I cannot give it a go in my setup as it's ESL, no caps in the signal path.Robh3606 said:[B
Give it a try. Then say it doesn't work. It worked for me. Like Earl says it is subtle but it's there and it's an improvement for sure.
Rob[/B]
Could that be because the signal is now running through two x caps, instead of one x maybe even different maker of cap? Caps do sound different "subtle"
Or as john k had measured an increase of thd when the bias is applied, up to .25%- 3% 2HD, when increased to over AC signal level.
Cheers George
I tried it about a year or two ago with my ESLs (using an Acoustat MK121 interface box and a few other mods). No difference that I could hear, but I wasn't using lousy handmade caps, either. For me, there was a MUCH bigger improvement when I got rid of the MK121 entirely, biamped, and went to a single step-up transformer with no capacitors.
Menno's paper is pretty speculative- it would have been nice to see some real data showing that the effect actually exists with reasonably well-made components. He assumes for his model a 3.5% variation in the plate to plate distance- what's the experimental justification for this number? It seems... awfully large. I don't doubt that there's some loosely wound "audiophile grade" hyper-expensive cap out there somewhere that might be this bad, and perhaps the resulting distortion is what makes that cap worth its price (cynical me!), but for any reasonably well-made industrial capacitor, that variation seems a few orders of magnitude too high.
The claim to have found a "major effect" doesn't seem supported by the paper's contents and at this point seems to have no experimental backup.
Menno's paper is pretty speculative- it would have been nice to see some real data showing that the effect actually exists with reasonably well-made components. He assumes for his model a 3.5% variation in the plate to plate distance- what's the experimental justification for this number? It seems... awfully large. I don't doubt that there's some loosely wound "audiophile grade" hyper-expensive cap out there somewhere that might be this bad, and perhaps the resulting distortion is what makes that cap worth its price (cynical me!), but for any reasonably well-made industrial capacitor, that variation seems a few orders of magnitude too high.
The claim to have found a "major effect" doesn't seem supported by the paper's contents and at this point seems to have no experimental backup.
I don't have data on dielectric modulus but 3.5% compression seems very unlikely.
If the majority of the plate distance change is due to compression of any remaining air in the plate to plate interface then 3.5% is very possible.
Compression will depend totally on whether the manufacturing process removes that air.
If the majority of the plate distance change is due to compression of any remaining air in the plate to plate interface then 3.5% is very possible.
Compression will depend totally on whether the manufacturing process removes that air.
Please re-read my post. I did not measure anything. I ran simulations using Menno's model to see if the simulated results agreed with my intuitive comments back in Post 19 , and they do. And I was mostly interested to see what the result would be for a DC bias in terms of the odd order distortion being reduced.
It shouldn't be surprising. Take a linear device; add a nonlinear effect; is there any surprise the results show nonlinear distortion? In fact, this argument is identical to the discussion of VC heating effect of Re and the resulting distortion.
The problem with Menno's paper is that he makes an assumption which leads to the result without providing any data that the assumption is valid. It's nothing more than an If; Then argument.
For example, if there is a limit to the compression, i.e. bump stops, then an applied DC bias of sufficient magnitude can eliminate all distortion associated with dielectric compression. But this bias will need to be such that the minimum of DC + input signal is sufficiently positive (for a positive DC bias) that the dielectric is always compressed against the bump stop.
Also consider AndrewT's comment. For metalized PP film cap most of the cap is PP dielectric. It would seem that if the force on the dielectric is sufficient to compress it by even 1% we should be able to measure the change in diameter of the cap (say a 50 mf PP film cap that is about 2" in diameter) when a sufficient DC voltage is applied. Get out you micrometers boys!
It shouldn't be surprising. Take a linear device; add a nonlinear effect; is there any surprise the results show nonlinear distortion? In fact, this argument is identical to the discussion of VC heating effect of Re and the resulting distortion.
The problem with Menno's paper is that he makes an assumption which leads to the result without providing any data that the assumption is valid. It's nothing more than an If; Then argument.
For example, if there is a limit to the compression, i.e. bump stops, then an applied DC bias of sufficient magnitude can eliminate all distortion associated with dielectric compression. But this bias will need to be such that the minimum of DC + input signal is sufficiently positive (for a positive DC bias) that the dielectric is always compressed against the bump stop.
Also consider AndrewT's comment. For metalized PP film cap most of the cap is PP dielectric. It would seem that if the force on the dielectric is sufficient to compress it by even 1% we should be able to measure the change in diameter of the cap (say a 50 mf PP film cap that is about 2" in diameter) when a sufficient DC voltage is applied. Get out you micrometers boys!
Andrew, let the red plates in this diagram be charged positive, the black ones negative. Except for the outermost plates, the forces on each one cancel, right? The attraction to the opposite plate is exactly counterbalanced by symmetry.
And yes, the forces are small. As I mentioned several posts back, 3.5% is a very questionable assumption. Dielectrics are not soft rubber!
And yes, the forces are small. As I mentioned several posts back, 3.5% is a very questionable assumption. Dielectrics are not soft rubber!
Attachments
I don't think so.
take the middle pair of plates. They attract each other. They try to compress what separates them.
The next two plates are attracted to the middle pair and so on to the edge. The whole sandwich compresses due to the attractive forces.
I may be wrong. Correct me if necessary.
I wonder if you are confused with the net force on any individual plate. If a net force were left then the plate would fly off into space. We know it doesn't. The net force must be zero. That does not stop an attractive force across the separator.
I agree, that 3.5% assumption cannot be compression of a solid dielectric. If there is air between the plates them maybe 3.5% is reasonable, bringing us back to "how are capacitors made"?
take the middle pair of plates. They attract each other. They try to compress what separates them.
The next two plates are attracted to the middle pair and so on to the edge. The whole sandwich compresses due to the attractive forces.
I may be wrong. Correct me if necessary.
I wonder if you are confused with the net force on any individual plate. If a net force were left then the plate would fly off into space. We know it doesn't. The net force must be zero. That does not stop an attractive force across the separator.
I agree, that 3.5% assumption cannot be compression of a solid dielectric. If there is air between the plates them maybe 3.5% is reasonable, bringing us back to "how are capacitors made"?
I disagree. The net attractive force on any plate comprises the vector sum of the attractive forces between it and its neighbors. If you take the middle plate, the force between it and the neighboring plate on the left is exactly counterbalanced by the force between it and the neighbor on the right. So the net compressive force is zero. You can make the same argument on each of the neighboring plates until you get to the outside of the stack.
The thing that gets me here is we aren't talking about some esoteric audiphile **** and bull story. The claim are straight forward: DC bias increase even order HD and reduced odd order. So how hard is that to measure? Build a crossover, measure the distortion at the input to the driver. Modify the crossover to include the DC bias and remeasure. The result would be definitive for that application. No ifs, ands or buts about it.
john k... said:
Hi John
I quite agree, and thanks for your comments and observations re Menno's paper. Even any dialectric deformation well below 1% should be easily measurable, this is the crux of the matter on one side. The other is the question: How audible is this? Shouldn't be hard to find out.
I have withheld my own listening observations. I was hoping that we would have a bunch of people just trying this and post their impressions.
With 1st and 2nd order there is only a single cap and provided enough capacitance is at hand not-with-standing choices of favourite brands. And to A-B just unclip one of the series batteries. What consensus might emerge? That is what interests me mostly.
What I heard was not just subtle in the context of the experience, but it may be subtle to others in their context. Mind you, the system it was heard in also bears context and it was an exceptional one. Simply the best digital front-end I know [Terra Firma http://www.audioasylum.com/forums/hirez/messages/25/252126.html] and amplification of matching and sterling quality. Whether subtle or not it was highly significant.
Now I like when there is a correlation between the observable (don't like using subjective much as that word now just means opinion) and the objective measurements. I consider that progress and support anybody attempting that, even if one has to pre-define parameters (and vary them - the fudge factor), but like you indicated this can reveal trends. But one has to start with some assumptions.
So when "I HEAR" it means "I OBSERVE" and then how do I define/quantify and explain in a physical world what I hear, what's the cause, because if you figure the cause you now have the potential to make it better.
So on this thread, I'd just like at least some of the people to first DO and then theorise the cause and measurability.
This is at the point where I am now (wish others were too) and for some that would be enough. Fine, enjoy. But I have a bunch of questions as it just interests me. That is the OTHER side of the issue.
Some must have heard the effect of DC bias. Did JBL or just a sales gimmick? I can see two persons who have posted positive observations, good. We are getting somewhere.
Now for my questions: It seems this must be measurable as you [John] said. If so, then distortion should vary with both bias level and signal amplitude as both affect the minimum DC seen by the cap. If so, then consider that a confirmation. But it must be a physical measurement, as you indicated.
The other question I have and one that Salas put out, the mechanical properties of the cap. What triggers it? Does DC bias prevent triggering resonances of vary Q and freq? Or is that not related? If a truly bad cap can "sing" then a good cap is simply attenuating the effect by -x dB and some freq shifting? An uncontrolled resonances versus a controlled one, what are the non-linear consequences?
Has anybody ever connected an accelerometer to a cap. Want to add a smilie here, but maybe not?
Joe R.
Just because you use your senses doesn't mean it isn't objective. For instance an ABX test is objective and yet involves only human senses. On the other hand, if it's not objective then it is by definition subjective. That only means that it is not free of opinion, not that it is solely opinion.Joe Rasmussen said:
What is an example of a handmade cap? To the best of my knowledge, most of the super-expensive, exotic audiophile caps are made by RTI Electronics.
http://www.rtie.com/copro.htm
John
Mr Evil said:
Let's not go there. But quickly, I agree that when I hear three acoustic guitars (tripple-tracked) when I thought there was two (double-tracked) after a genuine hardware advancement (as opposed to the usual swap-cooking-with-components tweaking), that's an objective moment.
Joe R.
SY said:
Clearly, if there are physical resonances within the construction of the cap, then dialectric compression is not the only thing happening. But let's say before we get close to 10-12,000 psi these things will show up? I think so. But if we hear something there has to be an explanation or two.
Joe R.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.