Capacitor Burn-In/Break-In

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2004
Paid Member
Here's a relatively simple way to reduce the number of variables involved in the experiment, by using scientific method.. Not all bases are covered, but it should provide some interesting ammunition for the assertions some of us make. (like me.. LOL)

Measure and select a pair or more of new, unused capacitors of the same make, lot code, tolerance value, and match to say 1% or better. If you can measure DA/DF at the start so much the better.

Get a make-before break switch spdt or rotary type and connect the capacitors such that you can select one or the other. Designate one as the "control" and the other as the "subject."

Install the caps and switch in the Elgar line conditioner.

Measure and record the spike amplitude of each capacitor at the first application of power, technically they should be the same, or certainly close enough that the scope's inaccuracy will mask any difference.

Note that all measurements should be performed at approximately the same time each day, short of using an expensive sinusoidal constant voltage supply this is the best way to get relatively consistent results. (Average local daily usage patterns should not differ that much.) Also measure the RMS voltage available at the outlet at the time of measurement, large variations may have an effect on the result.

Leave the "subject" capacitor in circuit for some defined number of hours, then remeasure both, making sure to always switch back to the "subject" capacitor as soon as the measurement has been taken. Continue to do this for at least 100 hours. If indeed something is happening to the capacitor with use, the two capacitors will no longer have matching spike amplitudes - if they match to within a couple of % you can reasonably safely conclude that something else is at work here.

At it's simplest the experiment requires no more than two closely matched same brand (etc.) caps and a switch, which for some transient reduction should preferably be make before break.

Improvements:
Using a good toroid power transformer for isolation, check for the presence of spikes and other noise on the AC line. (EI transformers have too much leakage inductance in many cases to be useful for this. Record any anomalies found, do at same time as capacitor measurements for consistency. (This measurement isn't strictly necessary, but may provide additional useful information.)

Note that if you measured DA/DF at the start you can remeasure at the end to see if there have been any changes.

Monitor and maintain consistent ambient temperature during test.

Comment:
In many years of component qualification, switching only the components being evaluated has proven to be far more accurate.
Comparing two complete devices is far less accurate because of all of the variability added by having two complex devices, which by their relative complexity can never be identical. Using one master device for comparison eliminates all of those (often unknown) variables.
 
Nice pix. Wonder what the reference sine feedback loop speed is? Probably too slow to begin to notice such a perturbation. Any idea of the peak capacitor current during the spike?

I'm sorta bemused by the notion that you turned it on and off at the same point in the waveform evening and morning, and advanced your timing from day to day. Formidable!

I'm amazed at the failure of the critics to point out that it is a worthless observation because you did not use a Braille oscilloscope. :D

I wish I knew more about the Cardas testing. The triangle waveform testing was particularly of interest, and I'd like to know more about the methodology. Unfortunately since retiring I no longer have access to such equipment. Testing a cable before and after burnin might be enlightening.
 
For some reason the majority of people in the US voted W into office, not once, but twice! What does that tell you about the wisdom of the masses?

I_F

Hah - good one, I_F. (Just saw this thread.)

What does that tell you about the wisdom of the masses? ... I suggest it tells you that the US "democratic voting system" is completely f****d and you need to re-do your Constitution! :D

Regards,

Andy
(Aussie)
 
Hah - good one, I_F. (Just saw this thread.)

What does that tell you about the wisdom of the masses? ... I suggest it tells you that the US "democratic voting system" is completely f****d and you need to re-do your Constitution! :D

Regards,

Andy
(Aussie)

Even more bizzare is that we elected a neophyte Senator from Illinois with no record of accomplishment in the Senate to the highest office in the land. Hope for change but the outlook is not promising!
 
Ok guys, back on topic now ok. I am into the midst of this hoopla concerning capacitor burn in with my crossovers. I re-capped them and installed a notch filter because a new tweet was added. It is a well known mod for my Polk SDA2Bs . Anyways they are bright as the sun now. Many that have the mod are telling wait 3-400 hours of listening and they will smooth out.

Heck I may be dead by then!
 
I changed the rectifier bridge from the old, heavy workhorse (circa. 1985) to a new, sprite FRED module. No change in the spike. I also swapped out the three circuit boards, exchanging them with those in my other Elgar, and no change.

That leaves: the wiring, the two banks of output transistors, and the transformer.

And the capacitors.

Here's today's visual.

Turn everything off and let the stored charge bleed down. Now place a moist paper towel around each of the new capacitors. Let the moisture content go up for say a day. Then try your test again. I suspect they will revert to like new conditions!
 
If, as example, old capacitors can be reformed using a slowly applied voltage, it doesn't seem to me to be outrageous that something similar could happen with newer capacitors. As a self-styled scientific type, I'd like to have some sort of an answer to actual mechanisms, though the alternative is probably more blissful.
BTW, in each and every case, if a difference is heard something is being measured.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.