PDA

View Full Version : Bypassed and Biased Capacitors



Techbot
10-23-2004, 06:22 AM
JBL began "bypassing" the mylar and non-polarized electrolytic capacitors in their frequency dividing networks beginning with the model L110A loudspeaker system. These bypass capacitors were smaller, higher quality polypropylene and polystyrene capacitors which exhibited greater linearity, spaciousness, and dynamics.

Our Library contains numerous documents and brochures with references to this practice. For example:

Page 3 - 1982 Loudspeaker Systems Catalog (http://lansingheritage.org/html/jbl/catalogs/1982-home.htm)

Page 2 and 3 - JBL L250 & B460 (http://lansingheritage.org/html/jbl/specs/home-speakers/1982-l250-b460.htm)

Page 7 - JBL DD55000 Everest (http://lansingheritage.org/html/jbl/specs/home-speakers/1985-everest.htm)

Techbot
10-23-2004, 06:24 AM
Original N110 network as used in the JBL Model L110 10-inch 3-way bookshelf loudspeaker system.

No bypass capacitors.

Techbot
10-23-2004, 06:25 AM
Second generation N110A network as used in the JBL Model L110A 10-inch 3-way bookshelf loudspeaker system.

0.068 uF polypropylene bypass capacitors in parallel with the mylar and N.P.E. main capacitors.

Techbot
10-23-2004, 06:28 AM
Second generation N112 network as used in the JBL Model L112 12-inch 3-way bookshelf loudspeaker system.

The 0.068 uF polypropylene bypass capacitors have been replaced with 0.01 uF polypropylene bypass capacitors.

Techbot
10-23-2004, 06:42 AM
Bypass capacitors have been a JBL standard for over 20 years now.

Top of the line JBL loudspeaker systems such as the L250, 250Ti, and XPL200 employ 0.005 uF polystyrene bypass capacitors in addition to the 0.01 uF polypropylene bypass capacitors.

Polystyrene and polypropylene tend to complement each other with polystyrene typically being more spacious and polypropylene typically being more dynamic.



A short list of some older JBL systems which can benefit from the installation of bypass capacitors include:

4301
4311
4313
4315
4331
4333
4343
4350

L19
L40
L50
L65
L150
L166
L212
L220
L300

Any Consumer or Pro system prior to ~ 1980.

Techbot
10-27-2004, 06:20 AM
In the early 90's JBL began experimenting with biased capacitors. (Greg Timbers [James B. Lansing Sound, Inc.], Ed Meitner [Meitner Audio])

Reference the links below:

http://www.lansingheritage.org/html/jbl/specs/home-speakers/1993-k2-s5500.htm

Specifically:

http://www.lansingheritage.org/images/jbl/specs/home-speakers/1993-k2-s5500/page10.jpg

http://www.lansingheritage.org/images/jbl/specs/home-speakers/1993-k2-s5500/page11.jpg

http://www.lansingheritage.org/images/jbl/specs/home-speakers/1993-k2-s5500/page12.jpg

Techbot
12-03-2008, 12:01 PM
I understand the technical aspects of the CC network and how it works in practice but there is one point that I dont understand at all. JBL compare how it works with Class A verse Class B amplifiers. For those that don't know, Class B crosses the "zero point" whereas in Class A the waveform is basically "pushed" away from the zero point by a fixed DC voltage. In this way the waveform never crosses the zero point so we get a purer sound. The bias viltage to get pure class A operation needs to be higher than the maximum peak input voltage though.

My question is this. If we apply a signal to our xover of say 40 volts, which for arguments sake is 200 watts into 8 ohms, how can a 9volt supply fully bias the main signal? Would it not need to be at least half of the peak to peak input voltage?

To my way of thinking a 9volt supply will only fully bias an 18volt peak to peak signal. Am I correct?


The capacitor biasing is something that has existed for many years. Tube equipment does it automatically since there is usually a large DC offset between stages. Some early transistor amps/preamps had two polarized caps in series with the center point going to ground through a large resistor.

I personally became aware of this technique for speaker systems through communications with Ed Meitner, currently of EMM Labs. He is a wealth of information regarding these "tricks" to help linearize or improve the sound of passive components.

It turns out that the bias trick actually increases measured IM distortion and the higher the bias voltage, the greater the increase. It is not by a great amount, but it is measureable. The sound imporvement (or change) is very rarely perceived as worse and is never linked with a increase in IM distortion. The sonic effect is one of smoothness, increased spaciallity, detail and stuff like that. IM has a muddling or confusing effect so I doubt that this particular steady state measurement is explaining the sound difference either way.

Simply put, we are striving to create a class A situation but as was just pointed out, depending on the bias voltage with respect to the voltage across the capacitor, we may only have an "A" condition up to a particular drive level. So if it makes you happier, consider the change to be class AB, but heavily biased to A. You must also keep in mind that the voltage across the input terminals of the crossover network does not tell you what voltage or current is applied to any individual component. Some parts block signal and others shunt signal so the loading on a particular part is not obvious. For the most part, the caps are well taken care of with 9 volts, even at healthy drive levels. The obvious choice for 9 volts is the small cheap battery and holders that are available. No current is involved so a smoke detector battery and holder is a natural choice.

We did do one system with 18 v (M9500). Certain of the Japanese reviewers thought it was an improvement. I can't personally tell any difference. I am also told on a regualr basis (again by our Asian customers) that the battery must be changed at least every 90 days and that the sound degrades after that. Once again, I have not been able to "hear" any difference after 90 days and the battery is certainly still good for many years from a voltage standpoint.

What playing around I have done with initial application of a battery to a biased circuit (that has not been previously powered up) is that it takes about 3-5 seconds for the soundstage to change. I have tried to measure the voltage level in that time period and it seems that several volts is all that is necessary to accomplish 90% or more of the improvement. Once a circuit has been energized, it is nearly impossible to return it to zero. You have to individually short out each cap and leave them shorted for a while or else they will creep back up somewhat. If you replace the battery with a short and play the system for a while, the caps will start to bias themselves, although not to anywhere near the same degree.

You can take this opinion for what it has cost you. I have been very pleased with biasing for many years. I use it in all applications that involve a capacitor and I have rarely been disappointed. Results may vary so if it doesn't do it for you that is okay too. It cost a bloody fortune to implement as it requires 4 times the capacitance and double the capacitor parts count. The network size gets huge as well. Inspite of this, I have never heard a capacitor type that didn't improve (or change) including the nearly perfect teflon variety..