Hiend Audio, DIY, Hi-Fi, Stereo, Electronics site, for lovers of high-fidelity music reproduction. High-End Vacuum Tubes. Silicon. Do-It-Yourself Audio Systems. Schematics: amplifiers, speakers, horns, CD, DAC. Circuits, Topologies, Acoustics, Cables, Speakers, Voltage Regulators, Equipment Upgrades, Modifications, Problems, Solutions, Tips, Tricks. ------ Dla pasjonatów Hiend Audio, najwyższej wierności odtwarzania muzyki na sprzęcie budowanym, modyfikowanym samemu. Wiedza: elektronika, układy lampowe i krzemowe, wzmacniacze, głośniki, kolumny, tuby, DAC, kable, DIY, porady, sztuczki, problemy, rozwiązania, tor audio.



Aug 06 2014

Print this Post

To Bypass or NOT to Bypass – that is the question

Readers Question:

Does it make sense bypass the caps in an amplifier to improve its sound?
And what do you think about bypassing. Specifically: in crossovers? 

Capacitors Paralell

Well, does bypassing make sense ? For me, very much so, indeed. Bear in mind that each capacitor is (unfortunately) also an inductance. Hey, and this is NOT VooDoo. The PDF file plainly states this fact, for about almost any type of capacitor, which is out there (providing that it originates from a manufacturer that actually does provide some kind of technical specifications).

If you take a big, bulky hog of a capacitor, one that is a smaller or larger cylinder, be sure of one thing: apart from the capacitance, it shall also possess a sometimes not so trivial amount of parasitic inductance.

So, to make a long story short:  A capacitor is made of the following (connected in series):
A capacitance,
A resistance (the parasitic ESR – “Equivalent Serial Resistance”), and

And guess what: What is the result of such a series connection ?
Yes, you got that right. It is a series resonant circuit.

Each capacitor is essentially an INDUCTOR (above a certain threshold frequency), whether you like what I am saying here, or not. Each capacitor possesses a specific border frequency, at which it stops being a capacitor, and starts being a resistance. But there is more to that: Above that certain border frequency, it becomes inductive.

But what has this to do with the title question ?  Should we, or should we not, use bypass capacitors, i.e. smaller valued capacitors, working side-by-side with the “main” capacitor ?

The inductive properties of the bulky capacitors, as I described above, can (and sometimes even should) be prevented by using smaller, bypass capacitors, or “shunts”.

Consider the way that a capacitor is built:  It is a set of two pieces of metal, separated by dielectric, and these are WOUND up in a roll. Guess what: this winding of foil constitutes an inductance. Obviously, there are techniques to mitigate this phenomenon (such as surface metallization of the sides of such foil based caps), but none of these methods alleviate the problem completely.

But there is one thing working against the odds, working to our benefit.  A SMALL capacitor (i.e. a small “roll”) will have a SMALLER parasitic inductance.

So, if you have, for examples sake, a big foil capacitor, like 10 uF, it will, at a certain frequency, cease to be a capacitor due to it’s high inherent parasitic inductance. But guess what. Exactly at the point, where it starts to “fail” us … the small little bypass capacitor, a decade smaller in value, say, a 1uF capacitor, shall kick in and take over command.  This is because that albeit it has a smaller capacitance, now we are talking about fairly high frequency, at which it’s impedance becomes fairly low. But at the same time, it’s parasitic inductance is still small enough so as not to mess up the workings of the tandem. Well, this small little bypass capacitor, has a much lower series inductance in comparison to the large one. As the frequency is high, it does not hurt that the capacitance is lower. We have a situation here, that at the point where where the large bulky capacitor “gives in” and fails us, this small little toddler, with its much smaller inductance, begins to take over the responsibilities. For a certain range of the higher frequencies, which the bulky one simply can not handle.

So, do I use bypassing ?   Yes, I do.

Always ?  No.

How do I do it ?  Hmmm … Well, depends.  On the actual application.

Let me distinguish between two possible, viable, bypassing schemes:

a) The “Decade System”  (quite good) or

b). The “Equal Valued System” (sometimes supposedly better).

The name of the “Decade System”, I invented myself. It just came to me when I was looking at my filter bank within a power supply of sorts. For examples sake, it looks like this:

47000 uF electrolytic || 10,000 uF electrolytic || 1000 uF electrolytic || 470 uF electrolytic || 100 uF electrolytic || 10 uF MKP || 4,7 uF Paper-In-Oil || 1uF PIO || 0,22 Teflon

Got it?  ”Decade System”, because I am going down with the values at a ratio of 10:1 at each hop, more or less. Sometimes the ratio is less than 10:1.

This scenario may prove to be beneficial in Power Supply filter banks. OK, OK, maybe I just exaggerated a bit, but surely, such an extensive Decade System will not hurt, at least from the filtering’s point of view.

But what about the “Signal Path” – as an alternative point of bypass application ?

Well, this is starting to get tricky now. For starters, I shall simply say that yes, I also do use bypassing of signal capacitors which are within the signal path of my tube amplifier, Such as for example:

4.7 uF Paper-In-Oil || 4.7 MKT || 1 uF MKP || 0.47 Polystyrene || 0.22 uF Teflon  || 0.22 uF Teflon  || 0.22 uF Teflon  || 0.22 uF Teflon  || 0.22 uF Teflon.

As you can see, I rather avoid listing any brand-name capacitors here. Just generally listed the type of technology of manufacturing. It may be discussed or argumented, that a particular make of capacitor … “from THIS manufacturer, rather than THAT manufacturer” … is better or worse, or something.  May well be, but that is a subject that has been covered time and again by countless other sources.

What I would like to stress here and bring your attention to is that I used a mix of different types of technologies within such a capacitor bank, or battery.   Each type of dielectric has it’s pros and it’s cons.  Each has specific technical characteristics, like conduction angle, memory effect, and many more. What I am saying here is that by combining different technologies “in parallel”, these individual pros and cons – may be substantially “compensated for”.

As for me – such a heterogeneous capacitor battery – as a result of countless listening tests and comparisons – shall play music much better, in most of the cases.  (Caveat: there indeed are certain exceptions – especially in the power supply conditioning for DAC chips, for examples sake. Generally, providing power to a DAC chip is all but a trivial task, to being with).

So, I actually said it. Yes, I do use the “Decade Bypass System, utilizing heterogeneous technologies, at that.  Almost always for Power Supply filtering.  Often for signal paths.

Hmmm …. I can feel the anxiety and tension rising. …..

OK, agreed. There is one big “BUT” to the aforementioned.  Especially in the context of the Signal Path. The thing is as follows: Different Dielectrics tend to have a different group velocity of propagation. So my adversaries shall now raise hell and shout: If you use different kinds of parallel connected capacitors, made of different technologies, or even of the same technology, but using different makes, or even the same make, but of different values, then …

… then you get an unfortunate side effect. They call it signal smearing. Argumentation is simple enough: If one cap offers a “higher” group velocity, but it’s neighboring bypass capacitor has a “lower” group velocity, and you want to convey a very sharp spike of a signal through such a tandem connected capacitor grouping, then “obviously” – the resulting signal will be “smeared” on the other side. Smeared due to the different velocities of propagation and the different “arrival time” of each fractional piece of signal, from the individual capacitors.

Following this line of reasoning, one would have to come to the conclusion, that bypassing: YES, but in a very restrictive manner. Only on the basis of:
a). one type of technology,
b). one and the same manufacturer,   or even (the extremists):
c). exactly the SAME values of each of the individual capacitors being part of the group, or even (the ultra-extremists):
d). it should be of the same manufacturing series.

Under certain conditions and applications, yes, I do agree that there are situations where I also prefer this “Same Valued System” instead of the “Decade System”.  With the exception of the ultra-extremist restriction d)., obviously.

Let us look at an example. The series capacitance of a mid range speaker crossover.

We need a capacity of 110 uF here. Foil based, obviously. Instead of using a single, large 110 uF capacitor, instead of using a haphazard decade of heterogenous technologies, I have indeed used, within my Jensen 1071 speakers, a set of “equal valued” smaller capacitors:

22uF + 22uF + 22uF + 22uF + 22uF = 110 uF.

These indeed came from the same manufacturer, are of identical voltage rating, of identical capacity, and (just by coincidence) purchased all at the same time, resulting in being of the same production series/lot.

The rationale is such that by using the “same”, or identical capacitors from the same manufacturer, we achieve:
a). equal group velocity and no blurring of the transients / pulses,
b). a lower total inductance of the system, because we now have a parallel connection of “little inductors” instead of a single “big one”. Unfortunately, this does not change the fact that although slightly smaller, each of them has a frequency limit, above which it does not work properly. This limit is well defined and almost identical for all the individual capacitors in the set. Well, errr ….  unless we bypass thes identical caps with … something smaller).

So: Who is right?
I do not know.

Each solution has it’s merits and it’s “most reasonable” applications.

P.S.  Tolerances of capacitors are quite loosely defined. Sometimes it is +/- 5%, sometimes it is +/- 10%, but there are cases where you can experience deviations as high as 20%.

So, setting aside the whole theoretical musing as above, there are times, (such as pairing the parameters of two sets of speaker crossovers) where you simply be force to using bypasses – but this time not as a “bypass” per se, but rather as a means to finely trim and even out the differences within the behaviour of your LEFT speaker crossover and your RIGHT speaker crossover.

The difference between Theory and Praxis is that in Theory, they are the same, but in Praxis, they are not. Do what you have to do.

Listen. Compare. Enjoy.

(Veni, Vidi, Vici ? )




Permanent link to this article: http://hiend-audio.com/2014/08/06/to-bypass-or-not-to-bypass-that-is-the-question/