Wax Paper Capacitor Test

[salvarsan]

In the absence of any other info, I'll continue to think it's a volumetric combination,

What is meant by a volumetric combination?

possibly enhanced or detracted by the psychosometric icosahedrality of the monopenes in the matrix, but only on alternate hydrogen bonds.

Why do you even bother writing that sh^H^H stuff? Rhetorical question, thank you.

Yeah. I'd be a lot happier about thinking about paper dielectrics, impregnants and voids if there was any evidence that there was any consistency to be had when it is made. A lot of my samples of the literature is focused on it varying all over the map. I'm guessing that subtleties of hydroxyl groups making it favorable for polymer strands to stack from hydrogen bonding may be washed out entirely by the equivalent of the old story about the night watchman spitting in the vat.

Heh. Fraid not. Unless you hydrolyze cellulose down to glucose units, the polysaccharide units stack ... unless they're in gas phase. Spit or not, cellulose is still going to act like a dielectric participant.

Click on the thumbnail. Everywhere is you see a dotted line is a hydrogen bond, and there are a lot of them. The basic cellulose unit is godawful complicated, particularly if you don't like chemical chickenwire drawings.



The paper dielectric articles before the 80's lament the variability, but things got better.

In that 2003 Indian paper, the very existence of units to specify pulp uniformity and liquidity tells us the characterization of paper manufacture improved enough that the void uniformity is not only quantifiable but predictable once you reliably digest out lignin and non-cellulosic material.

Other little bits like treatment with dilute EDTA to remove residual metal salts drastically lower conductive defects in the paper. Pre-WWII patents mention ~10 conductive defects per square inch or 15500 defects per square meter. That modern German capacitor paper specs at 50 defects per sq.meter worst case, a three orders of magnitude improvement. So, the industrial methods are here now and allow sophisticated predictive modeling, something impossible in the heyday of paper capacitors.

Undoubtedly, the best info is behind the IEEE paywall, but if you have university library access, SciFinder gets you some of it.

"Why do you do all this research?" is a very fair question.
Why does a dog lick itself?
The longer answer is that marketing hype and assertions of ineffable magic inhering in antique manufacturing offend me.
The mythopoetic superiority of the NOS JRC4558 is a case in point. (Sorry, didn't mean to grind my teeth so loudly)

Sounds like a great place for you to go do some novel research!

BTDT, got the T-shirt. Kill me now. Please.

[R.G.]

What is meant by a volumetric combination?

I have here in my hands two imaginary things. One is a capacitor with paper spacing, soaked in some dielectric oil. The other is a box with a three way switch.

In one position, the imaginary box makes the dielectric oil turn to purest vacuum. In the middle position, both oil and paper are still there, in place. In the third position, the paper turns to vacuum, but the oil magically remains in exactly the same positions it was when the paper was there, not collapsing to a fluid layer. So in each non-middle position, the opposing dielectric vanishes, replaced by free space, but otherwise the remaining dielectric is unchanged and remains in the same positions in the gap, not collapsing or shifting.

I also have an (alas...) imaginary meter that can tell me the real, no-fooling, dielectric constant of the cap insulation by unknown processes on the capacitor.

So we set the magic box to the middle position, corresponding to the real world as we know it, intermixed in position. We measure the dielectric constant with the imaginary, but maybe-possible-to-realize meter. We get some number for the impregnated paper and oil.

Now we flip the switch to the paper-only position, and measure. Having recorded that, we flip the switch to the oil-magically-held-in-position position, and measure again. We record this number as well.

Now we have to think, which always hurts.

1. How would one calculate from the DC of the paper itself and the oil itself what the three recorded measurements would be? In one position, they're intimately intermixed, in another they are a more or less diffuse "spray" of material, interspersed with vacuum.

That hurt enough, but we feel compelled to press on.

2. We change a setting on the switch box, and now in the paper-plus-vacuum setting, it rearranges the paper to be pure paper, no voids, on one plate, separated by vacuum from the other plate, both plates remaining at the same separation. The difference is that the paper is now solid, and the vacuum voids are all on one plate, the paper on the other. We measure and record the DC. Is the new DC the same, or different from the paper-matrix test result?
If so (or not!) why?

3. We now do the same for the oil, shifting all the oil to a layer on one plate, vacuum between the oil and the other plate, and the plates at the same separation as in the first tests. Is the newly measured DC the same or different from the oil-matrix test, and why or why not?

4. Finally, we try the middle setting of the switch. This setting puts the paper on one plate, no voids, and the oil on the other, no voids. The plates are at the same separation. Is the oil intermixed with the paper at the interstice? And is the new measured DC the same, different, or what, compared to the other tests?

That's what I was getting at with "volumetric combination". Is the paper independent from the oil, do they interact in the DC on a basis other than the portion of the volume they fill, and do they interact on a basis of other than their DC and the volume they fill?

Why do you even bother writing that sh^H^H stuff? Rhetorical question, thank you. Heh.

As you noted later, it has to do with dogs and their varied interests.

Unless you hydrolyze cellulose down to glucose units, the polysaccharide units stack ... unless they're in gas phase. Spit or not, cellulose is still going to act like a dielectric participant.

You're not familiar with the story. Back when one could make money making steel, one steel maker and the nearest matched competitor were at it hammer and tongs in the market. Their products were remarkably similar, but one of them was consistently just a ...bit... better. The burdened competitor furiously analyzed chemical and alloy compositions over a long period of time, matching compositions, treatments, batch sides, everything they could think of. Nothing reduced the slight lead of the other competitor. The burdened competitor finally resorted to industrial espionage. They placed a spy in the other company, and over a period of time got reports on an incredible number of details about the making of the slightly better steel. However well these were replicated, the difference remained. Finally, the spy was berated that there had to be SOMETHING different, and he retorted that he'd reported this; the only thing different any more was that the night watchman at the better competitor's place chewed tobacco and liked to spit into the vats when he made his rounds...

Click on the thumbnail. Everywhere is you see a dotted line is a hydrogen bond, and there are a lot of them. The basic cellulose unit is godawful complicated, particularly if you don't like chemical chickenwire drawings.

No need. I have a working acquaintance with it, sufficent for my needs.

The longer answer is that marketing hype and assertions of ineffable magic inhering in antique manufacturing offend me.
The mythopoetic superiority of the NOS JRC4558 is a case in point. (Sorry, didn't mean to grind my teeth so loudly)

You'd love this. My company bribed working guitarists (with sodas, popcorn and candy bars!) to come listen to a range of effects. We got them to vote on what they thought sounded better. But they had to do it with ears only. A live guitarist (who also didn't get to see which of what was being auditioned) played repetitive licks and the audience listened, voting for what they liked better. One of those tests was a series of different opamps including the NOS JRC4558. The result was that there was no clear preference above what chance would explain. Not a full-blown, rigorous blind test, but highly indicative. Several of them were astonished that they could not just hear the difference plainly.

[Steve Conner]

The result was that there was no clear preference above what chance would explain. Not a full-blown, rigorous blind test, but highly indicative. Several of them were astonished that they could not just hear the difference plainly.

Well, who would have thunk it. And yet we still have to sell gear to these people without offending them.

[nickb]

Great results, Salvarsan! I always suspected those waxed paper capacitors would be bad, but I never would have expected them to be as bad as that. My theory is that they are full of moisture. Water has a high dielectric constant and a hefty dissipation factor at low frequencies.

I guess they might have been better when new, but anyone using them now will be using old ones, so this is the relevant test, unless some audiophile company started making them again.

The electric music industry itself is just a side effect of the electronics industry, so it's hardly surprising that we take components intended for another job and use them for their musical side effects. I believe this is possible without sacrificing reliability or manufacturability, but some people would differ.

The UT612 looks like a fine geek toy.

I saw this thread took a distinctly materials science turn, but I'd like to throw this in the the mix, if I may, which I think reinforces the moisture absorption hypothesis further. At least one person, documented in this thread Reforming Waxy capacitors - UK Vintage Radio Repair and Restoration Discussion Forum, has used a vacuum to reform wax capacitors. He states later in the thread

" I have just finished a long run to examine what happens during my vacuum process. I seem to have reduced the leakage of a 0.01µF capacitor measured at 375V and 100°C from over 10mA to 10µA.

When cold it's leakage was initially 50µA and is now unmeasurable without resorting to more specialist equipment - which I will do when I get round to it.
It remains to be seen whether the results will last although previous tests suggest that it might.

It was interesting to see that the leakage dropped from huge to under 100µA in a very short time after applying vacuum (minutes). It then took a further 5 hours to reach a point where no further improvement could be seen over a 1 hour period although most of this improvement took place in 2 hours.

It was this complete flattening off of the improvement curve that made me wonder if I had reached the point of it's original performance."

Interesting, I thought.

[R.G.]

Good stuff, Nick!

There's no question that moisture absorption can cause changes in a cap's dielectric, and big ones.

Moisture absorption would account for a lot of the unknown drifts and oddities that people gain/lose by changing out older waxed caps. And the changes are not only measurable, but not on a cork-sniffing scale, either.

[salvarsan]

I saw this thread took a distinctly materials science turn, but I'd like to throw this in the the mix, if I may, which I think reinforces the moisture absorption hypothesis further. At least one person, documented in this thread Reforming Waxy capacitors - UK Vintage Radio Repair and Restoration Discussion Forum, has used a vacuum to reform wax capacitors. He states later in the thread

" I have just finished a long run to examine what happens during my vacuum process. I seem to have reduced the leakage of a 0.01µF capacitor measured at 375V and 100°C from over 10mA to 10µA." ...

Good catch and throw.

NIST has a definitive doc on water dielectric constant variation called
"A Database for the Static Dielectric Constant of Water and Steam".

It can be eye-glazing but less so when you remember that room temp is 298K and 1 atmosphere pressure is ~0.1MPa and ignore all those equations with the ε in 'em.

There's another article somewhere about estimating water contamination using test frequencies < 10Hz but I can't find it anymore.

[Rhodesplyr]

Drying the paper used for making paper capacitors was reportedly one of the challenging aspects of the manufacturing process. One would imagine all this would have to be done in a very clean, controlled environment.

Since paper is a natural product, its pH could vary depending on the source of the fiber, and one source I read mentioned a failure mode where water in the dielectric, in conjunction with an applied DC voltage, could dissolve and move metal ions through the paper, eventually resulting in a conductive path.

Even the machinery used to make the paper could leave behind tiny metal particles that would result in shorts.

[salvarsan]

Drying the paper used for making paper capacitors was reportedly one of the challenging aspects of the manufacturing process. One would imagine all this would have to be done in a very clean, controlled environment.

I wonder if the original capacitor manufacturing processes went on to become clean room processes.

Since paper is a natural product, its pH could vary depending on the source of the fiber

It varied depending on the pulping method, too. The Kraft process leaves the fiber length intact and produces a strong paper.

, and one source I read mentioned a failure mode where water in the dielectric, in conjunction with an applied DC voltage, could dissolve and move metal ions through the paper, eventually resulting in a conductive path.

The metal ions were initially an irreducible problem until someone realized chelation reagents (like EDTA) were a good idea.

A very early patent wound the capacitors using single-side aluminized paper with ~300VAC applied to burn out the conductive bits.

[Rhodesplyr]

The metal ions were initially an irreducible problem until someone realized chelation reagents (like EDTA) were a good idea.

Interesting. When did that practice begin?

[Steve Conner]

The vacuum drying process is interesting, but even 10uA of leakage is pretty useless for an interstage coupling capacitor. You want more like 10nA.

I said above that water had a lousy tan delta at low frequencies. From the information posted above, I see that seems to be because of conductive impurities, and if you could get perfectly pure water, it would be a pretty good capacitor dielectric.

Water is actually used for that purpose in some extreme pulsed power physics experiments. The Z-machine at Sandia Labs uses giant coaxial lines filled with deionised water for pulse compression. The water filling drastically lowers the impedance and increases the stored energy for a really impressive bang.

The lines are charged up just microseconds before firing, so they can get away with lower purity water and higher leakage currents. This famous image http://www.sandia.gov/z-machine/wp-c...and-sparks.jpg shows sparks shooting across the surface of the water as the charge redistributes itself after firing.

[Mick Bailey]

Something to consider is that the cap doesn't stand in isolation - its surface, terminals and mounting method contribute to losses, especially atmospheric moisture, board contamination and adjacent track proximity. Synth S&H circuits can droop quite badly if the atmosphere is moist. Old circuits with paper caps will often leak DC all over the place anyhow, far in excess of any cap leakage. Fabric covered wire, paxolin, fibre, beeswax and paraffin wax can all absorb moisture, so the circuit becomes the sum of its parts. If you replaced everything that was out of spec or defective in some way you'd have nothing left of the original.

[salvarsan]

Interesting. When did that practice begin?

Not sure when it started, but it had less to do with capacitor paper and more to do with not creating superfund cleanup sites around paper mills. I suspect it was in the 80's when the actuaries correlated Parkinson's Disease incidence with paper mills that used chlorine bleaches without waste water remediation, i.e., the bastards dumped their dioxin-laden waste water.

Bleaching paper historically used chlorine gas, then chlorine dioxide before moving on to less pernicious pollutants (and less effective bleaches) such as peracetic acid, peroxide, and pure oxygen+hydrodynamic agitation. The oxo bleaches decompose in the presence of metal and metal ions. Pre-treatment with EDTA removes the metal, allows for more effective bleaching. The remaining problem is EDTA extraction from waste water.

The incidental effect is that removal of metal and metal ions decreases the conductive defects in capacitor paper.

[Rhodesplyr]

Imagine the effect of water absorption on high impedance shielded cables with cloth insulation, which you do find in some older equipment.

[salvarsan]

Imagine the effect of water absorption on high impedance shielded cables with cloth insulation, which you do find in some older equipment.

Serious stuff. Given the obvious problems with paper as a dielectric, you realize how cutting edge the early stuff could be.

[R.G.]

Even more, how sharp the guys who recognized this without wikipedia to refer to!