Bypass cap's (a long one)

[Edu Silva]

I found the following text about bypass caps at Dan Hildebrand's page:
http://www.qnx.com/~danh/tubenote4.html

Well, it's not my opinion, but makes some sense. However, I'm still doing
the old way...

**********

Tube Amp Design Notes

This series of postings, by Henry Pasternack, is a discussion on the design
of a tube power amp.

From: "henry (h.) pasternack" 
Subject:  Tube Note 4. 

My design effort was temporarily stalled for a couple of weeks while I
mulled over some issues of component philosophy. For years, I have
reflexively paralleled all electrolytics with small film caps under the
audiophile assumption that these capacitors would augment the performance
of the electrolytics at high frequencies. 

The problems with using these caps are several-fold. They take up space;
they cost money; they require careful grounding if they are to be
effective; they have the potential to cause high-frequency ringing; they
may be ineffective; some people have reported that bypassing actually
increases audible coloration. 

In my last amplifier, I took the approach of treating the driver circuit as
a unit, with its own ground system and all power rails bypassed at high
frequency by 0.1uf film caps. The input signal ground and the bypass cap
ground were run separately back to the star ground in the power supply.
While the presence of those pretty red and blue gumdrops is very pleasing
to the eye and soothing to the conscience, I wonder if they really do any
good. 

Let's suppose there is some high-frequency noise riding on the power supply
rails. At 100kHz, the impedance of a 0.1uF cap is about 15 Ohms. At 20kHz,
it's five times higher. What is the benefit? 

I did some simple tests with a couple of capacitors, a 100 Ohm film
resistor, and a 10-10MHz signal generator. What I found is that the
impedance of even a lowly Sprague computer-grade electro- lytic is in the
range of tenths or hundredths of an Ohm out to frequencies in the hundreds
of kilohertz. The effect of a 0.1uF film bypass was entirely negligible.
The frequency response of the film motor-run caps I'm using in this amp was
even better. The bypass had no beneficial effect on the film cap, but did
result in resonances at medium RF frequencies. 

On the other hand, putting a 10uF SCR polypropylene cap in parallel with
the electrolytic resulted in a uniform and very low impedance right across
the audio band and out to AM radio frequencies. Modern electrolytic caps
are a lot better than those built two decades ago. It's probably time to
rethink the wisdom of automatically bypassing every electrolytic with a
small film cap. 

The wavelength of a radio wave in free space at 1MHz is three hundred
meters. The length of the wire carrying B+ from the filter caps to my
driver circuit is about three inches. This is also food for thought. 

Most solid-state power amplifiers are littered with bypass caps. A good
designer will consider using tantalums or ceramics because these types,
although they aren't that linear, have very low impedance and wide
bandwidth compared to films. Far from being superfluous, the bypass
capacitors keep the supply rail impedance low at frequencies in the tens of
Megahertz. This is necessary because the open-loop bandwidth of a
solid-state amp is often this high. As well, solid-state devices are more
prone to producing distortion in the presence of RF noise than tubes. 

On the other hand, the open-loop bandwidth of a typical tube amplifier is
only a couple of hundred kilohertz at best. They don't suffer from junction
rectification problems. 

Finally, I surveyed many schematics of classic tube amplifiers, vintage and
modern, and found that there were very few gratuitous bypass caps in them.
Of course, it costs money to install these parts. But I suspect there'd be
more of them if they really made a big sonic difference. 

The "underground" DIY literature was inconsistent. Some tweak articles
mandated elaborate networks bypasses-upon-bypasses. Others said the
bypasses screwed up the sound. Not much help there. 

I am sure there are many cases where changing a cap here and there can make
a significant difference in the sound in spite of objective predictions to
the contrary. The problem is, if we never build amplifiers without these
parts, we will never have a standard of comparison for tweaked-out designs.

So, I am using quality film filter caps in my power supply, and I am going
to forego most of the bypass caps unless I can convince myself there is a
good engineering reason for having them in the circuit. This makes the
circuit board layout much easier and keeps the cost down, as well. We'll
see how it sounds. 

**********

_____________________
Edu Silva - ES2 Audio
Salvador-Bahia-Brazil