Charge Coupled Networks

[Robh3606]

Hello

Can't answer the question but for what its worth the 9800 uses a Charge Coupled network that uses transcaping. The 4344 2 uses a CC network without?? I am going to build them with them.

Rob

[Earl K]

Hi mrdsp

I know you asked for references for "transcapping" , sorry, I can't help you there. For a reasonable laymans explanation of the benifits of charge-coupling, go and read the JBL K2 S5500 Product Overview . The overview info found there, "sells" the concept quite well. With Class A amplifiers being back in vogue, JBLs' "balanced capacitor" approach is in synch with current wisdom on the subject.

regards <. Earl K

[4313B]

Originally posted by mrdsp
Help... Giskard.

Do you have any references which can explain in simple terms the theory behind transcapping. Is this an earlier attempt to deal with "limitations of various dielectrics"? On the surface it would seem that transcapping is a variation of bypassing a capacitor with a smaller one in order to negate the effects of series inductance (ESL) and/or resistance (ESR) of the larger device. If so it would then seem that charge coupling would greatly increase the ESL and/or ESR , yet the results are reported to be better. One might ask the question whether paralleling that charge coupled pair with a pair of 0.02 uF capacitors in its own charge coupled configuration would improve things further. Thanks!

Well, JBL literature is full of little blurbs about bypass capacitors. Basically the large main capacitors (metallized mylar and NPE) introduce a slight time lag (hysterisis/dielectric absorbtion). The much smaller bypass capacitors (polypropylene or polystyrene) are much more linear and help to greatly reduce this. The theory was borrowed from advanced circuit design and was first incorporated in the L110A. Nearly all JBL loudspeakers since the L110A have incorporated this topology as SOP. Some of the statement systems, such as the 250Ti, used both polypropylene and polystyrene to increase both dynamics and spaciousness. There was a website that showed the differences in linearity between the different dielectrics but I can't seem to find it any longer. In any case, teflon, polypropylene and polystyrene are among the most linear while mylar and NPE's are lower down on the list.

Expense and size are also of consideration. A good metallized mylar main capacitor (usually superior to just mylar (polyester)) bypassed with a polypropylene or polystyrene is a very cost effective way to achieve nearly the same performance of a high quality polypropylene main. It's also a smaller package.

These days there seems to be as many different dielectrics and differences in quality between the same dielectrics as there are theories on how to use them. Some expand on the bypass theory and think cascading is the best solution. Others feel that the finest film and foils completely negate the need for bypassing at all. Quality is another issue. Just because a capacitor is film and foil doesn't necessarily mean it is automatically superior to metallized. One brand of metallized polypropylene may be preferred over another brand's more expensive film and foil, etc. The various combinations of dielectrics either work or don't work, and only experimentation will dictate which falls in which category.

This is one reason why I have advocated simply using the stock JBL bypass capacitors. No, they aren't the best that are out there but they have been chosen to offer the best price/performance with the stock mylar and NPE capacitors JBL has used over the years. In other words, the R&D has already been done. They are the most economical in both time and money. A person embarking on a mission to find a better solution has to be aware of the time and money involved. Not every polystyrene sounds the same, not every metallized polypropylene sounds the same, etc.

As for charge coupled theory, Earl has posted the link for a decent cover of the subject.

[4313B]

Originally posted by Robh3606
Hello

Can't answer the question but for what its worth the 9800 uses a Charge Coupled network that uses transcaping. The 4344 2 uses a CC network without?? I am going to build them with them.

Rob

Hi Rob,

After further conversation it has been suggested that one start off using polypropylene mains without bypass capacitors. If one HAS to use mylar or NPE, those will have to be bypassed. If one HAS to use NPE, at least 10% of the total value should consist of polypropylene.

[Robh3606]

High Rob

OK Thanks

Rob

[4313B]

I've been looking at some of the large format 4-ways and the capacitor values get daunting

One thing I would like to revisit is a charge coupled high pass on a 2405/077. Those things respond so well to polypropylene and polystyrene bypass caps the charge coupled approach could be very interesting. Maybe I'll have to toss a pair of 077's at a pair of LE8T-H's just like old times

[4313B]

Oh, I also forgot to add that it is probably best to use the capacitors at least for 48 hours before installing the battery. This allows them to "break in" and also allows one to become quite familiar with the program material one will be using for comparison. Once that period is over and one connects the 9V battery one should begin to hear a difference as the battery charges the capacitors. The resistors used should be 2.0Mohm to 6.0Mohm. The capacitors need to charge slowly. The batteries should last as long as their shelf life since no current is being drawn from them.

Some people have stated that series capacitors are "more important" than parallel capacitors, in "real life" this is false. So charge coupling only the series capacitors and leaving the parallel capacitors "normal" isn't a brilliant idea.

[mrdsp]

Per the groups recommendation I read the K2-S5500 discussion about charge coupling. I was aware of the piezoelectric nature of capacitors, but from a reciprocal point of view. I have had ceramic capacitors induce tiny voltages from microphonic effects (ala condenser microphones) in feedback loops for amplification stages which is why I don't use them for that application. I was not aware that voltage biasing them could make film capacitors be and sound "tighter" (re: K2-S5500 brochure). In my speakers the crossovers are mounted in the woofer area. I can certainly imagine that it gets quite noisy from the capacitors point of view and that the vibrations may affect its physical "tightness". It makes me wonder if I should attempt to isolate the crossover in a sound tight box or am I making a mountain out of a molehill?

[4313B]

I think the network is isolated in the K2-S9800.

[MJC]

I will probably start out charge coupling a pair of stock JBL networks that employ the standard metallized mylar and non-polarized electrolytic capacitors bypassed with polypropylene and polystyrene topology such as used in the original 250Ti. (polypropylene excels in speed and dynamics, polystyrene excels in spaciousness and three-dimensionality)

What website did you use to purchase polystyrene capacitors?

[4313B]

http://www.capacitors.com/index.html
http://www.welbornelabs.com/rel.htm
http://www.percyaudio.com/
http://www.soniccraft.com/
http://www.partsconnexion.com/
http://www.hndme.com/
http://www.partsexpress.com/

Look for AudioCap Polystyrene-&-Tin Foil Series RT or MultiCap Polystyrene Series RTX

I think you can still get 0.005 uF polystyrene capacitors from JBL Pro Parts

[Ian Mackenzie]

I am in the process of ordering some Solen capacitors to build a 3145 Equivalent Charge-Coupled Network per the recommendations above.

I will post progress on the project in due course.

Ian

[MJC]

I am in the process of ordering some Solen capacitors to build a 3145 Equivalent Charge-Coupled Network per the recommendations above.

I will post progress on the project in due course.

Ian

My experience with using Solens in a CC network produced very good results, indeed.

[Robh3606]

Bump

Rob

[badman]

http://www.madisound.com/catalog/pro...oducts_id=1463


I just ordered 100 of these, as a 10uF connection will require 4, and a 20uF 8, and so forth- I anticipate being able to burn through them pretty quickly. Naturally they're not useful for smaller values, but the potentially massively paralleled larger values should have some performance benefits outside of the charge coupled dielectric.

All values will have greater power handling than the smaller cap without charge coupling (2 larger caps with 2x the plate area each=6dB more power)., and anything over the nominal value of the individual parts used for the series/parallel network will begin to see improvements in ESR and ESL as well.

I'm looking forward to giving it a try, I have a lot of caps invested in notch filters and such.