Black Gate capacitor reforming and leakage

[AndrewT]

Hi,
just started re-forming the 220uF Black Gate in the MyRefC Negative FeedBack loop.

62V applied through a 1M0 started raising the DC voltage to ~3V, this took about 5mins, and then stuck there for a further 10mins leaking >15uA through the cap.

Changed the 1M0 resistor to 100k and continued.
Over about half a minute the voltage rose steadily to 20V and has risen very slowly to 20.5V over the next 20mins It appears that reforming current is about >400uA about half (245uA) of this passes a resistor string parallel with the cap. Leakage ~200uA

This is an odd reforming Voltage with Time characteristic that I have never experienced with conventional electrolytics.
I have another to do to see if it behaves similarly.

Any thoughts on why it behaves like this?
Could this reforming behaviour be linked to the excessive in circuit time that is reported until they sound right?

edit,
second showing similar but no parallel resistor string this time.
My estimate is very approximate, but it looks like from 3V to 4V it slows charging and again around 24 to 26V is slows.

[linuxguru]

No idea about the reforming characteristics of Black Gates, just a conjecture - a portion of the current through a BG is probably the tunneling current through graphite sheets/filaments in close proximity. This will not contribute to electrolytic forming, and can be considered to be leakage. Depending on the proportion of the tunnel current to the electrolytic current, it could be a reason for the extended time required for the aluminium oxide insulating barriers to form near the foil.

Another possibility is that graphite (a common industrial catalyst) can catalyze the reduction of the aluminium oxide barrier back to Al ions in electrolyte solution, thus retarding the formation of the barrier.

One thing I can confirm - for the first few hours of in-circuit break-in of Black Gates, they sound extremely harsh and grainy. It has an audible similarity to shot-noise, but mysteriously quietens down in a few hours. It continues to sound harsh and edgy, but the highs smooth out a bit in a few hours, followed by the upper mids (which takes much longer) and then the lower mids (which is very gradual).

[Ian Finch]

Also conjecture:
From construction and some likely potentials within the cell(s), Linuxguru is on the track. The second potential plateau you note at 20 V (or is that 24-26V?) indicates that a significant process requiring considerable emf is occuring. It had better not be a part of the electros normal AC transfer. That it presents this way instead of a standard electros. curve is likely due to the much greater effective conductance of the electrolyte. i.e. It is only now apparent. You could also express this in terms of shorter ionic paths as do Jelmax. It is a pity that the manufacturers did not evidently comment on this whilst it would almost certainly be evident in manufacture.

I assume then, that this marks the formation of the AlO layer you are hoping for.
Can we guess that this is a 25-63V rated electro being charged from a supply rail?
BTW, Is it possible to retry this on same caps well after discharge to see if any hint of delay at 3V level is still evident? This might be a real pain.

Reported "graininess" with NOS caps might make perfect sense if actually unformed.
'can't accept that should normally occur with new stock, though.
I don't know about ambient temp. catalysis with straight graphite here, either.

[AndrewT]

the Black Gates are 220uF 50V.
The 1M0 feed resistor keeps the capacitor voltage low while the cap is charging and/or passing leakage and/or reforming.

As the cap stops charging and the reforming current drops then the cap only has leakage passing. That's when one needs to check that one does not overvoltage the capacitor. I reduce the PSU voltage as the cap approaches 50V.

I have inadvertently reformed to much higher than rated voltage in the past. With ordinary electrolytics this did not seem to harm them.

I asked, but got no answer, "can I re-rate a reformed electrolytic" eg. if it's standard voltage is 63Vdc and reformed to 70Vdc. Can it be used long term at 68Vdc, just as an example.

[Mooly]

Originally Posted by AndrewT

I asked, but got no answer, "can I re-rate a reformed electrolytic" eg. if it's standard voltage is 63Vdc and reformed to 70Vdc. Can it be used long term at 68Vdc, just as an example.

I suspect that depends as much on the individual cap as anything else. I am a sure you can get away with it on some, certainly for steady state DC conditions, perhaps problems might show if the cap was used at it's rated temperature and ripple current causing the dielectric film to degrade quickly.

Strange question from you Andrew Maximum ratings are there for a reason, and must be adhered to if reliability is to be ensured.

[DRC]

I asked, but got no answer, "can I re-rate a reformed electrolytic" eg. if it's standard voltage is 63Vdc and reformed to 70Vdc. Can it be used long term at 68Vdc, just as an example.

I have absolutely no idea !!
I assume the capasitance must go down somewhat, if a thicker oxide layer is formed ...

What differences are there in the manufacturing process for electrolytic caps with different voltages ?

There doesn't appear to be many variable to play with ...

dc

[Todd-uk]

Hi
I performed this excercise on the blackgates (feedback and input coupling) in a Rotel RA935BX I bought recently. I did it because I suspected from the schematic that the BGs have only 100 0r so mV across them which is very low and the amp had not been run for while. It took a couple of hours for the leakage to stablise and involved removing the caps from the pcb. Sound improved markedly after this procedure and continued to improve with a few tens of hours running time.
It seems rather careless of any manufacturer to utilise 'lytics where they receive bugger all bias but I suppose 10 and 100uf film caps are bit bulky....
Re burn in time, the blackgates in my tube amp took about 100 hours to bed in and indeed they need a good few minutes to recover after a few days rest.
Fglorious things blackgates, its almost a crime that they have been discontinued.....

Regards

Todd

[Damon Hill]

Would similar effects be expected in bipolar (non-polar?) N and NX series Black Gates? I use them in the feedback circuit where they see millivolt DC and not very much AC either. Note that Nichicon (and other) AC caps have lower DF as their voltage rating goes up, but I've always used the lowest voltage parts available.

I've always made the assumption that these particular parts are true AC caps and not two polar caps in series.

[Ian Finch]

Originally Posted by AndrewT

the Black Gates are 220uF 50V.

Thanks, Andrew

As far as I can determine, the voltage rating of electros and most other caps is set by plate spacing. i,e. membrane, plastic film etc. thickness.

The thickness of the electros dielectric formed in process is not used to rate voltage. However, it is certainly increased incrementally to provide greater rupture proofing across higher voltage ranges, at a penalty to size and ESR.

Blackgates electros are different in that they have a partly conductive membrane but I imagine the electro rating systems would still apply.

[hamsubboy]

excuse me, i also have a question on the usage of cap.
I have attached an application circuit which the amp can be used as BTL or single end.
The capacitors C6 & C7 should be used as cut off low frequency.
I dont understand why we dont need to add these cap when using as BTL