The bad leakage current and how was meassured

[Tecnófilo]

Hello you eestorians. I´ve been scrolling your chat and searching posts. I am surprised nobody has made comments here (well, I did not see them) about the "testing equipment" specified in the EEStor report of jan 28. And someone is calling for an explanation of how a "leakage current" is meassured.

As I think you are guessing about me. I am a spanish physicist and my speciliaty is in electronics.  And, yes, I have ZEEN shares at 0.87CAD/sh. I shall try to explain you my opinion about the recently published leakage and its measure.

First problem I see: as "Weir's style" implies, we don’t know which "testing equipment" was used; we only know that "testing equipment included"... so maybe more testing equipment were used.

"Testing equipment included: QuadTech 1715 LCR Digibridge LCR, Keyence GT2-212K, Yokogawa WT3000, Stanford Research PS350/5000B-24W"

QuadTech 1715 LCR Digibridge LCR (quadtech - specifications sheet): is useful for testing impedances (resistance, capacity and inductance). So Weir should have an analysis of the impedances at "Test Frequency: 100Hz, 120Hz, 1kHz and 10kHzv (9.6kHz)". But it is not related with current leakage. Also, for energy storage, which matters must to ZENN, it is without interest the response at 1KHz: charge and discharge will be at DC or cuasi-DC, unidirectional current.

Keyence GT2-212K (GT2 - GT2 specifications - GT2 applicationes): GT2 is a series of "contact probes". I think were used to messure the thickness and maybe wardpage (electrodinamic response with applied voltages). It also has nothing to do with the leakage current.

Leakage current is very simple concept: it is the current between plates through the dielectric insulator, even when the capacitor is disconnected. Even disconnected, the capacitor has a voltage between plates (culprits are the charges accumulated in both plates as they attract), so there will be a current that (self)discharge the capacitor: a very bad thing if you pretend to storage charge in it. The energy storaged in a capacitor is only the potential energy due electrostatic attraction between charges storaged in plates; of course you can speak about the changes in dielectric (polarization, field, fux, etc.), but at the end the storaged energy is that potential energy.

So, how is measured the leakage?: it is also very simple. The capacitor will be connected between a constant voltage source which is equal to the voltage it is intended to maximum charge the capacitor; then the capacitor is let charging until it has the intended maximum charge, which occurs when the capacitor has the same voltage than the source but with inverted polarity. 
Then the current is messured, with the same constant voltage source applied! That is so because this current should have the same value that the leakage current in the first instant when capacitor is disconected. If you disconnect the voltage source, at the first moment the leakage current is equal to the current measured as I stated here; after disconnection the leakage current decreases exponentially as it does voltage between plates, and so the storaged charge and energy (you have beautiful graphs posted in this blog).

So the "testing equipment included" that matters for leakage are: Yokogawa WT3000 and Stanford Research PS350/5000B-24W (we don’t know if they used other equipment).

Stanford Research PS350/5000B-24W (PS300Sereis Catalog - PS300Series Manuals): is the voltage source, and it can force up to 5000V (5kV) between plates. It is very sophisticated and can measure the current with "current Resolution 1uA" (micro A), it has display for current and can be controlled with computer. But the problem is obvious: preliminary EESTOR results of leakage currents are 0.0002, 0.0003 and 0.0007 mA: well out of its resolution, 1 orders of magnitude(*).

PS350 cannot meassure neither 0.2uA nor 0.7uA: it should display 1uA or 0uA.

So there is no doubt for me: leakage current was NOT measured with the Stanford Research PS350/5000B-24W.

Yokogawa WT3000 (WT3000): it is a power analyzer. An impressive meter for sure, can be computer controlled (interfaces GP-IB, Ethernet, RS-232 and USB ), frequency up to 1MHz. Of course it can measure the current; I am not familiar with it, but from its specifications I understand the current ranges that can be selected are for the 2A input element (I understand it means up to 2A):
5mA, 10mA, 20mA, 50mA, 100mA, 200mA range.

 

So, the best option I see in specifications is to select 5mA range; again I understand up to 5mA. What is the DC accuracy for that range (5mA):
"DC accuracy: 0.05% of reading+0.05% of range+2uA"

Forget the reading: 0.05% of range+2uA=5*10^-4*5*10^-3 + 2*10^-6=(2.5+2)*10^-6=4.5uA. So, at best case current could be measured with +-4.5uA: too much error margin.

Furthermore, several current sensors are proposed in specifications sheet, with its accuracy:
"751521....DC to 100 kHz (-3 dB). -600 A to 0 A to +600 A (DC) Basic accuracy:(0.05% of rdg* + 40 mA)"
751523... idem
CT1000...DC~300 kHz, ±(0.05% of reading +30uA)
CT200...DC~500 kHz, ±(0.05% of reading +30uA)

What do you think about: "basic accuracy 40mA"? "30uA"?

With those specifications I'll give you my opinion:

The Yokogawa WT3000 has a margin error non less than 4.5uA, probably much more. Neither a scientist nor an engineer should claim a 0.2, 0.3 or 0.7uA +- 4.5uA or +-(much more uA)!

So, the specified 0.7uA was not actually measured.

I am not saying Weir is lying; he is not stupid, only a foolish could lie about that data.

Another possibilities must be explored.

1.- I am plain wrong.
2.- There is an error in published date. I think it is highly improbable: leakage has been specified for 4 samples (and also in previous public reports).
3.- Could be 0.7mA (0.7*10^-3A) correct? Impossible, 0.7uA is too much leakage, but 0.7mA is impossible (for one layer only).
4.- A multi layer capacitor was actually measured.
5.- Another one.

I think the most probable possibility is the 4th.

Even within the 4th case there are two possibilities:
a.- The specified leakage current is for whole multilayer capacitor. I would like this one, and it is possible, but it not appears to be the real case. “Layer” is explicit three times in table data. Futhermore, there is a conclusive prove it is not the real case, at least not with a lot of layers: if the published leakage was for a whole multilayer capacitor with a lot of layers we must had testified a much more powerful “leakage”: the share price should be now above the sky.
b.- The specified leakage current is the mean leakage for its layers: total leakage divided by the number of layers.

So I think most probable possibility is that was measured a multilayer capacitor and then published the mean data of leakage current.

I it is true, the bad leakage data is more representative: it is not an only one bad layer of 0.7uA; it is a statistically significative data, because is the mean of multiple layers.

There is no question this leakage is huge. With this leakage there is not possible EESU, the multilayer capacitor cannot storage energy for a long time (only a few thousand of seconds, as has been posted in this blog); Weir, EESTOR, ZEEN and us have a problem of several orders of magnitude, three. If it is not solved the Technology Agreement is not worth too much.

 

Disclosure:

I am not familiariced with specified equipment, muy opinion is based only in my interpretation of the specifications sheets.

I was lucky and had a good entry point, and have not too much money, so I shall not sell (now) :)

 

[JCatania]

I tend to think it wasn't a measurement, although measurements were made, they just weren't accurate. In John Galvagni/Zenn's previous report it was stated that the good ammeter was broken and that the one used could only measure to a uA or so but I believe it was stated that the value was believed to be less. And now we are hearing 2nA and such. It would be easy to calculate the from voltage decay. As far as I'm concerned a 10% drop in voltage might be a good rejection criterion. Actual current leak measurements may not be sensitive enough as done by Weir. He should not be ashamed if he derives this number from voltage.

[Y_Po]

can measure the current with "current Resolution 1uA" (micro A), it has display for current and can be controlled with computer. But the problem is obvious: preliminary EESTOR results of leakage currents are 0.0002, 0.0003 and 0.0007 mA: well out of its resolution, 3 orders of magnitude.

Check "3 orders of magnitude" part.

[Wynja]

Thanks, Y_Po

[Tecnófilo]

Thank you, Y_Po. Just corrected it, I had in mind we need 3 orders of magnitude improvement in leakage.

I'll  modifiy a bit my comments, and I have in mind to add images, later no time now.

Y_Po said:

can measure the current with "current Resolution 1uA" (micro A), it has display for current and can be controlled with computer. But the problem is obvious: preliminary EESTOR results of leakage currents are 0.0002, 0.0003 and 0.0007 mA: well out of its resolution, 3 orders of magnitude.

Check "3 orders of magnitude" part.

[ee-tom]

One point:

Measuring cap leakage is not as simple as it seems, because any change in dielectric characteristics that changes k will provide a pseudo-leakage current if the capacitor is fed from a voltage source of I = d(CV)/dt = V dC/dt  (dv/dt here is 0). App notes about measuring insulation will all note that you need to wait for the dielectric to reach equilibrium, and that temperature changes can cause issues. at higher k see is larger so percentage changes in C more significant and this is a greater problem.

So the obvious method (ammeter + supply) is problematic though maybe not impossible (you have to try it to see). For example the Yoga has a sensitive current sense input for use with an external CS resistor that could be used to measure low currents, and this could integrate over time so reducing errors.

But there is a much more accurate method, requiring no difficult equipment:

(1) connect supply.

(2) disconnect supply, wait 10 minutes.

(3) measure voltage.

The self-discharge over the 10 minute gap with nothing connected gives an accurate measure of leakage.

Given EEstor's like of presenting figures to unreasonable accuracy you might however expect the leakage current to be directly measured, since otherwise any calculation would less to more than one significant figure which EEStor (noting the k measurement) would not round. In that case an ammeter with 0.1uA resolution maybe was used for direct measurement. The Yoga integrating would surely provide better resolution than this. But we are speculating too much when we try to identify specific equipment.

[Tecnófilo]

I've added an image, attached files and links; and corrected the error pointed by Y_Po, wich not changes the arguments.

[Eenigma]

Y_Po / Tom stop with the non sense, you both know better. Measuring/calculating leakage is so easy a child that can read could do it.

1. Charge up to 1,500V
2. Monitor charge until it drops to half power or in this case 1,060V

At 1/2 life do the calculations based on known energy storage of the fully charged layer.

that's it..........end of story!

Measuring/calculating leakage within a tolerance good enough for a press release is a non issue. The only real issue with this press release is seeing the 3rd party report directly from the 3rd party.

[Y_Po]

Stupid dumbass, I have never said the quoted current was hard to measure. I said it was easy.

Also long time ago I said that original claims of ridiculously low leakage 0.1% per year or so (If I remember correctly) is impossible to reliably measure  without few years.

[JCatania]

It still says 3 orders! Why is 3 wrong?

[wcushman]

If the leakage is a ridiculously low rate such as 0.1% a year, then measuring the voltage drop over a month would show the leakage is low enough to be valuable although you would not accurately know the leakage rate. For example, if the voltage dropped from 1500 to 1400 over a month, then the stored energy would have declined by ~13% in 1 month, but you would have a very usable EESU if this were the only problem. Good enough for government work.

[PeterP]

On Another thread I Wrote:

When the SEAL guy said to Dick: “How accurately do you want leakage measured” Dick was amused. He said:

“This is a battery Fred, holds it charge for months, you tell me how you want to do it”.

A pause then: “Let’s charge it up, wait 5 minutes and measure whatever we get with the Fluke”.

Dick said “Done”

As he walked away he chuckled to himself: “Let’s see what the fool from Notre Dame makes of that”, then he started to laugh.

 

[wcushman]

And what leakage rate was reported?

PeterP said:

On Another thread I Wrote:

When the SEAL guy said to Dick: “How accurately do you want leakage measured” Dick was amused. He said:

“This is a battery Fred, holds it charge for months, you tell me how you want to do it”.

A pause then: “Let’s charge it up, wait 5 minutes and measure whatever we get with the Fluke”.

Dick said “Done”

As he walked away he chuckled to himself: “Let’s see what the fool from Notre Dame makes of that”, then he started to laugh.