"Charged Coupled" crossovers?

[pierce]

I've seen numerous references here to 'charged coupled' crossovers. When I first saw this, I was thinking "huh, CCD based delay lines or something?" but I've since seen a couple of schematics, indicating they are really just DC biased capacitors.

I've read various threads like... http://www.audioheritage.org/vbulletin/showthread.php?t=43

Frankly, with my somewhat naive understanding of analog electronics and filter networks, I don't get it, and would love to see some discussion of the theory behind why it works. URLs to reading material are fine, too.

[4313B]

Frankly, this is all you need to know:

Bypassed and Biased Capacitors (http://audioheritage.org/vbulletin/showthread.php?t=3555)

Read Greg's various explanations about it in his product papers as linked above.

At this point all we have to say is bias or don't bias, do whatever you want, but leave us out of it. It's remarkably simple. :)
No emails, no phone calls, no PM's. Thank you and have fun! :)

[John]

Frankly, this is all you need to know:

Bypassed and Biased Capacitors (http://audioheritage.org/vbulletin/showthread.php?t=3555)

Read Greg's various explanations about it in his product papers as linked above.

At this point all we have to say is bias or don't bias, do whatever you want, but leave us out of it. It's remarkably simple. :)
No emails, no phone calls, no PM's. Thank you and have fun! :)


Who is we and us ???:blink:

[4313B]

Who is we and us ???:blink:
Whenever this topic comes up someone sees fit to bother Greg or myself. I think we just went through this once again four months ago. I'm not real sure why it keeps coming up. He has graciously offered up everything he has and feels that should be sufficient.

If this is just another thread to yammer on about it and it won't result in emails, PM's and phone calls asking why this topology works then by all means carry on. I'm not trying to be rude, I'm simply trying to put an end to all the behind the scenes discussions before they even start this time around.

My earlier post was made in haste, as they often are, and I've been told that puts people off. I apologize for that. From my perspective, unless someone has actually charge-coupled some networks and given them a thorough listen they very probably have nothing of interest to offer on the subject. Challenging posts and PM's such as those that have arisen in the past are just trolling as far as I'm concerned.

[John]

Whenever this topic comes up someone sees fit to bother Greg or myself. I think we just went through this once again four months ago. I'm not real sure why it keeps coming up. He has graciously offered up everything he has and feels that should be sufficient.

If this is just another thread to yammer on about it and it won't result in emails, PM's and phone calls asking why this topology works then by all means carry on. I'm not trying to be rude, I'm simply trying to put an end to all the behind the scenes discussions before they even start this time around.

My earlier post was made in haste, as they often are, and I've been told that puts people off. I apologize for that. From my perspective, unless someone has actually charge-coupled some networks and given them a thorough listen they very probably have nothing of interest to offer on the subject. Challenging posts and PM's such as those that have arisen in the past are just trolling as far as I'm concerned.

I agree on all the points you have laid out, thanks. :)

And as far as the CC networks go, the proof is in the pudding. :blah:

Thanks Giskard and Greg.:applaud:

[pierce]

Frankly, this is all you need to know:

Bypassed and Biased Capacitors (http://audioheritage.org/vbulletin/showthread.php?t=3555)

Read Greg's various explanations about it in his product papers as linked above.

indeed, that was what I needed to know. dielectric hysteresis..

don't worry, I won't PM you :)

I am curious how many millivolts and/or microseconds this hysteresis is for the normal sorts of nonpolar caps used in these crossovers, but I'm sure I can find some technical component data on this.... And, as I saw someone else asked on some other thread, it does strike me that the bias voltage of the classic circuits has to be at least as high as 50% of the peak to peak voltage, which might be as high as 50 or 60 volts for high wattage transients from a big amp (assuming the cap is directly across or in series with the load)

[pos]

some interesting info:
http://www.diyaudio.com/forums/showthread.php?threadid=141929

[4313B]

some interesting info:Thanks for the link. :coolness:

*****

I see that it was started 03-31-2009. Interesting that it can go on for 68 posts. I only did a quick scan but it looked like the first few posts were worthwhile and then everything up to post 68 could be whacked as typical Internet noise. It'd be nice to see more posts along the lines of 68. My experience though is that people just do it and then enjoy the results. It's curiously difficult to get them to post something on an Internet forum.

One question jumped out and is exactly the question Greg posed to me several years ago during one of our conversations.

But why is it not more common?

He was rather surprised that competitors hadn't jumped on it. I think Ian has since pointed out that some other manufacturers are doing it.

In any case, I personally bias the entire network including any shunts. In this instance I don't worry about cost or space issues.

Here's one of the latest pair I need to finish wiring up. The boxes are finely crafted by another forum member for his 4355's. Maybe someday he'll post about them. :dont-know

[Loren42]

So, this improves the phase delay for capacitors, I am thinking, but what about the inductors, which are the worst offenders in a passive crossover network? DC bias would not help that.

I guess anything is an improvement.

[4313B]

True. Go active if the inductors are bothering you.

[robertbartsch]

I have a pair of vintage JBL speakers with charge-coupled X-overs and I am not impressed. I'm about to rip the x-overs out and replace them.

[hjames]

I have a pair of vintage JBL speakers with charge-coupled X-overs and I am not impressed. I'm about to rip the x-overs out and replace them.

How about some specifics to go with your accusatory anecdotal info??



Doesn't surprise me ...
You're the guy that tried to re-engineer the Crown with the fan to bypass it and wasn't impressed when it didn't work right ...
;)
Seems you get upset a lot ...

[4313B]

I have a pair of vintage JBL speakers with charge-coupled X-overs and I am not impressed. I'm about to rip the x-overs out and replace them.You effed up dude! You should have made them external like I did as shown above so you could just shove them off the top of the enclosures onto the floor. I don't have to pull any drivers.

[pierce]

So, this improves the phase delay for capacitors, I am thinking, but what about the inductors, which are the worst offenders in a passive crossover network? DC bias would not help that.

I guess anything is an improvement.

air wound inductors should have little to no reluctance... at least compared with ferrite or other such cored inductors ? At least, thats my somewhat naive understanding. If you take away the reactance thats left, its not an inductor anymore... :dunno:

[Zilch]

I only did a quick scan but it looked like the first few posts were worthwhile and then everything up to post 68 could be whacked as typical Internet noise.Somehow, I believe you'd leave Rob's post, too, though there may be merit in not inviting more noise to move over here.

[Those guys are sure good at noise generation in lieu of the "Do" part of DIY.... ;) ]

[4313B]

Somehow, I believe you'd leave Rob's post, too.... ;)Yeah, I saw Rob in there fighting the good fight... ;)

He's a better man than I am though. I wouldn't set foot in that forum without a case of hand grenades, a straight razor and a bottle of tequila.

[pierce]

You effed up dude! You should have made them external like I did as shown above so you could just shove them off the top of the enclosures onto the floor. I don't have to pull any drivers.

and, external coincidentally avoids any 'microphoning' effects on the bigger caps induced by high woofer SPLs, I bet. :deal:



I wouldn't set foot in that forum without a case of hand grenades, a straight razor and a bottle of tequila.

If thats Don Julio, I'll cover your back!
ok, ok, I'm *so* outgunned in these forums, all I'd be good for would be deflecting bad jokes like this...:baby:

[Loren42]

air wound inductors should have little to no reluctance... at least compared with ferrite or other such cored inductors ? At least, thats my somewhat naive understanding. If you take away the reactance thats left, its not an inductor anymore... :dunno:

They do have impedance among other things.

[pierce]

They do have impedance among other things.

well, impedance is the sum of AC reactance, and DC resistance, and of course, is a rather complex (pun intended) entity, as its quite variable with frequency :deal:

I repeat my statement... if you remove the reactance, its not an inductor anymore...

[Ian Mackenzie]

If you are familiar with the sonic virtues of SE class A sound there certainly are analogous effects to Charge-Coupling capacitors in loudspeaker applications.

Of course the merits of this technique become fully apparent when you use SE biased amplifiers as I do ie Passlabs X250.5 (this amplifier uses a trickle of SE bais in addition to significant push-pull class A bias operation, the result is superior high frequency performance and imaging...the 2405/077 never sounded so good)

[Russellc]

I have a pair of vintage JBL speakers with charge-coupled X-overs and I am not impressed. I'm about to rip the x-overs out and replace them.
Hmmmmm. I guess you should be prepared to be even slightly more unimpressed. ;)

Russellc

[jblsound]

If anyone has any doubt about C-C, build one normal XO and a C-C XO for the same speaker. Then listen to them in near field.

The first set of LCRs (L212) I built, I built the normal XO using many JBL original parts(new). Two years later I built a pair of C-C XO for the L/R speakers. I then I replaced the XOs in the first pair of L212s I bought in '79 with the first set of XOs I built.

Well for a period of time I had those original L212 wall mounted in this room, they were a bit harsh in the mid-range. For surround duty they were quite good.
At this time I have the pair with the C-C XOs wall mounted in this room, and they are as smooth as silk.

[pierce]

so how come the biased cap designs I've seen haven't also had bypass caps?

[/me ducks and runs laughing maniacally..... is it Friday yet? :p]

[4313B]

so how come the biased cap designs I've seen haven't also had bypass caps?

[/me ducks and runs laughing maniacally..... is it Friday yet? :p]I guess you just don't know the history? :dont-know

[jblsound]

so how come the biased cap designs I've seen haven't also had bypass caps?

[/me ducks and runs laughing maniacally..... is it Friday yet? :p]
If all the caps in a C-C network are Polypropylene there's no point to adding bypass caps.
As the reason JBL started using poly bypass was the main caps were mylar caps. And adding the poly bypass makes the sound more dynamic.

One exception might be for using polystyrene bypass to make the highs more open.

[Steve Schell]

I finally got around to trying charge coupling a few months ago and was immediately impressed with the results. I added my thoughts to a thread on the Audio Asylum:

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=141098&highlight=charge+coupled

[spwal]

A few words from a complete non-techie who just loves his speakers:

Fwiw, I love my CC network. It is external and wood mounted with adjustable knobs and all Solen caps.

It is made for the 4345 I believe, but I am having no issues at all running it with my 4341.

I don't feel the need to mess with the knobs 90+% of the time from the 12 o'clock position (neutral).

One question that I know has been answered before -- how often do you change the 9V batteries????

And how about another thank you to the forum -- I wouldnt have such a wonderful sounding stereo if it wasn't for everyones help and input! I don't come here enough these days... (off the audio upgrade merry go round).

I am reluctant to link hi res pics of the crossover and my setup in a public forum, but if you are curious to see pics, just send me a pm and I will shoot them over to you.

[pierce]

One question that I know has been answered before -- how often do you change the 9V batteries????

the obvious answer is, when they aren't outputting 9V anymore. I'd bet they are good for 3-4 years assuming the series resistor is on the order of several megohms as I've seen in most of the diagrams. I'd use a volt meter on them every so often to confirm they are still OK

[spwal]

thanks

[spkrman57]

Usually the shelf life of the battery is the same as with several million ohms of resistance in series.

I would check them once a quarter to verify the voltage to be sure.

Also, the capacitors when hooked up the 1st time will need some time to fully charge to their potential. Sometimes that may take up to 24 hrs to maximum performance. Also, removing the battery will take about the same time to notice the lack of bias.

I hope I'm right with this information.

Regards, Ron

[pierce]

Usually the shelf life of the battery is the same as with several million ohms of resistance in series.

I would check them once a quarter to verify the voltage to be sure.

Also, the capacitors when hooked up the 1st time will need some time to fully charge to their potential. Sometimes that may take up to 24 hrs to maximum performance. Also, removing the battery will take about the same time to notice the lack of bias.

I hope I'm right with this information.

Regards, Ron

a capacitor is charged to 63% after R*C seconds, so 2.2 megohms into a 20uF capacitor (typical for what I've seen) would be 63% charged in 44 seconds. another 44 seconds would charge 63% of the remaining 37%, or 87% in 88 seconds. another 44 secs makes it to 95% (in about 3 mins total) whereupon we're deep into the diminishing returns (95% of 9V is 8.55V, which should be sufficient bias...)

:deal:


the 72uF caps in that design of the k2.5500 above would take almost 4 times this long, but still not hours.

[Hoerninger]

... but still not hours.
You have not taken into account the strange behaviour of the electrones at the surface of the foil. It will take more time for the full effect than the classical physics want to tell you.

Btw. the nominal 9 volt batterie should not have less than 7 volts.
____________
Peter

[pierce]

You have not taken into account the strange behaviour of the electrones at the surface of the foil. It will take more time for the full effect than the classical physics want to tell you.

I do not believe in Magic.

[Hoerninger]

I do not believe in Magic.
It is not magic but solid-state physics.

a capacitor is charged to 63% after R*C seconds ... (and so on)This law can't describe it all as there is only a definite number of charges so loading can not go on undefinitely.

Further the dielelctricum (I called it foil) is regarded as an isolator, right? But when the voltage is too high than there is current. It is acceptable that there is more to the charges in the dielectricum than simply polarisation.

Charge coupling demonstrates that there is an influence in a very small amount of voltage when the voltage crosses zero volts. This is audible at low levels but not at louder ones. 9 volts seems to be sufficient.
____________
Peter

[Earl K]

I do not believe in Magic.

- IME, it's best to just implement the topology ( it's certainly cheap enough ).
- Delay trying to answer the "why" questions till later / then ones' personal observations can contribute something meaningful to the discussion .

>< cheers :)

[Allanvh5150]

I would have to agree with EarlK. Thinking about it too much you will just come up with ideas why it won't work. The major hurdle for me, and I have asked why and I got a resonable explanation, was the fact that a 9v bias would only work perfectly if the input voltage was below 9v. One would think that it would be prudent to raise the bias level abouve what the maximum input voltage would be. I would like to try a higher voltage but I think the additional benefits would be very small. Also, if something goes wrong with the crossover, the 9v source has much less of a chance at killing you drivers. Just try it. It does work.

Allan.

[jblsound]

I would have to agree with EarlK. Thinking about it too much you will just come up with ideas why it won't work. The major hurdle for me, and I have asked why and I got a resonable explanation, was the fact that a 9v bias would only work perfectly if the input voltage was below 9v. One would think that it would be prudent to raise the bias level abouve what the maximum input voltage would be. I would like to try a higher voltage but I think the additional benefits would be very small. Also, if something goes wrong with the crossover, the 9v source has much less of a chance at killing you drivers. Just try it. It does work.

Allan.
According to Greg Timbers, no matter what the input voltage is no single conponent in the XO will have a voltage more than 9 volts. I believe that post is in this thread a year or two ago. Tho it might be a different thread.

[Steve Schell]

As I understand it the input signal peak to peak voltage would have to be twice the polarizing voltage, or 18V in the case of 9V bias, before the signal would swamp the bias. One direction of swing will momentarily increase the bias while the other direction will decrease it. If the input signal happened to reach or exceed 18V PtoP, all that would happen would be that one side of the signal peaks would be subject to the zero crossing nonlinearities that have been avoided with the bias. It is doubtful that this would be audible. In fact, if any of the speakers I listen to regularly receive an 18V swing I don't want to be anywhere close!

[pierce]

my comment about Magic was in reference specifically to Hoerniger's statement ...


You have not taken into account the strange behaviour of the electrones at the surface of the foil. It will take more time for the full effect than the classical physics want to tell you.

odd how a foil which can change polarity at megahertz speeds would have a memory for a dc charge that required conditioning it for hours. :deal:

I have no issues with the basic concepts of biasing capacitors... And I can see how the dielectric hysteresis that this bias circumvents are only an audible issue for small signals, that if you have a large enough signal to exceed 9V peak across the filter caps, the tiny amount of distortion would be insignificant.

[Robh3606]

With a DC bias why would the cap loose the bias due to an AC signal??? AC and DC have a different set of rules. You also have very long time constants involved. I don't see how you could swing the cap through 0 volts with a 10ms peak no matter what the voltage when you have a 20 second time constant to work against. That battery is always there so it is always trying to charge the capacitor. The AC see's the battery and resistor as an open and the load as a couple of ohms. It's no surprise where the AC is going.

Rob:)

[pierce]

With a DC bias why would the cap loose the bias due to an AC signal??? AC and DC have a different set of rules.

actually, AC and DC follow exactly the same set of rules, its just that AC is changing in time while DC isn't, so this complicates the circuit analysis, and takes it from simple arithmetic into the realm of calculus and differential equations when you add reactive signal elements like inductors, capacitors, transformers (and, of course, at a high enough frequency, *everything* has reactance)..

the DC 'bias' is added to the AC signal. if the AC signal is +/- 5 volts, and your DC bias is +9V, the net affect is a AC signal that ranges from +4 to +14 volts. if the AC signal is +/- 12 volts with the same 9V bias, then your net signal will be -3 to +21 volts.

[Robh3606]

actually, AC and DC follow exactly the same set of rules,

You can pass DC through a capacitor or a transformer?? What about the time constant to charge the cap?? Does it just go away when you apply an AC voltage??

Rob:)

[pierce]

You can pass DC through a capacitor or a transformer?? What about the time constant to charge the cap?? Does it just go away when you apply an AC voltage??

you cut off the interesting part...


its just that AC is changing in time while DC isn't, so this complicates the circuit analysis, and takes it from simple arithmetic into the realm of calculus and differential equations when you add reactive signal elements like inductors, capacitors, transformers (and, of course, at a high enough frequency, *everything* has reactance)..

[Robh3606]

its just that AC is changing in time while DC isn't, so this complicates the circuit analysis, and takes it from simple arithmetic into the realm of calculus and differential equations when you add reactive signal elements like inductors, capacitors, transformers (and, of course, at a high enough frequency, *everything* has reactance)..


When you look at a circuit you look at it for both AC and DC analysis. For example you can have multiple sections in an amplifier that are capacitively coupled but are electrically isolated for DC and work under different DC voltage and bias conditions.

Maybe I am wrong but I think you are looking at it as if the AC and DC are the same or act the same in the same circuit. If you do a simple DC and AC analysis to a charge coupled capacitor circuit they look very different.

In your example you have the AC voltage crossing zero volts. What is the DC voltage doing?? Seems to me it's not moving, based on simple addition or subtraction used in the example. If the DC bias doesn't change the capacitor never goes to zero volts.

Rob:)

[pierce]

In your example you have the AC voltage crossing zero volts. What is the DC voltage doing?? Seems to me it's not moving, based on simple addition or subtraction used in the example. If the DC bias doesn't change the capacitor never goes to zero volts.


AC is volts changing over time, while DC is constant volts. volts is volts.

Electronics engineers tend to prefer to simplify things and treat DC and AC as distinctly different things, it certainly makes the math more tractable, but in fact, there's a continuum of behaviors, as the frequency approaches zero, the AC approaches DC :deal:

[Steve Schell]

Rob, the DC polarizing voltage and AC signal voltages are both imposed on the biased capacitor plate, and its potential at any given instant is their sum. This situation is very similar to the DC biasing voltage applied to a vacuum tube grid. The audio signal is superimposed on this potential, and when the signal swings more positive than the grid is biased negative then the grid potential becomes positive in relation to the cathode and the grid draws current from the cathode. If DC and AC are so different and cannot coexist in the same circuit then how does grid current occur?

[Robh3606]

If DC and AC are so different and cannot coexist in the same circuit then how does grid current occur?

Hello Steve

I didn't say they couldn't coexist. All I am saying is they act differently in the same circuit. If they didn't act differently how would you get an amplfier to work?? Once you set-up the DC bias on a transistor it just sits there and does nothing. It takes an AC signal to get things going and the AC signal is what gets amplified. Obviously without the DC bias it couldn't work. The DC in this case is the power source and the AC the signal.

When you get the tubes all tuned up and working does the AC signal negate the DC bias?? The DC bias stays as set and power supply does it's best to maintain a constant DC voltages to the tubes and bias voltage all set by the resistors. This happens independent of what signal is passing through the amplifier.

Once you have that capacitor charged its just like a DC voltage source for at least the first 2 time constants. Just like the power supply capacitors in an amplifier. When you run an AC signal through the capacitor will the DC bias just disappear?? The AC may change with reference to ground, 0 volts because it is alternating, but why would the DC voltage change with respect to ground??

If you look at the previous example of say 12 peak or 24 peak to peak you would get what +9 and 12 so +21 volts on the low side you have -12 volts plus +9 so -3 volts. But that's the negative peak AC voltage not the +9 volts DC that has not changed. The AC voltage is riding the DC rail but the rail voltage isn't changing. It's still +9 volts. It can't change any where near as fast as the AC because you have the time constant of C x R which in this case is measured in seconds. Just like in a good stiff power supply. Figure a 10uf times 2.2 meg is what 22 seconds so 66 seconds for 3 time constants. That time constant is only there for a DC voltage.

With an AC signal in the capacitors pass band it looks like a short. To DC it's an open so the AC passes freely and DC is blocked by the capacitor. Once those caps are charged that DC voltage has no place to go. The only way to get rid of it is to manually discharge them, with low leakage in the poly caps they will stay charged for quite some time if left alone.

Well that's my take anyway. What do you think??

Rob:)

[Steve Schell]

Rob, I think my statements to this point in the discussion are at least basically correct. I believe that you are overthinking what is a fairly simple matter, but I don't have the energy for a prolonged debate. Electrical engineering was never my strong subject anyway. Does someone else want to jump in and help Rob determine if +9VDC combined with -9VAC equals something other than zero volts?

[pierce]

+9VDC and a +/-9V AC signal (aka 18V P-P aka 6.4V RMS) would equal a 0-18V signal. you can call that what you want, AC and DC are just names and abstractions.

its all volts, its just that some of the volts are changing, and others are constant. if you add a constant to a changing value or function, you get a changing value whats got a different offset. :deal:

[Zilch]

http://www.audioheritage.org/vbulletin/showthread.php?p=130034&#post130034

197 posts in that thread alone, 2.5 years ago, 14 pages.

[We REALLY need to get past this.... ;) ]

[Allanvh5150]

As I understand it the input signal peak to peak voltage would have to be twice the polarizing voltage, or 18V in the case of 9V bias, before the signal would swamp the bias. One direction of swing will momentarily increase the bias while the other direction will decrease it. If the input signal happened to reach or exceed 18V PtoP, all that would happen would be that one side of the signal peaks would be subject to the zero crossing nonlinearities that have been avoided with the bias. It is doubtful that this would be audible. In fact, if any of the speakers I listen to regularly receive an 18V swing I don't want to be anywhere close!

18V P-P is only 20watts RMS into 8 ohms. My TV has bigger amplifiers...:)

Allan.

[Allanvh5150]

Rob, I think my statements to this point in the discussion are at least basically correct. I believe that you are overthinking what is a fairly simple matter, but I don't have the energy for a prolonged debate. Electrical engineering was never my strong subject anyway. Does someone else want to jump in and help Rob determine if +9VDC combined with -9VAC equals something other than zero volts?

Yup. I will jump in and agree with Rob. 9VDC doesn't combine with 9VAC. The DC sits there and does basically nothing but hold the capacitor above OV so it always has a charge on it. The AC voltage simply flows past with the impedence of the capacitor decreasing with increasing frequency. At DC the capacitor presents a very high impedence. The AC and DC ALWAYS remain separate. If they could somehow combine I have the feeling that the DC would flow through the capacitor as well. Probably not a good thing. I cant even imagine what would happen if the DC bias on the grids of my tube amps somehow combined with the drive voltage.....

Allan.

[Allanvh5150]

+9VDC and a +/-9V AC signal (aka 18V P-P aka 6.4V RMS) would equal a 0-18V signal. you can call that what you want, AC and DC are just names and abstractions.

its all volts, its just that some of the volts are changing, and others are constant. if you add a constant to a changing value or function, you get a changing value whats got a different offset. :deal:

hmmmmm.......I think 9VDC is actually 9V RMS. Calculations made for DC use basic Ohms law. Calculations only take into account Volts, Current and Resistance and Power is thrown in for good measure. AC calculations, on the other hand, take into account Frequency and reactance. AC and DC are totally different animals and never the twain shall meet!

Allan.

[Robh3606]

Does someone else want to jump in and help Rob determine if +9VDC combined with -9VAC equals something other than zero volts?

Hello Steve

Well from my standpoint you are trying to oversimplify what's happening and not taking the time to think it through.



Rob:)

[4313B]

[We REALLY need to get past this.... ;) ]I'm way past it to the tune of a dozen already built and several dozen left to build. I flat out can't build them fast enough. They rock! I'm quite glad Greg saw fit to hook me up.

He wants to make it clear though that he really doesn't care whether or not anyone coats their diaphragms or biases their networks. They aren't some kind of requirements for the general public. These are things he does and these are things he's shared with me. If I'd known the end result I'd never have mentioned any of it on a public forum.

It could have been pretty fun, instead it's been kind of a drag.

[jblsound]

I'm way past it to the tune of a dozen already built and several dozen left to build. I flat out can't build them fast enough. They rock! I'm quite glad Greg saw fit to hook me up.

He wants to make it clear though that he really doesn't care whether or not anyone coats their diaphragms or biases their networks. They aren't some kind of requirements for the general public. These are things he does and these are things he's shared with me. If I'd known the end result I'd never have mentioned any of it on a public forum.

It could have been pretty fun, instead it's been kind of a drag.

I think people can tend to over think way too much, and not take it for what it is. C-C is a huge improvement to the basic crossover. Period. It works.

If anyone doubts that, try my previous suggestion. Setup speakers for near field first with the basic and then w/bias and hear the difference.

[pierce]

hmmmmm.......I think 9VDC is actually 9V RMS.

RMS is normally used on AC voltages, it stands for Root Mean Square. for a DC, or unchanging constant voltage, RMS would be somewhat meaningless, and in fact, any DC offset (bias) is usually removed from a signal before measuring its RMS voltage.

For a pure sinusoidal AC voltage, the RMS voltage is equal to .707 of the peak voltage (eg, 1 over the square root of 2)... So, a 18 V P-P sine wave, which goes from +9V to -9V, would be about 6.4V RMS (9/sqr(2)). This represents the integral of the absolute value of the voltage over time, eg, the average of the absolute value of the instananeous voltage, and is what you would get if you full wave rectified the AC signal (aka ABS(V)), and then filtered it for DC.


Calculations made for DC use basic Ohms law. Calculations only take into account Volts, Current and Resistance and Power is thrown in for good measure. AC calculations, on the other hand, take into account Frequency and reactance. AC and DC are totally different animals and never the twain shall meet!

AC follows Ohms law too, V=I*R, and watts law (P=V*I), etc. Its just that AC is constantly changing in time, and some of the Impedances (those with non-zero Reactance, ie capacitors and inductors) are dynamic dependent on frequency, and introduce time delays where current lags voltage, and so forth, which leads to interesting things like resonant circuits from which oscillators, filters, etc can be derived. Z = (R + j*X), Impedance is a Complex number if reactance (X) is non zero. A purely DC world would be very very boring :banghead:


side note: none of this has anything to do with how 'charge coupled capacitors' based crossovers sound to the ear... this is a purely pedantic discussion where I'm trying to correct what I percieve as some common misconceptions about fundamentals of electronics. I tend to take a Physics approach to all this, rather than that of a practicing electronics designer, and I think Ive stated my exact same point ni about 5 different ways so far in this thread, and I should probably let it rest :deal:

[Zilch]

18V P-P is only 20watts RMS into 8 ohms. My TV has bigger amplifiers...:)It's the first watt that counts.

[Especially when running compression drivers.... :p ]

[Steve Schell]

4313B, I'm pleased with the discussion thus far, and if it exposes more folks to the charge coupling concept or leads to a better understanding then so much the better. Some seem to be quite skeptical, which is okay. Based on my experiments and listening I'm quite biased in favor of this technique, +9 volts at least.

[Zilch]

Timbers discussed experimenting with higher bias voltages.

Bottom line: "Unnecessary."

[Allanvh5150]

Timbers discussed experimenting with higher bias voltages.

Bottom line: "Unnecessary."

18v would be better technically but in this world of dimishing returns, would anyone hear the difference? I think not....

Allan.

[Russellc]

Usually the shelf life of the battery is the same as with several million ohms of resistance in series.

I would check them once a quarter to verify the voltage to be sure.

Also, the capacitors when hooked up the 1st time will need some time to fully charge to their potential. Sometimes that may take up to 24 hrs to maximum performance. Also, removing the battery will take about the same time to notice the lack of bias.

I hope I'm right with this information.

Regards, Ron
I seem to have noticed this as well. Initially they sounded a little fuzzy, but snapped into focus. This was during a build on another forum, and when I commented positively, the general reaction was like a 16th century witch hunt! Mostly from people that never tried it, just had a million reasons why it "wouldnt work". I think it does!

Russellc

[Uncle Paul]

Charging time is a function of voltage, resistance, and capacitance.

There are several calculators for this including this one: http://www.csgnetwork.com/rctimecalc2.html

To give you an example of charging time, it takes ~27 seconds to charge a 6 MFd capacitor in series with a 1 M Ohm resistor to ~9V using a 9V battery.

[Uncle Paul]

I'm way past it to the tune of a dozen already built and several dozen left to build. I flat out can't build them fast enough. They rock! I'm quite glad Greg saw fit to hook me up.

He wants to make it clear though that he really doesn't care whether or not anyone coats their diaphragms or biases their networks. They aren't some kind of requirements for the general public. These are things he does and these are things he's shared with me. If I'd known the end result I'd never have mentioned any of it on a public forum.

It could have been pretty fun, instead it's been kind of a drag.

4313B, I for one am grateful for what you have shared with us. I built a few CC and non-CC L200T3 networks to compare when Zilch was doing the Q&D fest. I don't think I'd do a non-CC network for any speaker system that used decent drivers. CC networks do rock, as do so many of the things you have shared.

[Allanvh5150]

Charging time is a function of voltage, resistance, and capacitance.

There are several calculators for this including this one: http://www.csgnetwork.com/rctimecalc2.html

To give you an example of charging time, it takes ~27 seconds to charge a 6 MFd capacitor in series with a 1 M Ohm resistor to ~9V using a 9V battery.

Yup. I thought that as well but others may not agree. Some people also beleive that cables need to burn in for a few weeks as well.............

Charge couple the suckers and if it aint better, send me the bill.

Allan.

[pierce]

Yup. I thought that as well but others may not agree. Some people also beleive that cables need to burn in for a few weeks as well.....

indeed, many people believe in Magic, as defined by Arthur C Clarke: Any sufficiently evolved technology, meaning, something beyond the understanding of the observer. :deal:

Magicians perform ritualistic tasks hoping to achieve their goal without understanding why. Many technicians are essentially magicians as they lack a fundamental understanding of the systems they work on, and instead rely on repeating what they know worked before. :blah:

To most people, these computers we use are nothing short of magic, as is their consumer electronics, all they know is they push a button and sounds or pictures appear. As a kid in the 1960s, I had to know how everything worked, took stuff apart to see what made it tick, built Heathkits, read endless books on science and technology. Both my teen kids are perfectly happy to just accept stuff without understanding its inner workings, this saddens me. Admittedly, things like modern HDTV are bogglingly complicated, with layers and layers of nearly incomprehensible technology, all of which must work in perfect unison to deliver that picture into your living room...

[Hoerninger]

http://en.wikipedia.org/wiki/Superposition_theorem

This procedure is followed for each source in turn, then the resultant responses are added to determine the true operation of the circuit. The resultant circuit operation is the superposition of the various voltage and current sources.
____________
Peter

[Ian Mackenzie]

I've seen numerous references here to 'charged coupled' crossovers. When I first saw this, I was thinking "huh, CCD based delay lines or something?" but I've since seen a couple of schematics, indicating they are really just DC biased capacitors.

I've read various threads like... http://www.audioheritage.org/vbulletin/showthread.php?t=43

Frankly, with my somewhat naive understanding of analog electronics and filter networks, I don't get it, and would love to see some discussion of the theory behind why it works. URLs to reading material are fine, too.

True it is simply dc biasing but many would not have thought of that approach to topology.

As to the rest of the discussion its perhap useful to understand exactly how a polarised and non polarised capacitors works and then related that to the application of a Dc potential. Black Gate used to have a very useful explanation of the facts . Try Google. http://en.wikipedia.org/wiki/Capacitor
http://www.elna-america.com/tech_al_principles.php
http://www.technologystudent.com/elec1/capac1.htmhttp://en.wikipedia.org/wiki/Capacitor_(component (http://www.technologystudent.com/elec1/capac1.htmhttp://en.wikipedia.org/wiki/Capacitor_(component))


A lot of research as been done recently on voltage (AC signal) induced vibrations within capacitors used in audio crossover applications.

According to the reseach its the voltage not the current that induces the vibrations within the structure and this directly effects sound quality.(Vibrations caused by sound within the enclosure have no effect what so ever according to the research)

Depending on the manner of the windings within the structure a DC voltage potential may help control these vibrations

[Russellc]

I agree with charge coupling, but have a implimentation question: Does one need to CC the cap in the woofer zobel? I didnt do that on the one I built and was wondering. CC the zobel cap or not?

russellc

[1audiohack]

I always thought of a capacitor as an electron hotel where the lobby normally would accomodate the same amount of electrons as the beds, the difference being the electrons in the lobby could not get out the back door and the electrons in the beds could not get out the front. If you pushed an extra electron into the lobby the imballance would push one out of bed where it would wait outside the back until it could get back in. The problem is the halls (the dielectric) act like a spring accumulator causing delay and possibly giving that electron somewhere to hide and him not going outside to do the work he was supposed to, or at least making him late.

All right so I rode the short bus to electronics class.

Quoting from the Handbook for Sound Engineers, Fourth Edition 2008 on dielectric absorption or DA

"Dielectric absorption is a reluctance on the part of the dielectric to give up stored electrons when the capacitor is discharged. If a capacitor is discharged through a resistance and the resistance is removed, the electrons that remained in the dielectric will reconvine on the electrode, causing a voltage to appear across the capacitor. This is also called memory.

When an AC signal, such as sound, with its high rate of attack is impressed across the capacitor, time is required for the capacitor to follow the signal because the free electrons in the dielectric move slowly. The result is compressed signal. The procedure for testing DA calls for a 5 minute capacitor charging time, a 5 second discharge, then a 1 minute open circuit, after which the recovery voltage is read. The percentage of DA is defined as the ratio of recovery to charging voltage time 100" Close quote.

So to mitigate this issue you take a battery and two like capacitors and cram the lobby AND the halls full so when one electron gets pushed in the front door, another gets kicked right out the back and into your tweeter to do the work you intended him to do and back again. Making a more ideal device. CC is a very simple and elegent soulution.

The funny thing (to me) is that even seems to bridge the gulf between "pierce and Hoerninger", on the subject of can it theoretically ever be fully charged and if so how long. 5 time constants is considered 100%, your ears may tell you different.

OK I am off to find a flack jacket, and a bottle of tequila and a straight razor.

[jblsound]

I agree with charge coupling, but have a implimentation question: Does one need to CC the cap in the woofer zobel? I didnt do that on the one I built and was wondering. CC the zobel cap or not?

russellc
I C-C all the caps. If a current is flowing through a cap w/o C-C then it is producing a phase shift.

[Hoerninger]

... the gulf between "pierce and Hoerninger"
Thank you for your contribution.

I have never seen a "gulf" between Pierce and me. We only had to clarify our viewpoints, it is from the scientific side for both of us.

First of all I go with Earl K and Allan. They pointed out to try it first and seek an explanation afterwards.

For those who are interested in investigations about distortion in capacitors there are some compiled links :
http://audioheritage.org/vbulletin/showthread.php?p=252124#post252124
____________
Peter

[pierce]

For those who are interested in investigations about distortion in capacitors there are some compiled links :
http://audioheritage.org/vbulletin/showthread.php?p=252124#post252124


ah, yes, Bob Pease. that man is a genius. I miss reading his regular columns in Electronics Design.

[Allanvh5150]

I C-C all the caps. If a current is flowing through a cap w/o C-C then it is producing a phase shift.

A current flowing through a capacitor or inductor will always produce phase shift. Charge coupling will actually double the phase shift up to somewhere near 180 degrees.

Allan.

[jblsound]

A current flowing through a capacitor or inductor will always produce phase shift. Charge coupling will actually double the phase shift up to somewhere near 180 degrees.

Allan.

Wrong, I suggest you go back and do some searches for what Greg Timbers has to say on the matter, and look at the pics of an audio sinewave in a cap w/wo C-C.

When the sinewave crosses the zero dielectric point in a non C-C cap the wave shifts. When C-C the sinewave is above the zero dielectric and continues w/o the shift.

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

[4313B]

I agree with charge coupling, but have a implimentation question: Does one need to CC the cap in the woofer zobel? I didnt do that on the one I built and was wondering. CC the zobel cap or not?

russellcI just bias the whole mess as a matter of course. I hardly ever use N.P.E.'s in conjugates either.

[Russellc]

I just bias the whole mess as a matter of course. I hardly ever use N.P.E.'s in conjugates either.
Agreed. Thanks for the response.

Russellc

[Earl K]

A current flowing through a capacitor or inductor will always produce phase shift. Charge coupling will actually double the phase shift up to somewhere near 180 degrees.

- I really don't know what you're thinking here, but because of that second sentence, your understanding of basic electrical theory appears to be in doubt .

- 2 caps in series ( with that high an "R" value given to the feed resistor ), will ultimately produce essentially the same amount of phase shift ( as a single cap of equivalent value ) / down into the "stop-band" .

- ie; 90 degrees .

- Above F3, phase shift diminishes to near zero ( about a couple of octaves up ) .

- F3 is indentified as a 45 degree phase shift ( for a single passive component ) .

>< cheers :)

[pierce]

the phase shift as measured in degrees depends on the frequency, capacitance, and impedance of the circuit. a circuit that causes a 90 degree phase shift in a 8khz signal will only shift a 60 hz signal a degree or less. two 20uF caps in series is electrically equivalent as 1 10uF cap (neglecting the details of dielectric absorption and other such minutia).

[Robh3606]

Here's what Leap says about a pair of series caps and a single both the equivalent value. No difference the curves overlap

Rob:)

[pierce]

of course, LEAP probably models a theoretical perfect capacitor and doesn't cope with things like dielectric absorption, internal resistive leakage, parasitic inductance, or any of the rest of the little nasties of the real world, eh?

[note I said probably, I've never used LEAP, I'm just guessing]

[Hoerninger]

... or any of the rest of the little nasties of the real world ...

I was curious, I did not expect this

■ Capacitor
This component also includes parasitic models for series resistance and inductance. It also includes a exponential frequency dependent model to simulate various types of dielectric behavior.

which would be an overshot for calculating series capacitors.

Download EnclosureShop Brochure (http://www.linearx.com/cgi-bin/filebot.pl?/pdf/EncShop_Brochure.pdf)
____________
Peter

[Steve Schell]

Rob, I am fascinated by your argumentative stance in this thread. I would really like to understand your thoughts on the subject. Do you think that biasing capacitors is without merit? If so are your objections based on theoretical grounds, or have you experimented and failed to hear an improvement?

There is of course plenty of silliness in audio, but IMO based on direct experience, charge coupling does work and produce positive sonic results. In two elaborate, all horn systems with which I am very familiar, the sound improved rather dramatically within a few seconds of connecting the 9V batteries. Part of what motivated me to build up some CC crossovers was Greg Timbers' enthusiasm for the technique, and I know him to be about as far from a wild eyed tweaky audionut as one can get, though blessed with razor sharp perception. I'm just curious why you are "fighting the good fight" on this one.

Aren't your eyelids becoming heavy? All we want you to do is to go to sleep... go to sleep and become one of us...

[pierce]

I was curious, I had not expected this
...
which would be an overshot for calculating series capacitors.

indeed. and the other link...

http://www.linearx.com/cgi-bin/filebot.pl?/pdf/XvrShop_Brochure.pdf
[the crossover shop brochure...]


I'm impressed. I'd seen various LEAP graphs here, but not being an audio designer, hadn't looked deeper, and was (stupidly) assuming it had Spice style simplified models of the components. Sadly, at $1500, its not likely I'll be playing with it any time soon...

[1audiohack]

LinearX is currently having a 20% off sale, usually a Febuary thing. Their support as well as their product is outstanding, a bargain even at regular prices I think.

Through 06-15-2009.

[midlife]

In simple terms what does charge coupling do? And it effects the listening experience how? Are there draw backs? :bouncy:

[4313B]

In simple terms what does charge coupling do? And it effects the listening experience how? Are there draw backs? :bouncy:It's exactly like adding kiwi to a strawberry drink. The drawback is, if you don't like kiwi or strawberry you're screwed.

[midlife]

It's exactly like adding kiwi to a strawberry drink. The drawback is, if you don't like kiwi or strawberry you're screwed. Perfect, thats all I'll ever need to know about CC, very insightful.

[4313B]

Perfect, thats all I'll ever need to know about CC, very insightful.;)

http://audioheritage.org/vbulletin/showpost.php?p=251055&postcount=2

[pierce]

In simple terms what does charge coupling do? And it effects the listening experience how? Are there draw backs? :bouncy:

while I think 4313b may have nailed the ghist of it, I'll try to answer literally in the most non technical terms I can come up....


Capacitors are a key component used in crossover networks to filter out the frequencies so the highs go to the tweeters and the lows go to the woofers (and the mids to the mid).
Audio signals are 'AC' in that they alternate between + and - voltages many times a second.
When a capacitor changes from + to - or back, there's a little tiny glitch in the signal, on the order of a few millivolts, caused by an effect known as "Dielectric Hysteresis".
"Biasing" is a techie term for adding a DC voltage to a AC signal.
By biasing these capacitors, the capacitor stays entirely on the "+" side, avoiding this glitch entirely.

the net effect is, the speakers 'sound smoother'.

To do this, you replace a single capacitor of X microfarads (uF) with two capacitors in series of 2*X uF (which has the same net capacitance), then connecting the midpoint of these two caps with a several megohm resistor to the + side of a 9V transistor radio battery.. The battery - side goes to ground.

the simplest crossover is a cap in series with a tweeter, this blocks low frequencies that would damage the tweeter. here's two versions of this, the lower one modified with this 'charged coupled' design.

http://hogranch.com/files/Bitmaps/biasxover.png

(pardon my crappy drawing skills, I don't have any decent schematic editing software and was forced to use "Dia").

[jblsound]

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


Above is the last paragraph from that link. And I've mentioned to others you just don't C-C just any ole speaker. B/C of the cost of the C-C circuit the speaker system in question needs to be of a high quality to start with.

The cost to C-C a pair of L212s was about $300. So one certainly wouldn't spend that kind of money on speakers that only cost $300/pr new.
With the current JBL Consumer series I wouldn't spend the time, much less the money, to C-C any series below the Performance Series.

Now for bypass, its quite different, its a small cost. But, if one is thinking of bypassing an old system of say 30 years, give or take, if the XO in question is original, it would be best to build new XOs and bypass them, imo, instead of just adding bypass to old worn out XOs.

[midlife]

I like it, where do I get mine? :bouncy:

[pierce]

I like it, where do I get mine? :bouncy:

http://www.mouser.com/Images/mouser-reg-logo.gif http://dkc3.digikey.com/hp/en/ts_dklogo.jpg

(eg, you buy the parts, you build your own)

[Robh3606]

Rob, I am fascinated by your argumentative stance in this thread. I would really like to understand your thoughts on the subject. Do you think that biasing capacitors is without merit? If so are your objections based on theoretical grounds, or have you experimented and failed to hear an improvement?


Hello Steve

I don’t have an issue with CC networks. I have been using them exclusively on any projects I have built since 2003. As a matter of fact if you look at the first post in this thread the linked thread is the original CC thread I started back in 2003. I was an early adopter and have never looked back. I agree with you and think they really do make a difference that is worth any added part costs. So I am one of you guys in that respect.;)

The reason we do this is to keep the capacitors dielectric from crossing 0 volts on each cycle of the AC waveform. The DC bias prevents this because the capacitor is charged to 9 volts and will remain that way as long as the battery is in the circuit.

My bone of contention is the perceived loss of bias when the sine wave peaks go beyond -9 volts. If you use superposition you would just add and subtract the voltages. But that doesn’t really work here because it ignores the time domain and does not address what is happening inside the capacitor. If you had a 12V PP signal on the negative side you would have the voltage across the capacitor cross 0 volts at -9V and peak at -3 volts at the -12 peak.

The voltage across the capacitor does indeed cross through zero and go negative. That said the dielectric has not crossed through 0 volts because it has not had enough time to discharge. You would have to discharge the 9 volt charge for the capacitors dielectric to cross back through 0 volts. So even though the capacitors see’s a negative voltage the capacitor remains biased due to the long charged and discharge rates of the capacitor resistor combination and the battery.

The capacitors charge and discharge rate is not linear. It’s an exponential curve so it is a complex function of time, not just a static DC voltage. If you compare the discharge curve to the duration of any program material peaks, in the time domain, it is clear that the short peaks beyond 0 volts have little effect on the biasing considering the battery is there to maintain the +9v charge on the capacitor.

So there you have it. I just don’t see where the capacitors dielectric ever crosses back over the 0 volt reference after the DC bias is applied.

My eyelids are getting heavy now

Rob:)

[spkrman57]

It's exactly like adding kiwi to a strawberry drink. The drawback is, if you don't like kiwi or strawberry you're screwed.


I've been adding kiwi to my grapefruit juice!!!:blink:

Ron......

[pierce]

I'm inclined to suspect that another reason why 9V is enough regardless of the signal negative peak voltage is that a few millivolts of glitch on a large signal is insignificant, while that same few millivolts of glitch on a small signal would be obvious.

[Robh3606]

while that same few millivolts of glitch on a small signal would be obvious.

Good point and that's where you hear it as well on the low level stuff. Same with the aguaplas, helps resolve the lower level details. They make a great combination.

Rob:)

[Hoerninger]

If you use superposition you would just add and subtract the voltages. But that doesn’t really work here because it ignores the time domain and does not address what is happening inside the capacitor. ...


As I had thrown in the idea of Superposition I want to reply to your post.

Superposition is an often used approach in physics when a problem can't be solved directly. In electrical engineering it is inevitable for calculating electrical networks. This holds not only for DC but for sinusoidal AC and all other forms of current changes.
When using for example Leap or Spice (LTSpice) (http://audioheritage.org/vbulletin/showthread.php?p=190342&highlight=spice#post190342) for calculating this is done automatically for the user.

When the capacitors are charged by the 9 V batterie it can be switched off afterwards as long as there is no loss due to leakage. When an AC current is applied the initial charging changes. The time constant for the former charging is irrelevant as the AC current flows through the capacitors and the speakers with about 8 Ohm and not through the ca 2.2MOhm resistor and the 9 V batterie (now regarded as short circuit). The speaker has a much lower resistant and thus the time constant is much lower. Now the frequency dependend effect comes into account.

But just like you I discuss here only the variations in time and not the frequency dependency. So I am with you in the time domain but not in the frequency domain. (The corresponding mathematical descriptions can be transformed vice versa.)

It can be said that the dielectric has a voltage. But in the context here it is not helpful as the difference in charge concentration between the two sides of the electric (equivalent to voltage) is caused by the charges on the plates. The dielectric will be polarized. When there is no charge on the plates of the capacitor than there is no polarisation in the dielectric.
(Just ignoring capacitor soakage (http://audioheritage.org/vbulletin/showpost.php?p=252113&postcount=13) for the moment.)
____________
Peter

[Ian Mackenzie]

Hello Steve

I don’t have an issue with CC networks. I have been using them exclusively on any projects I have built since 2003. As a matter of fact if you look at the first post in this thread the linked thread is the original CC thread I started back in 2003. I was an early adopter and have never looked back. I agree with you and think they really do make a difference that is worth any added part costs. So I am one of you guys in that respect.;)

The reason we do this is to keep the capacitors dielectric from crossing 0 volts on each cycle of the AC waveform. The DC bias prevents this because the capacitor is charged to 9 volts and will remain that way as long as the battery is in the circuit.

My bone of contention is the perceived loss of bias when the sine wave peaks go beyond -9 volts. If you use superposition you would just add and subtract the voltages. But that doesn’t really work here because it ignores the time domain and does not address what is happening inside the capacitor. If you had a 12V PP signal on the negative side you would have the voltage across the capacitor cross 0 volts at -9V and peak at -3 volts at the -12 peak.

The voltage across the capacitor does indeed cross through zero and go negative. That said the dielectric has not crossed through 0 volts because it has not had enough time to discharge. You would have to discharge the 9 volt charge for the capacitors dielectric to cross back through 0 volts. So even though the capacitors see’s a negative voltage the capacitor remains biased due to the long charged and discharge rates of the capacitor resistor combination and the battery.

A zero or 0 Dc voltage condition on a capacitor can only exist where there is a zero Dc potential present as in a normal series "single" capacitor without a Dc voltage present.

The capacitors charge and discharge rate is not linear. It’s an exponential curve so it is a complex function of time, not just a static DC voltage. If you compare the discharge curve to the duration of any program material peaks, in the time domain, it is clear that the short peaks beyond 0 volts have little effect on the biasing considering the battery is there to maintain the +9v charge on the capacitor.

So there you have it. I just don’t see where the capacitors dielectric ever crosses back over the 0 volt reference after the DC bias is applied.

My eyelids are getting heavy now

Rob:)

Great post Rob.

My take is that 9 volts is arbritrary and can be any voltage up to the DC working voltage of the capacitor

As such the ac voltage has a linear +-4.5 peak-peak swing within the 0-9 volt dc range of the battery. This is much like the Dc bias current of a SE class A amp but we are talking voltage here not Dc current.

The point being to move or shift the anode/cathode 0 volt crossing up so the ions are no longer moving in the "zero volt" non linear region. ie a non polarised "ideal" capacitor..

This principle is used in so called biasing of many device such as BJT, mosfets and Jfets where we pick a region on the voltage/current gain curve where the device is most linear and apply a constant current (bias) or constant voltage (cascoding) to ensure the non linear distortion caused by the device is minimised.

In practise for domestic applications 0-9 volts allows a large majority of hi resolution (low voltage) audio signal to pass within the linear range and beyond that at the larger superimposed a/c voltage swings that are typcially bass transients are less critical.

A/c voltages across the whole audio band sums for a total peak A/C voltage. In this respect the peak voltages sum additively to increase the magnitude of the peak to peak A/C voltage. They are only superimposed on any the DC voltage present that actually shadows the peak ac voltage but this is not additive. ie ac and dc voltage to not sum together as such nor do ac and dc currents.

In respect to the maximum linear voltage swing as Greg comments elsewhere they tried +18 volts and some Japanese audiophiles feedback was its was an improvement but this is perhaps difficult to validate.

As to what happens and why plug your crossover capacitors into an A/c circuiy and apply a sine wave and listen to the capacitors sing. The singing is the plates rattling around. Because the plate move this causes non linear behaviour and is often termed noise and is in essence distortion. The charge-coupling helps control this issue but it does not control inductance due to the wrapping of the plates much like a coil of foil.

[jblsound]

As to what happens and why plug your crossover capacitors into an A/c circuiy and apply a sine wave and listen to the capacitors sing. The singing is the plates rattling around. Because the plate move this causes non linear behaviour and is often termed noise and is in essence distortion. The charge-coupling helps control this issue but it does not control inductance due to the wrapping of the plates much like a coil of foil.

I seem to remember reading, from a few years back, and I think it was Greg's statement, that applying DC v to a cap tightens up loose caps, which you're referring to. As such removes that distortion.

[Ian Mackenzie]

See this interview:

http://www.sonicstate.com/news/2008/05/27/aes08-if-the-cap-fits-it-must-sound-better/

This take nothing away from Charge-Coupling and I see it as an alternative means of improving capacitor performance.

Of course direct amplifier connection (pure-active) eliminates all these issues.

But Charge-Coupling cleans up a lot of the most audible undesirable characteristics without resorting to over the top (expensive) capacitors.

Ian

link fixed

[Hoerninger]

ICW (UK) have proved conclusively that internal vibrations inside a capacitor effect sound quality.
Concerning this point there is a AES-paper by
Menno van der Veen & Hans van Maanen
"Non-linear distortions in capacitors"
http://www.mennovanderveen.nl/nl/publicaties.html

But from the logical point of view it is not finally determined whether vibrations are fixed by CC.

But then it might be possible that CC affects Dielectric Absorbtion (DA)
http://audioheritage.org/vbulletin/showthread.php?p=252336#post252336

Just some food for thought.
____________
Peter

[Zilch]

The source impedance for the DC bias voltage is very high, several megohms.

That of the signal is orders of magnitude lower. Since the R in R*C is low, the time constant of the AC signal operating on the capacitor is correspondingly shorter; it swamps the DC.

Put a low-impedance, higher voltage source on the capacitor, and it will charge and discharge lickety-split to and from that higher peak relative to the time constant of the battery and multi-megohm resistor. Put a 1-Ohm resistor in, instead 2.2 Meg, and the signal's going nowhere; it's the impedance differential that allows the circuit to pass the AC.

[Ducatista47]

Smart fellow. From Download 4 of Peter's link:


7. CONCLUSIONS
We have discovered a major cause of nonlinear
distortion in capacitors and we have shown
that this effect can create significant –and thus
audible- distortions in sound signals. Especially in
passive cross-over filters this effect can introduce
high levels of intermodulation distortion and thus
lead to audible differences between capacitors.
The underlying causes of these distortions
have now been unveiled and therefore, an
objective way to determine the audible quality of
capacitors can be developed. This is a major step
forward, compared to endless listening tests with
often incomprehensible results. Also, the design of
electronics can be optimised with the non-linear
properties of capacitors as parameter.
We can only hope.

I have not been a fan of meter men because they ignore hearing when ears present evidence to the contrary. If they develop their art sufficiently to measure what we actually hear, I might change my opinion. (And if they also come on board with things like phase shift affecting the listening experience to the degree that it actually does.)

Clark

[Ducatista47]

See this interview:

http://www.sonicstate.com/news/2008/05/27/aes08-if-the-cap-fits-it-must-sound-better/

This take nothing away from Charge-Coupling and I see it as an alternative means of improving capacitor performance.

Of course direct amplifier connection (pure-active) eliminates all these issues.

But Charge-Coupling cleans up a lot of the most audible undesirable characteristics without resorting to over the top (expensive) capacitors.

Ian

link fixed

I find it especially interesting that it seems to be JBL he is talking about working with on this issue. (All the hints point to them. A very large speaker manufacturer using very efficient drivers and having been trying to deal with this problem for some time.) Doesn't JBL use Solens in its high end crossovers? Perhaps they will switch to Clarity Caps now. Caps designed and manufactured to eliminate this at the source are preferable to any conceivable remedial fix.

Talk about proof of the phenomenon being audible. They designed a test capacitor so afflicted in this respect that it turned into a speaker. I agree that charge coupling, designed to address an entirely different problem, would do nothing for this one. One thing that will never change, the best crossover would be no crossover. Things in the signal path can degrade the signal, but never improve it. Neutral would be great, but rarely even possible.

It is amazing that some of these fundamental suppositions had never been investigated for veracity. The example cited of it being voltage rather than current that causes the problem by a ratio of one million to one is downright embarrassing. And the sound/vibration induced distortion in caps proved to be a groundless myth. In other words, it ends up that caps are not microphonic in the least. Since it is the Clarity Cap people that did the research I am glad I am already on their bandwagon, as I would surely have switched after seeing this.

My thanks to Ian for telling me about Clarity Caps in the first place and for pointing to this interview.

Clark

[Hoerninger]

Using LTSpice (http://www.linear.com/designtools/software/#Spice) I took a look at the voltages at a 6 dB/oct highpass with 2 kHz.
I would have liked to add the circuit file. But I am not used to Windows and I do not know how to perform. With Ltspice Getting Started Guide (http://www.linear.com/designtools/software/#Spice) you have a very good entry into LTSpice.

Charging up:
The first plot shows how the voltage across one of the paired capacitors is rising.
After about 300 seconds both have a voltage of 9 volts.
After these 300 seconds there is a burst of 10 cycles at 3.5 kHz. The source has a voltage swing of +/- 18 Vpeak (12,7 Veff; 36 Vpp).

Going on I want to give you two automatically calculated examples which show other results some have discussed before.
____________
Peter

[Hoerninger]

Cut off region (lower than 2 kHz):

A sine wave of 20 Hz and an amplitude of 17.5 Vpeak is applied (12.4 Veff, 35 Vpp).

20 Hz is far within the cut off region and so there is only a small voltage at the speaker, smaller than 0.2 Vpeak - red V(n003).
At C1 rsp. C2 the voltage swings are between 0 V and ca. 17.5 V – green V(N002,N001) and blue V(N002, N003). Each capacitor has about half of the generators voltage swing.

For a further increase of voltage reversed polarity will occure. This may cause distortion. But normaly the overtones will be attenuated with increased frequency and only a minority part will belong to the pass band.
____________
Peter

[Hoerninger]

Pass Band (above 2 kHz):

A sine wave of 20 kHz and an amplitude of 17.5 Vpeak is applied (12.4 Veff, 35 Vpp).

The voltages at each capacitor are relativly small between 7.9 V to 9.5 V – green and blue. Most of the voltage is at the speaker – red V(n003), nearly the whole swing of the generator.

In the pass band the input voltage can be greatly increased until the capacitors are driven into reversed polarity. The source voltage can rise up to 38.8 Vpeak (26.9 Veff; 77.6 Vpp).

There is only little chance that in the pass band distortion by polarity change will occure. It would be fairly loud, but then the ears produce distortion as well.
____________
Peter

[Allanvh5150]

The charge-coupling helps control this issue but it does not control inductance due to the wrapping of the plates much like a coil of foil.

Good point Ian. I am not sure of the inductance of say, a 5uf cap is, but if we put 2 10 uf caps in series for our CC network we will get a larger series inductance. However, if we use 4 x 5uf caps in series/parallel, the series inductance will be the same as the single 5uf cap. Correct me if I am wrong. I am not sure what the reasoning is for not using bypass caps in a CC network but generaly, a small bypass cap of 0.01uf will make the series inductance very small. One could just use 1 bypass cap for the whole group. I don't think the term "bypass" should be used because the cap doesn't bypass anything at all. The impededence of the 0.01uf cap is hugely higher than the actual network capacitor. It does not bypass the high frquencies around the main capacitor as some would suggest. It mearly "cancels" the series inductance. You would use a "bypass" capacitor to route the high frquencies around the L-Pad in a crossover; aka compensation network.

p.s. So why don't we use a 0.01mH in series with the inductors to cancel their parallel capacitance?

Allan.

[Robh3606]

Hello Peter

Thanks for running and posting the Spice plots.

Rob:)

[Hoerninger]

Rob, thank you. :)

In post #108 there is a small mistake when I started calculating by hand.
The source voltage may rise to more than 38.8 Vpeak as the new spice plot demonstrates.
The consequence is the same, the ears produce distortion as well - as long as the driver will not get burned. :hmm:
____________
Peter

[4313B]

Here's an example for the 4355. I can't take credit for the nice box which another forum member built. He wanted his networks external. I think the connectors alone were something like $400 for all sixteen ...

[4313B]

of course, LEAP probably models a theoretical perfect capacitor and doesn't cope with things like dielectric absorption, internal resistive leakage, parasitic inductance, or any of the rest of the little nasties of the real world, eh?

[note I said probably, I've never used LEAP, I'm just guessing]Here's the editor box in LEAP.

This particular file is Greg's LEAP file from the Everest II which I am modifying for use with older dual 1500AL's instead of the Everest II's newer 1501AL's.

[pierce]

Here's the editor box in LEAP.

This particular file is Greg's LEAP file from the Everest II which I am modifying for use with older dual 1500AL's instead of the Everest II's newer 1501AL's.

I'll bet its fun figuring out those values for real world caps from the typical vendor spec sheet... or do you measure them on the bench?

[midlife]

Here's an example for the 4355. I can't take credit for the nice box which another forum member built. He wanted his networks external. I think the connectors alone were something like $400 for all sixteen ...
That is an immaculate piece of work, thumbs up.

[jblsound]

Here's an example for the 4355. I can't take credit for the nice box which another forum member built. He wanted his networks external. I think the connectors alone were something like $400 for all sixteen ...

Just beautiful
That pic makes me want to re-do the C-C XOs for my L212s, as I've got the three cards stuffed into the tops of the boxes. One mounted to the right, one left and the third across (vertically) the back's opening.

[jblsound]

Having used a C-C pair of L212s for 5 years, in many different ways. freestanding stereo setup, mains in a MC HT, and wall mounted nearfield, the nearfield might be the best at showing off what C-C can do.

As I type this, I'm listening to those speakers all of 4 ft, or less, from my chair. As all the other walls in this room are more than 4 ft from the speakers the nearfield setup eliminates the room from the sound. I hear details of songs I know very well, that I never noticed before, and so smooth, and clean, clear.

And as they are wall mounted and gain 3db @ the low end, no need to use the sub.

[Hoerninger]

I'll bet its fun figuring out those values for real world caps from the typical vendor spec sheet...
Looking at LTSpice there is a window for capacitor properties as well. The database of stock capacitors does unfortunately not show those capacitors we would like to have in a deviding network.

There seems to be no known possibility to simulate dielectric absorbtion (capacitor soakage) which is CC applied for. The usual modelling with paralleled R-C combinations will only show linear dissortion but not the nonlinear ones.
[The effect of CC could be shown in a simulation :thmbsup: ]
Any idea?
____________
Peter

[John]

Here's an example for the 4355. I can't take credit for the nice box which another forum member built. He wanted his networks external. I think the connectors alone were something like $400 for all sixteen ...

That picture is a repeat. :(

Would be nice to see some other angles of that nifty cabinet. ;)

[midlife]

Does charge coupling yield as good of results on horn loaded systems as it has on more conventional speakers?

[jblsound]

Does charge coupling yield as good of results on horn loaded systems as it has on more conventional speakers?

JBL uses Charged-Coupled crossovers on the K2s and Everest II, need I say more?