Reforming caps?

[Phil_S]

I've searched TAG on this subject but only found a partial answer. What I found is not new information -- that e-caps can be reformed, but the discussions all suggest this is done one cap at a time while the cap is not installed.

Over a period of several years, I was actively building. After about 12-15 amps, I stopped building abruptly. This includes a last almost finished amp that has just sat for about 10 years along with all the others. Now, I've got this odd collection of amps that have stale caps in them. (I've also been on a guitar playing hiatus due to age-related problems on the left hand, arm, neck and back. Physical therapy works up to a point, but it's been a see-saw. This is the main reason my amps don't get used. I'd like to see if I can get some chops back.) When I fire up an amp, it works OK for a few minutes and the it exhibits a loud, intermittent, very unpleasant crackle static-like sound. I'm guessing cap jobs all around might be needed.

Before getting into changing all the electrolytic caps, I am trying to determine if the existing caps can be reformed while remaining in the amp. Like an electron, I'd prefer to avoid pulling every chassis for a cap job. Is there a way to do this that might succeed in some cases?

I don't have a variac and am not interested in buying one. I do have a light bulb limiter and a modest supply of incandescent bulbs. I've been thinking there might be a way to do this, but am not really clear. Suggestions would be appreciated.

[Stevem]

I don't think your issue is E cap related, I think you have noisy/ plate load resistors.

Does the amp still have the issue if you pop each preamp tube out one by one and do so moving back upstream from the output tubes?

[R.G.]

Re-forming caps in situ in the amp is a dicey process. The general advice of "bring it up slowly on a variac" is quite tricky to do well in amps with vacuum rectifiers because the rectifier itself doesn't start applying DC until its filaments get hot enough to conduct a bit. That means there is no forming at all going on until you get to ??? 75%? 80% of line voltage, and then the voltage/current comes up very quickly with tiny moves on the variac.

The key to forming or re-forming capacitor oxide layers is low, slow current. A high DC voltage through a big resistor limits the current until oxide forms in weak spots. When the voltage across the big resistor drops down enough to indicate low leakage, the resistor can be taken out.

One way I just made up to do this inside an amp is to take solid state rectifiers loose from the first filter cap and run this into a temporary new/known good filter cap. This can be smallish, maybe 10uF or so. If the amp has a vacuum rectifier, you can bridge in solid state diodes on the tube socket or some such. The new, added filter cap then has a 470k or so resistor from its + to the + of the in-situ first filter cap. Remove all the tubes and clip a voltmeter across that resistor. Turn the rig on, being prepared to leave it on for hours to days if needed.

The SS diodes and new/good cap make a high voltage DC supply which feeds the big resistor into the amp's own filter caps. The amp's caps are connected through the dropping network, so they all form in place. The resistor limits current to forming levels, not destructive levels in any weak spots. Hmm. I gotta write this up with pictures, I think.

[didit]

Agree fully R.G., as that was conjured image I'd painted in mind, less invasive but still needs some brief limited access within the chassis ..

[Phil_S]

Thank you for the replies. I see the best course of action is to recap each amp, or, pull the caps and reform them out of circuit. I have a pile of NOS good quality caps that I could start to reform for the cap jobs. If these turn out bad, I can purchase what I need. I'm mostly retired at this point, so, really, I have the time to plod through this. I was just hoping for instant gratification. We all know that isn't a thing.

Stevem, I appreciate your alternate diagnosis. When I pull a chassis from its cab, I'll look for that, too. However, none of these amps have been gigged, and all have seen only very light use. My thinking is that noisy plate load resistors would have been obvious when the build was completed.

[Guy77]

Thank you for the replies. I see the best course of action is to recap each amp, or, pull the caps and reform them out of circuit. I have a pile of NOS good quality caps that I could start to reform for the cap jobs. If these turn out bad, I can purchase what I need. I'm mostly retired at this point, so, really, I have the time to plod through this. I was just hoping for instant gratification. We all know that isn't a thing.

Stevem, I appreciate your alternate diagnosis. When I pull a chassis from its cab, I'll look for that, too. However, none of these amps have been gigged, and all have seen only very light use. My thinking is that noisy plate load resistors would have been obvious when the build was completed.

Hi Phil. I am curious, how old are these electrolytic capacitors.

Edit : Just noticed you mentioned 10yrs
I believe checking them with and ESR meter would be a good option here, besides visual inspection for any physical defects.

Cheers
G

[R.G.]

I did a little looking. Turns out that there is a USA military specification on storage and reforming of electrolytic caps. It's here:

https://www.w7ekb.com/glowbugs/Military ... lytics.pdf

[didit]

While out for a walk another variation crept in. If there's a socket for vacuum rectifier one could box up a small high voltage transformer with silicon rectification, that small 10uF capacitor, power resistor, and meter or meter test-points all in a fairly safe package that'd plug into that socket and deliver a low current high voltage onto the B+ chain. This skips using the amp's built in PT and should obviate opening the chassis, that is unless one concludes the reformations failed.

Or is there a gap missing?

..

[didit]

Turns out that there is a USA military specification on storage and reforming of electrolytic caps.

But of course there is, excellent ..

[nuke]

Go browse eBay for a used electrophoresis power supply. Cheap as dirt, adjustable, regulated 0-500VDC power supply, plenty of current. Most are not particularly clean for powering an audio amplifier, but they are plenty good for testing capacitors at voltage and applying a re-forming voltage with voltage and current control.

I too, have a bunch of caps that are old, but new in the package from when I was more actively working on things. I have used the electrophoresis power supply to reform them before using them on my own. I have a pile of Sprague atoms from the early 2000's, (when they were still well made) and F&T's and so on.

I won't use them in repairing anyone's amp, but I use them for my own experiments, builds and stuff.

I usually put several on a clipboard jig I made with a current sensing/limiting resistor on each capacitor and do about 10 at a time. Bring up the DC slowly, monitor each capacitor until I reach the rated voltage, then let them run for 24-48 hours. Usually within a few hours, the leakage is very minimal, and gone in 24 hours.

I've only seen 1 bad LCR multi-section cap fail in all the runs I've done. It just had excess leakage and got slightly warm to the touch.

The old Sprague Atoms were intended for very long shelf-life at repair shops and what not. The old ones seem to live up to that. After I think about 2006, something changed and they don't seem to hold up as well.

Weirdly enough, leaky caps usually test *higher* than their label value, especially on multimeters with a capacitor test function. It has to do with the way typical multimeters do the test.

[R.G.]

Weirdly enough, leaky caps usually test *higher* than their label value, especially on multimeters with a capacitor test function. It has to do with the way typical multimeters do the test.

That has to do with what is happening in the oxide layer. The forming process actually grows aluminum oxide on the surface of the aluminum, then uses voltage to force it to grow thicker. It's an analog to what happens when anodizing aluminum for a protective layer. Any gaps in the oxide layer get plugged by the electric field being stronger in the gap, and there being more force in the gaps to get another grain of oxide to grow.

The oxide is initially quite low-voltage capable, but fills up with time and oxide growth. As this happens, higher voltage is needed to grow oxide in the leakiest spots. The oxide layer forms up until all of the surface will withstand the forming voltage without leaking.

But an oxide layer is a forced condition, so it "relaxes" slowly and aluminum oxide breaks down and thins the oxide layer. This may happen unevenly, with some thinner, higher leakage spots. But overall, with time, the whole oxide layer gets thinner.

And that accounts for both the higher leakage and for the bigger capacitance. A thinner insulating layer directly increases the capacitance. You can think of each tiny square area of the foil being a small capacitor in parallel to all the others. Some have thin spots on them, so they contribute more capacitance, but resist voltage less well.

Yeah, capacitance goes up as the oxide ages and thins.

[Guy77]

This increase in capacitance is a another great reason to test caps with an ESR meter . You can see the change in capacitance as well as the increase in resistance which is another tell tale sign that they are going bad.

Cheers
G

[nuke]

That has to do with what is happening in the oxide layer. The forming process actually grows aluminum oxide on the surface of the aluminum, then uses voltage to force it to grow thicker. It's an analog to what happens when anodizing aluminum for a protective layer. Any gaps in the oxide layer get plugged by the electric field being stronger in the gap, and there being more force in the gaps to get another grain of oxide to grow.

The oxide is initially quite low-voltage capable, but fills up with time and oxide growth. As this happens, higher voltage is needed to grow oxide in the leakiest spots. The oxide layer forms up until all of the surface will withstand the forming voltage without leaking.

But an oxide layer is a forced condition, so it "relaxes" slowly and aluminum oxide breaks down and thins the oxide layer. This may happen unevenly, with some thinner, higher leakage spots. But overall, with time, the whole oxide layer gets thinner.

And that accounts for both the higher leakage and for the bigger capacitance. A thinner insulating layer directly increases the capacitance. You can think of each tiny square area of the foil being a small capacitor in parallel to all the others. Some have thin spots on them, so they contribute more capacitance, but resist voltage less well.

Yeah, capacitance goes up as the oxide ages and thins.

Yeah, I've also seen similar problems testing bad film and PIO capacitors. Including some "boutique" PIO guitar tone caps. They're not electrolytic of course. However, my Fluke multimeter is totally confused by them. The PIO .022uf tone cap from a boutique maker, reads about 6-7 megohms of DCR. (brand new, multiple examples). My Fluke 289 multimeter reads them as 0.047uf, or about twice their actual value.

However, a digital LCR bridge measures it as a .022uf, with the mentioned DCR in parallel with it.

Something about how the multimeter measure capacitance is vulnerable to error when there's a lot of leakage. I think it is some sort of ramp/time delta measurement, rather than the AC mode method the LCR bridge uses.

But, a good thing to be aware of is that a cap that reads more than it is supposed, may be bad, just as much as one that reads lower.

[Helmholtz]

It has to do with the way typical multimeters do the test.

Yes, leakage increases the capacitance reading on most DMMs

[R.G.]

But, a good thing to be aware of is that a cap that reads more than it is supposed, may be bad, just as much as one that reads lower.

Yes - it's one of those hints that something just might be hiding from you.

Another side note in general for people who consider re-forming caps; the manufacturers run aluminum foil through anodizing/forming baths of dissolved chemicals to enhance uniformity of the films and fast forming. The baths, forming voltage and current are fine tuned for best uniformity and production speed at large scale.

The foils are then wound into their desired format, with a mechanical spacer (i.e. porous paper/films) and an un-formed foil, then put into the outside can, which is filled with gelled electrolyte. The electrolyte is chemical-balanced for best performance in terms of capacitance, low ESR, etc.

But the performance electrolyte may not be the best stuff for forming the oxide, just like the high speed forming bath may not be the best for final performance. Back in the Golden Age, the two solutions were closer, but modern electros have been highly optimized for, among other things, not being fluid and runny so they don't boil and evaporate, that kind of thing.

Re-forming in the can may not be as effective or fast as the manufacturing process was. It's not optimized for that.

I have come to view re-forming capacitors as a less desirable path that I would use if I had to, but I'd really prefer to replace the caps.