post #3676 of 3676
Originally Posted by Don Hills View Post

I don't think he really knows. It's the most awful mish-mash of misapplied jargon and misunderstood theory I've seen in a long time, and that's putting it politely. Capacitors have dynamic range?

If electrolytics were really as bad as he says, a lot of electronic equipment that we depend on simply wouldn't work.

Well, electrolytics have one thing in common with moving magnet phono cartridges. They do "kind of" work "well enough" - and they once , like the moving magnet cartridge, were the only reasonable game in town.


There is, of course, a difference between the two. Moving magnet cartridge, or at least the majority of them, have problems due to the limited frequency response.

It is due to the LRC filtering that occurs WELL WITHIN the audible frequency range for the vast majority of the MM cartridges - yet it can be made to look superbly flat up to and slightly beyond 20 kHz with the use of the precisely made mirror image resonance in the mechanical part of the transducer. Such a cartridge, even if and when electrically terminated with load that does produce the flattest measured response, will never sound nearly as good as a decently designed cartridge that is not forced to use mechanical resonance in order to measure flat; one resonance in audible range can thus be avoided and with it all phase shift anomalies that go with it.  The net result is usually FAR better imaging - despite both cartridges having for all practical purposes the same frequency response up to 20 kHz. There is another, far more important benefit - MUCH lower record wear. Who bears the brunt of that mechanical resonances - if not YOUR vynil groove walls, precisely where it is the most critical to begin with, that is to say high frequencies where lion's share of a phono cartridge work is ?


( Please note I am NOT against MM principle in general - only poor implementations. There were/are a few MM designs that did adres this in an exemplary fashion, one happens to be the fastest cartridge of them all . )


Electrolytic capacitor is an even more sneaky thing. From your response, it is clear you have not even considered a capacitor could possibly have dynamic range - assuming it was perfect/infinite !? Now , PLEASE do click on the link

 - and each and every sub-link as well. To your surprise, there will be mechanisms that lead to dynamic range limitations in (electrolytic) capacitors described - along the methods for measuring them. And that has been available in print since 1980 !


Extremely condensed conclusion - below say - 40 dB ( 1% of the maximum voltage level ), due to all ill effects of a REAL electrolytic capacitor, there is ever greater error as the signal is getting lower - because the capacitor is unable to discharge correctly/perfectly below that -40 dB or 1% voltage and signal has to "pull" it down instead. That IS painfully audible - if one does follow the advice given according the possible improvements, one is rewarded with MUCH better definition of sound; and yes, that gets better the lower the level. After having heard the "dynamic noise" ( all the distortions due to the capacitor issues ) removed (or at least brought down from 1 % to 0.01 % (still somewhat reasonable in price) or lower ( sky is the limit ) by the the judicious use of quality capacitors ) - there is no going back. 


Trouble, as already mentioned countless time, is the cost. Simply go to any electronic components distributor and check the prices of film and electrolytic capacitors of the values needed for any real world circuit. And please do use manufacturer's reasoning regarding pricing - it is approx that MSRP is about 5 times all the expenses to produce it. So, going from one single capacitor that costs the manufacturer instead of 0.50$ for an electrolytic capacitor say to a film type for 3 $ does not increase MSRP by the difference of $ 2.50, but 5 times that - $ 12.50. Multiple that with say an average 50-100 times as in real world designs ( I count here replacement of small value ceramic capacitors for film types as well ) - and it is clear why nobody does it anymore. Furthermore, really good small value capacitors , the polystyrene, are extremely "hostile" to solder, being utterly unsuitable for anything like automated stuffing/soldering . Taking the time to solder one or two by hand being equal of automated soldering of the entire PCB...


An additional consideration that is in practice the main drawback, is the size/volume limitations. Film capacitors are in general FAR bigger than whatever one is replacing them with. So, for optimum results, a new PCB is the only correct way, meaning it is better to design a PCB with these large(r) components in mind from scratch. Extremely expensive and worth of consideration only if there is a large(r) unit run to be made. I know of extremely few high end devices that did take this in concern - at least they allow for replacing that dreaded electrolytic with film, as they did provide for enough space on the PCB. Doing it on "normal" PCBs designed for normal size electrolytic capacitors IS EXTREMELY difficult and time consuming. Replacing say a single 47 uF electrolytic with film type can set you back for a couple of hours - before it can *somehow* be squeezed into space available AND perform as it should - there are drawbacks regarding bigger size, increased lead spacing etc, resulting with higher inductance, possibly leading to oscillations - these ill effects have to be dealt with, otherwise the result may well end up as worse than staying with the original electrolytic ... - I have encountered cases where this simply could not be made to work properly :(.


But results are worth it. And they end the dilemma "Do the amplifiers sound different" - for good. Despite both stock and modified amp having EXACTLY the same schematics - just because that C on the schematic is assumed  to be good enough/perfect for the job, it does not necessary mean it is .


It is precisely this what is wrong with capacitors; they do not work in a "digital" fashion, either work or not; they perform with various degrees of accuracy.


And to make matters worse, loudness wars with their compression of masters ( gee, today's Linn 24 Bits of Christmas free sample clearly being one of those - dynamics being next to non-existent ... ) operate well ABOVE where the effects of poor capacitors start to become audible. 


It takes some listening - repeated by more listening, preferably to live music - in order to be able to to hear the difference. If one relies on recordings by others, there is no way of knowing how and on which equipment they have been made - and there is absolutely zero guarantee they are free from capacitor induced limitations. And you can ABX  those recordings :deadhorse: and , quite correctly so, hear no difference. Which does not mean it does not exist on properly made recordings being auditioned on gear of high enough quality. 


As can clearly be seen by now, it is anything but easy to further the pursuit of SQ; one needs to fulfill one condition, then next, etc - and some are very well "hidden", like this capacitor thing. And is too complex to be meanigfully presentable with current measuring practice - how on earth can one measure something if the device under test is order(s) of magnitude better than the measuring device ?


( Capacitor thing is grim enough to the point there are no/few? amplifiers that can drive the pure capacitive load for the purpose of testing capacitors at anything above small signal conditions - one audio amp specifically designed for capacitive load lasted the whole of couple of seconds when driving 10 uF/1 kVAC capacitor at 30 VRMS - although stable, FAR bigger heatsinks would have been required in order for the temperature shut down protection not to be tripped . Most others would not provide stable performance or blow their fuses instantly. If manufacturers of capacitors are struggling with this - what can be said of the amateur side ? )


And yes, in one of the links to recent AES papers regarding audibility of 16 vs 24 bit, there were TWO MORE papers; one dealing with the audible effects of digital filtering, the other with the conclusion that CD/redbook is NOT transparent for certain audible signals. 


My approach is fundamentally different from the CD crowd. I try to eliminate from the realm of possibility of anything that possibly might be audible ( bandwidth in electronics up to MHz level, sampling frequencies, bit depths, etc, etc )

- and can of course be accused of using overkill and high cost. True. But at least can not be accused of using too little - once the recording is over, it is too late to realize it was "too little". Despite not being in recording for long, I do have recordings of people who are no longer with us - no way of repeating them with "good enough according to whatever AES might come up in the future". The biggest topic of the AES in 80s was - MP3 and what can be left out of already insufficient CD in order to squeeze so and so much music on so and so much of limited storage - now it seems to be gradually picking up to approx the level of quality we had back in the analog days.


How many good musical artists have passed away in the intervening 30 or so years - which were dominated with digital recording, only recently being significantly improved over redbook ?

Edited by analogsurviver - 12/20/14 at 4:34am