic bad properties which in this case is relatively short lifetime, and often the inductive reactance (which can be negated by paralleling a small value mica or film.)
Ok, where was I? Caps in parallel are fine, never determined a difference from one cap, except it might be ugly (10 film .1’s in parallel, back before they made a 1 mFd film). Wiewound resistors – get em out of the signal path, they are fine in power supplies. Well, I am sure I had more I wanted to comment on, but as I predicted I forgot, I ain’t gonna read the whole thread again, wife took the twins to waterpark for their birthday, I have a very rare day off, and I am going fishing, flats or offshore?, always decisions. I hope I have given some of you experienced people a little more stuff to understand as you are working your hobby. Thanks for reading this far if you did, now I gotta see if this forum can take a message this big or if I gotta put on one of my servers and link – joe
Shoot! I just remember I came to this forum seeking advice on a subwoofer for my stereo. sorry if i bored you, one of my pet subjects, you know."