brian@sdcsvax.UCSD.EDU (Brian Kantor) (08/11/87)
I'm sitting here at my workbench surrounded by a pile of bad electrolytic capacitors that I just got through ripping out of my stereo. Here's the story. I've been noticing a gradual deterioration in the sound quality of my stereo for some time, but I'd thought it was my ears, age, lack of hair, too many earrings, excessive exposure to punk, cat fur, etc. The other day, however, I played a CD for a friend, and we looked at each other and agreed that it really sounded bad. That whirring sound was Bach spinning in his grave at the distortion. Ok, add "fixing the stereo" to the list of things that need to be done. Got to it last Sunday. First, let's take the power amplifier, a Dynaco ST80. Not a bad amp back in 1973 when I bought it. Hmm. DC voltages are fine. Let's measure the response and distortion. Fine too. Nothing wrong here - they're just the same as when I measured them after I put it together. Channels match, too. Bridge the power supply caps with some .1uFs anyway, to bypass the internal inductance of the large filter caps. Preamp - a Dyna PAT-4. Also pretty good for 1973. Check some DC. Voltages ok, but there's some ripple on the power supply. Ok, well the filter capacitor is a three-section electrolytic, so bridge some test capacitance across it, and see what happens. Hum gets worse. Damn, there is inter-section leakage. Rip the thing out and replace it with three separate electrolytics. Might as well bridge them with .1 caps while I'm at it, cut down the high-frequency power supply impedance. Hmm. Better check the other capacitors. Dig out the old capacitance bridge (yes, it has a magic-eye tube in it!). Well, all the 50uF coupling capacitors are too leaky to measure what value they still are, and the 5uF ones are about 1uF, and the 100uF bypasses are about 25uF, and the 330uFs are shorted. Sigh. I didn't even test the rest of the capacitors. I replaced EVERY electrolytic capacitor in the preamp - about 4 hours work, and $10 worth of parts. I used tantalum electrolytics where I could get them - at the local surplus store, the tantalums are about 10cents more than the aluminum ones, and they're reputed to have less internal inductance, so why not? Next, check the tuner - a Dyna FM-5. Pilot lights are burned out, change them. Now the hum comes up, ok, the filter caps in the supply have dried out, and the increased load of running the pilot lamps killed the supply, so change them too - another multisection. Replaced it with separate caps too, just to be safe, and bridged them with .1uFs also. Now, how about the bypasses in the audio stage. Yup, they're dying too, so change all of them and the coupling capacitors too. $3 worth of capacitors, sigh. Put the system back together, and it sure sounds a lot better, but the bass still isn't as solid as I remember it being on the 1812 cannons.... Oops, I forgot the Bose speaker equalizer. Yup, all the electrolytics in here are bad too. Back to the surplus store for another $3 worth of capacitors, and another 2 hours to change them all. For a 15-year-old stereo, it sure sounds good again, and the cannons still scare the cat, and my hair is growing back, and the sun just started to come up, and I'm going to bed now. Next year I buy a new stereo. I >>> HATE <<< electrolytic capacitors. Brian Kantor UC San Diego "There is more harmony in films than in life." - Francois Truffaut
brian@sdcsvax.UCSD.EDU (Brian Kantor) (08/11/87)
>Really want to make it sound MUCH better? Pull all the electrolytics out of >the preamp and power amp and put in 2 uF mylars or polyprops if you can >find them. Bullsh*t. Most of the capacitors in the preamp and power amp are on the order of 50 to 100 uF; the smallest are 4.7uF. Replacing them with 2uF polypropylene caps is going to change the sound, that's sure enough. I don't think I'd want to listen to it anymore, though. And I've got great and glorious visions of finding a 5,000uF polypropylene speaker coupling capacitor.... Try an experiment sometime - take your LCR bridge, feed a 10kHz generator into it (so that it's not running at 60Hz, which many do by default), and measure the inductance of a large-value (4 - 10 uF) mylar capacitor. You may find, as I did, that it's quite a bit more than the same value tantalum electrolytic's inductance, so if you want to go around installing low-pass filters between all the stages of your amplifier, feel free. Look at the dissipation factor, Q, leakage current, etc. Try several type, and get a feel for it. Yes, the mylar and polypropylene capacitors have many superior characteristics, but they are not 100% across-the-board better. It will, in some measure, depend on where they are in what circuit. True, they'll have a whole lot less leakage current than old electrolytics, but a solidly-biased transistor stage can easily withstand the few microamps of leakage that a new tantalum electrolytic provides. Remember, we're talking about old equipment here, when it wasn't unusual for transistors to have hfe spreads of 10:1 in a single lot. It's Dynaco, circa 1973. Right out of the back of the RCA transistor manual. I wonder how many people upgrading old equipment try replacing the old parts with new ones first, before changing the type of the part. I mean, an old amplifier full of bad capacitors can't help but sound better with new parts in it, no matter what flavor they are. I'm not insisting that the mylar or polypropylene capacitors don't make a difference, I just wonder if I'll ever be able to hear it. My ears aren't golden by a long shot. By the way, I tested every one of the new capacitors at 10kHz for dissipation factor, leakage, and correct value before I put them in to replace the old parts. Worth it, too, since about 1/3 of the new-off-the-shelf parts were unacceptable too. Most of the bad ones had ten-year-old code dates; it pays to shop carefully. - Brian
max@zion.berkeley.edu (Max Hauser) (08/12/87)
This is a nit-picking response; I must make clear that I completely agree with Brian Kantor's recent articles and I think his indictment of electrolytic capacitors was eloquent and sound. I only want to address a peripheral remark. In article <3650@sdcsvax.UCSD.EDU> brian@sdcsvax.UCSD.EDU (Brian Kantor) writes: > >... a solidly-biased transistor stage can easily >withstand the few microamps of leakage that a new tantalum electrolytic >provides. Remember, we're talking about old equipment here, when it >wasn't unusual for transistors to have hfe spreads of 10:1 in a single >lot. It's Dynaco, circa 1973. Right out of the back of the RCA >transistor manual. First, it is still usual for transistors to have hfe spreads of 10:1 in a single lot, depending on the part number you buy. Indeed most modern (i.e., planar) transistors inherently have wide hfe (common-emitter current-gain) spread when manufactured, since the hfe depends more or less exponentially on the Gummel number (which they teach in school) and the emitter-region lattice integrity (which they do not), and both of these are difficult to control in ordinary bipolar fabrication processes. Often if you get a discrete bipolar with a tighter spread it is a selected version with its own part number. This is a feisty point with me only because I have designed some bipolars and seen what happens. Second, this doesn't matter. Well-designed circuits employing bipolars will be insensitive enough to the infamous hfe that they can tolerate wide spreads without much change in overall specs. So in the end, I agree completely that transistor stages can tolerate electrolytic leakage (and also that the old Dynacos, bless them, are right out of the venerable RCA transistor manual -- old farts unite!). I just disagree about why. M. Hauser, curmudgeon-in-training UUCP: ...{!decvax}!ucbvax!eros!max Internet (old style): max%eros@berkeley Internet (domain style): max@eros.berkeley.edu
bblue@crash.CTS.COM (Bill Blue) (08/13/87)
In article <3650@sdcsvax.UCSD.EDU> brian@sdcsvax.UCSD.EDU (Brian Kantor) writes: >>Really want to make it sound MUCH better? Pull all the electrolytics out of >>the preamp and power amp and put in 2 uF mylars or polyprops if you can >>find them. > >Bullsh*t. Most of the capacitors in the preamp and power amp are on >the order of 50 to 100 uF; the smallest are 4.7uF. Replacing them with >2uF polypropylene caps is going to change the sound, that's sure >enough. I don't think I'd want to listen to it anymore, though. And >I've got great and glorious visions of finding a 5,000uF polypropylene >speaker coupling capacitor.... Electrolytics in the audio path are awful. There's no if's and's or but's about it. In many many cases they are much larger than they need to be -- it all depends on the load impedance that they feed. Some simple r/c calculations can be made to determine applicability of lower value non-electrolytics. In many cases a 5 or 10mf polyprop or wonder cap is more than sufficient for linear response to below 10hz. Electrolytic bypasses are generally ok, if they are bypassed with .01's and .1's of reasonable quality. Electrolytics (and tantalums) as dc blocking in an audio circuit are just flat unacceptable. >Try an experiment sometime - take your LCR bridge, feed a 10kHz >generator into it (so that it's not running at 60Hz, which many do by >default), and measure the inductance of a large-value (4 - 10 uF) mylar >capacitor. You may find, as I did, that it's quite a bit more than the >same value tantalum electrolytic's inductance, so if you want to go >around installing low-pass filters between all the stages of your >amplifier, feel free. Look at the dissipation factor, Q, leakage >current, etc. Try several type, and get a feel for it. Yes, the mylar >and polypropylene capacitors have many superior characteristics, but >they are not 100% across-the-board better. It will, in some measure, >depend on where they are in what circuit. You *don't* want inductance or any kind of leakage in any of these applications. >True, they'll have a whole lot less leakage current than old >electrolytics, but a solidly-biased transistor stage can easily >withstand the few microamps of leakage that a new tantalum electrolytic >provides. Remember, we're talking about old equipment here, when it >wasn't unusual for transistors to have hfe spreads of 10:1 in a single >lot. It's Dynaco, circa 1973. Right out of the back of the RCA >transistor manual. > >I wonder how many people upgrading old equipment try replacing the old >parts with new ones first, before changing the type of the part. I >mean, an old amplifier full of bad capacitors can't help but sound >better with new parts in it, no matter what flavor they are. I'm not >insisting that the mylar or polypropylene capacitors don't make a >difference, I just wonder if I'll ever be able to hear it. My ears >aren't golden by a long shot. Golden or not, even untrained listeners will readily appreciate improvements made by proper component selection. It really matters how much you enjoy audio. If it's just background music it really doesn't matter a whole lot. And of course, the better the circuitry design the more evident these types of changes will make. >By the way, I tested every one of the new capacitors at 10kHz for >dissipation factor, leakage, and correct value before I put them in to >replace the old parts. Worth it, too, since about 1/3 of the >new-off-the-shelf parts were unacceptable too. Most of the bad ones >had ten-year-old code dates; it pays to shop carefully. Certainly it's good to be selective with any replacement components, but besides the gross differences described above, a great many of the characteristics that make an audio component sound good aren't measurable by methods readily available. Probably one of the most overlooked areas in audio equipment improvement programs is the power supply. Invariably, it will be extremely anemic and irregular in its frequency response (actually impedance curve). Dramatic improvements can be made in this area with relatively minor changes. --Bill
jeffw@midas.TEK.COM (Jeff Winslow) (08/13/87)
In article <1520@crash.CTS.COM> bblue@crash.CTS.COM (Bill Blue) writes: [After a man with a lot of good sense wrote:] >> Yes, the mylar >>and polypropylene capacitors have many superior characteristics, but >>they are not 100% across-the-board better. It will, in some measure, >>depend on where they are in what circuit. > >You *don't* want inductance or any kind of leakage in any of these >applications. Uh, sorry to break it to you, but *all* capacitors have series inductance, series resistance (frequency dependent, to boot), and leakage. Some just have more or less than others. Jeff Winslow