Reforming Electrolytic Capacitors

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
emrr said:
here's a veer.....can you slow bake the absorbed moisture (noise) back out of NOS carbon resistors? 

I have heard stories of folks baking carbon comps and then sealing them (shellac or varnish?) so they don't absorb moisture again.  Sounded like audiophool mythology but I've been wrong before!
 
www.cde.com/resources/catalogs/AEappGUIDE.pdf        <great article

blurb-o-matic>



Reforming Electrolytic Capacitors

Home

Warning: The voltages involved with most electrolytic capacitors used in valve based equipment are lethal. If in doubt seek expert help. The old adage - keep one hand in your pocket while it is switched on or the capacitors are charged - should be followed.

1. Current and Voltage Limited Method

The electrolytic capacitor is a critical part of both old and modern electronic equipment and it must be  used correctly in order to get the longest and safest operational life and is particularly important with high voltage versions of these components. Electrolytic capacitors rely on a chemical process to provide the insulator between the two metal plates and this process can degrade over a period of years if the capacitor has not had power applied. The result is that the working voltage of any electrolytic capacitors in equipment gradually falls. If full power is applied to long unused equipment then the electrolytic capacitors can pass excessive amounts of current that could cause a catastrophic failure to the entire equipment and a potential fire hazard to surrounding property.

The correct course of action is to ensure that each electrolytic capacitor’s insulation layer is ‘reformed’ by the application of a current and voltage limited DC supply to each individual capacitor. Current limiting ensures that the heat generated within the capacitor is kept at a sufficiently low level that damage does not occur. My preferred method is to carefully disconnect each electrolytic capacitor and apply a voltage, equal to the working voltage of the respective capacitor, via a suitable current limiting resistor to that capacitor.

For example, for a 450V working capacitor, I apply 450V DC, observing the correct polarity, via a 470K 2W resistor to the capacitor and measure the voltage drop across the resistor with a volt meter - see the following circuit.

The circuit assumes that the negative line of the power supply is connected to ground but this is not mandatory. Should the positive line be grounded instead then move the 470K resistor to the positive side of the capacitor. Care is needed if the capacitor case is internally connected to one of the capacitor wiring tags.

Over a period of time, which can be up to 24 or more hours for older components, the voltage across the resistor will fall and eventually stabilise at some much lower value. My rule of thumb is that if the voltage drop across the resistor after 24 hours is significantly more than 22V (indicating a leakage current in excess of 50 microamps) than I repeat the reforming process. If no improvement is obtained then I replace the capacitor with a new one. You may also find that very old capacitors have dried out and cannot be reformed in which case they must be replaced. A similar process may be required for new electrolytic capacitors that were manufactured only one or two years ago - I always check to make sure.

Capacitors with higher values will have higher natural leakage currents and may require a correspondingly lower value current limiting resistor. For lower voltage capacitors like 10,000uF 25v I use a 10K series resistor on a 25V DC source.

Once the reforming process is complete then the capacitor may be fully discharged with a resistor (not a short circuit), disconnected from the reforming supply and reconnected to its original circuit. As soon as any further inspections or tests on the equipment are completed then it may be powered up.

NB Another well known but separate problem with older equipment is that solid carbon resistors will gradually show an increase in their resistance and it is not unusual to see increases from 50% to 500% after periods in excess of twenty years. All resistors outside of their original tolerance should be replaced - check that the wattage and voltage ratings of the replacement resistors are suitable for the proposed application. Carbon film resistors do not appear top suffer from this problem.

2. Alternative Reforming Method

It has been suggested that a variac and a low power incandescent llight bulb wth the same working voltage as the incoming supply connected in series with the AC supply to the unit in question may be used to reform electrolytic capacitors without the need to remove them from circuit. The unit is connected to the variac and bulb and powered up with the variac set to zero. The variac is increased to say10V or sufficient to make the bulb glow weakly and then left for a few minutes after which the bulb should be glowing less brightly. The process is repeated until the variac is set to output the correct mains voltage for the unit under test.

This may well work but my concern is that the lamp may not provide sufficient current limiting to prevent damage.  One downside of this method is its use where there is a constant current drain by the unit under test - for example where there are valves present as this will cause the lamp to glow continuously and damage may take place to the valve heater emissive area when operated on low voltages.

There is an additional problem in selecting the bulb with the correct  power rating for each application.

My preference is for method 1, even though it may be more complicated and require more patience.
 
Whoops said:
Tried to find specs just for fun, on the Fluke 179 meter specs they state + - 1,2%  accuracy.

On the Chinese Meter I have no specs but the similar models I find online state 0.5% for values under 200uf:

Accuracy                         
  ±0.5%    (<2000uF)
  ±1.0%    (2000uF)
±2.0%    (20mF)

200.0pF (0.5%),
2nF, 20nF, 200nF, 2µF, 20µF, 200µF (0.5%),
2,000µF(2%)
20,000µF(4%)


So this might explain the China meter reading lower than the Fluke under 1000uf, and the values being similar above that

This is as much accuracy as one would ever need for electrolytics. If your meters' readings are within 10 percent of each other, they're doing okay and you're getting the answers you need.

Are all these measurements before the capacitors are reformed? I'd go ahead and reform them, toss out any that leak too much current (does the Philips datasheet say what the max leakage should be?), and THEN measure the capacitance and ESR of the ones that reformed well.
 
> end of life mechanism that can make electrolytic value go high? 

I would say that if the oxide is re-dissolving, C will rise, also leakage, and blow-up voltage will drop.

Little caps like these can't safely dissipate 1 Watt. At 35V that is 28mA. So get a 35V supply and a 2K resistor. Soak for 5 minutes and check for heat-- I'd say if they heat they will never be reliable and should be tossed (perhaps salvaging the wraps for faking vintage look on new caps). If not gonna blow, soak them a few hours to a few days. The cap voltage should come up within mV of the supply voltage, indicating near-zero current (10mV in 2K is 5uA, and 1uA/uFd is a common spec).
 
benb said:
Are all these measurements before the capacitors are reformed?

I'd go ahead and reform them, toss out any that leak too much current (does the Philips datasheet say what the max leakage should be?), and THEN measure the capacitance and ESR of the ones that reformed well.

Yes, I still haven't done any reforming.
I followed ln76d advice "First measure capacitance as also ESR if you have an option, then think about "reforming them".  "

I tried to find Philips datasheet, but can't find it for this low voltage, axial, -10% to 50% capacitors
These capacitors are not produced any longer.
I found datasheets for Philips Axial High voltage, and for Philips axial +-20%, not the same ones I have

I don't know if Vishay took over the capacitor department of Philips or BC, but when I search for Philip caps Vishay datasheets are always the first ones to appear.
 
here is a  Dept of Defense file>

http://everyspec.com/MIL-HDBK/MIL-HDBK-1000-1299/MIL_HDBK_1131_1788/

looks like e-caps are shrinking in size because they rough up the foil>

"Dry" aluminum capacitor
A "dry" electrolytic capacitor with 100 µF and 150 V

The ancestor of the modern electrolytic capacitor was patented by Samuel Ruben in 1925, who teamed with Philip Mallory, the founder of the battery company that is now known as Duracell International. Rubens idea adopted the stacked construction of a silver mica capacitor. He introduced a separated second foil to contact the electrolyte adjacent to the anode foil instead of using the electrolyte-filled container as the capacitor's cathode. The stacked second foil got its own terminal additional to the anode terminal and the container no longer had an electrical function. This type of electrolytic capacitor combined with an liquid or gel-like electrolyte of a non-aqueous nature, which is therefore dry in the sense of having a very low water content, became known as the "dry" type of electrolytic capacitor.[

With Ruben's invention, together with the invention of wound foils separated with a paper spacer 1927 by A. Eckel of Hydra-Werke (Germany) the actual development of e-caps began.

William Dubilier, whose first patent for electrolytic capacitors was filed in 1928, industrialized the new ideas for electrolytic capacitors and started the first large commercial production in 1931 in the Cornell-Dubilier (CD) factory in Plainfield, New Jersey.  At the same time in Berlin, Germany, the "Hydra-Werke", an AEG company, started the production of e-caps in large quantities.
Miniaturisation of aluminum electrolytic capacitors driven by progress in the anode foil etching process

Already in his patent from 1896 Pollak wrote that the capacitance of the capacitor increase by roughening the surface of the anode foil. Today (2014), electrochemically etched low voltage foils can achieve an up to 200-fold increase in surface area compared to a smooth surface.[ Advances in the etching process are the reason for the dimension reductions in aluminum electrolytic capacitors over recent decades.

For aluminum electrolytic capacitors the decades from 1970 to 1990 were marked by the development of various new professional series specifically suited to certain industrial applications, for example with very low leakage currents or with long life characteristics, or for higher temperatures up to 125 °C."

forgot about ECS, (exploding cap syndrome)>

Water based electrolytes

With the goal of reducing ESR for inexpensive non-solid e-caps from the mid-1980s in Japan, new water-based electrolytes for aluminum electrolytic capacitors were developed. Water is inexpensive, an effective solvent for electrolytes, and significantly improves the conductivity of the electrolyte. The Japanese manufacturer Rubycon was a leader in the development of new water-based electrolyte systems with enhanced conductivity in the late 1990s The new series of non-solid e-caps with water-based electrolyte was described in the data sheets as having "Low-ESR", "Low-Impedance", "Ultra-Low-Impedance" or "High-Ripple Current".

A stolen recipe for such a water-based electrolyte, in which important stabilizing substances were absent, led in the years 2000 to 2005 to the problem of mass-bursting capacitors in computers and power supplies, which became known under the term "capacitor plague". In these e-caps the water reacts quite aggressively and even violently with aluminum, accompanied by strong heat and gas development in the capacitor, and often led to the explosion of the capacitor.
Electrical characteristics" 

even ceramics lose value with age, except NPO's  looks like heat can fix those>



 
different testing techniques will give different results>

The standardized measuring condition for e-caps is an AC measuring method with 0.5 V at a frequency of 100/120 Hz and a temperature of 20 °C. For tantalum capacitors a DC bias voltage of 1.1 to 1.5  V for types with a rated voltage ≤2.5 V, or 2.1 to 2.5 V for types with a rated voltage of >2.5 V, may be applied during the measurement to avoid reverse voltage.

The capacitance value measured at the frequency of 1 kHz is about 10% less than the 100/120 Hz value. Therefore, the capacitance values of electrolytic capacitors are not directly comparable and differ from those of film capacitors or ceramic capacitors, whose capacitance is measured at 1 kHz or higher.

Measured with an AC measuring method with 100/120 Hz the capacitance value is the closest value to the electrical charge stored in the e-caps. The stored charge is measured with a special discharge method and is called the DC capacitance. The DC capacitance is about 10% higher than the 100/120 Hz AC capacitance. The DC capacitance is of interest for discharge applications like photoflash.

The percentage of allowed deviation of the measured capacitance from the rated value is called the capacitance tolerance. Electrolytic capacitors are available in different tolerance series, whose values are specified in the E series specified in IEC 60063. For abbreviated marking in tight spaces, a letter code for each tolerance is specified in IEC 60062.

    rated capacitance, series E3, tolerance ±20%, letter code "M"
    rated capacitance, series E6, tolerance ±20%, letter code "M"
    rated capacitance, series E12, tolerance ±10%, letter code "K"

The required capacitance tolerance is determined by the particular application. Electrolytic capacitors, which are often used for filtering and bypassing, do not have the need for narrow tolerances because they are mostly not used for accurate frequency applications like in oscillators.
 
"Series-equivalent circuit model of an electrolytic capacitor

The electrical characteristics of capacitors are harmonized by the international generic specification IEC 60384-1. In this standard, the electrical characteristics of capacitors are described by an idealized series-equivalent circuit with electrical components which model all ohmic losses, capacitive and inductive parameters of an electrolytic capacitor:

    C, the capacitance of the capacitor
    RESR, the equivalent series resistance which summarizes all ohmic losses of the capacitor, usually abbreviated as "ESR"
    LESL, the equivalent series inductance which is the effective self-inductance of the capacitor, usually abbreviated as "ESL".
    Rleak, the resistance representing the leakage current of the capacitor"

ESR effects>

"The equivalent series resistance (ESR) summarizes all resistive losses of the capacitor. These are the terminal resistances, the contact resistance of the electrode contact, the line resistance of the electrodes, the electrolyte resistance, and the dielectric losses in the dielectric oxide layer.

For electrolytic capacitors generally the ESR decreases with increasing frequency and temperature.

ESR influences the remaining superimposed AC ripple behind smoothing and may influence the circuit functionality. Related to the capacitorm ESR accounts for internal heat generation if a ripple current flows over the capacitor. This internal heat reduces the lifetime of non-solid aluminum electrolytic capacitors or influences the reliability of solid tantalum electrolytic capacitors.

For electrolytic capacitors, for historical reasons the dissipation factor tan δ will sometimes be specified in the relevant data sheets, instead of the ESR. The dissipation factor is determined by the tangent of the phase angle between the capacitive reactance XC minus the inductive reactance XL and the ESR. If the inductance ESL is small, the dissipation factor can be approximated as:

    tan ⁡ δ = ESR ⋅ ω C

The dissipation factor is used for capacitors with very low losses in frequency-determining circuits where the reciprocal value of the dissipation factor is called the quality factor (Q), which represents a resonator's bandwidth."

ok, so here is your vector sum, equiv circuit and effect on Z curve, which will then change after reform, e-caps form a series resonant circuit due to self inductance> L, C and R (esr)>
 
"Lifetime
The electrical values of aluminum electrolytic capacitors with non-solid electrolyte change over time due to evaporation of electrolyte. Reaching the specified limits of the electrical parameters, the time of the constant failure rate ends and it is the end of the capacitor's lifetime. The graph show this behavior in a 2000 h endurance test at 105 °C.

The lifetime, service life, load life or useful life of electrolytic capacitors is a special characteristic of non-solid aluminum electrolytic capacitors, whose liquid electrolyte can evaporate over time. Lowering the electrolyte level influences the electrical parameters of the capacitors. The capacitance decreases and the impedance and ESR increase with decreasing amounts of electrolyte. This very slow electrolyte drying-out depends on the temperature, the applied ripple current load, and the applied voltage. The lower these parameters compared with their maximum values the longer the capacitor's “life”. The “end of life” point is defined by the appearance of wear-out failures or degradation failures when either capacitance, impedance, ESR or leakage current exceed their specified change limits.

The lifetime is a specification of a collection of tested capacitors and delivers an expectation of the behavior of similar types. This lifetime definition corresponds with the time of the constant random failure rate in the bathtub curve.

But even after exceeding the specified limits and the capacitors having reached their “end of life” the electronic circuit is not in immediate danger; only the functionality of the capacitors is reduced. With today's high levels of purity in the manufacture of electrolytic capacitors it is not to be expected that short circuits occur after the end-of-life-point with progressive evaporation combined with parameter degradation.

The lifetime of non-solid aluminum electrolytic capacitors is specified in terms of “hours per temperature", like "2,000h/105 °C". With this specification the lifetime at operational conditions can be estimated by special formulas or graphs specified in the data sheets of serious manufacturers. They use different ways for specification, some give special formulas, others specify their e-caps lifetime calculation with graphs that consider the influence of applied voltage.  Basic principle for calculating the time under operational conditions is the so-called “10-degree-rule”.

This rule is also known as Arrhenius rule. It characterizes the change of thermic reaction speed. For every 10 °C lower temperature the evaporation is reduced by half. That means for every 10 °C lower temperature the lifetime of capacitors doubles. If a lifetime specification of an electrolytic capacitor is, for example, 2000  h/105 °C, the capacitor's lifetime at 45 °C can be ”calculated” as 128,000 hours—that is roughly 15 years—by using the 10-degrees-rule.

However, solid polymer electrolytic capacitors, aluminum as well as tantalum and niobium electrolytic capacitors also have a lifetime specification. The polymer electrolyte has a small deterioration of conductivity caused by a thermal degradation mechanism in the conductive polymer. The electrical conductivity decreases as a function of time, in agreement with a granular metal type structure, in which aging is due to the shrinking of the conductive polymer grains. The lifetime of polymer electrolytic capacitors is specified in terms similar to non-solid e-caps but its lifetime calculation follows other rules, leading to much longer operational lifetimes.

Tantalum electrolytic capacitors with solid manganese dioxide electrolyte do not have wear-out failures so they do not have a lifetime specification in the sense of non-solid aluminum electrolytic capacitors. Also, tantalum capacitors with non-solid electrolyte, the "wet tantalums", do not have a lifetime specification because they are hermetically sealed and evaporation of electrolyte is minimized.

Electrolytic capacitors with solid electrolyte do not have wear-out failures so they do not have a lifetime specification in the sense of non-solid aluminum electrolytic capacitors."
 
Back in SA when quality components where very rare due to sanctions we used the reform ALL caps, onlycan remember one failing when in service.

wE used to use an old Heathkit IT 11 cap tester for all duties, I picked one up from e-bay quite cheap, check it out...
 
This has become a really interesting thread with a lot of useful info.

I have a lot of work at the moment, but I will spend my Christmas reading all this info and learning more.

Thank you CJ and thank you all for the information and advises shared

 
A friend point me to some evil bay meter,
I was checking this one:
http://www.ebay.co.uk/itm/Portable-MK328-LCR-ESR-Tester-transistor-inductance-capacitance-resistance-meter-/171927184554?hash=item2807a8a4aa:g:MdoAAOSwk0pVgETb

In the description point 6.15 they say  "For capacitors with a capacity value above 5000pF the voltage loss after a load pulse can be determined. The voltage loss give a hint for the quality factor of the capacitor."

What is this voltage loss, are they talking about DC leakage?
If it's suposed to be a leakage measurement wouldn't the meter need much more voltage than what the internal battery can provide?

If it's not leakage what is it? And how helpful can this cheap meter be for capacitor testing?
 
Whoops said:
A friend point me to some evil bay meter,
I was checking this one:
http://www.ebay.co.uk/itm/Portable-MK328-LCR-ESR-Tester-transistor-inductance-capacitance-resistance-meter-/171927184554?hash=item2807a8a4aa:g:MdoAAOSwk0pVgETb

In the description point 6.15 they say  "For capacitors with a capacity value above 5000pF the voltage loss after a load pulse can be determined. The voltage loss give a hint for the quality factor of the capacitor."

What is this voltage loss, are they talking about DC leakage?

Just a guess, possibly they charge the cap, and dump the charge into a fixed resistor for a short period of time. Then they measure voltage remaining.

A low esr cap will lose most of it's charge with a higher current into the load, and a high esr cap will have a lower discharge current leaving most of the charge in the cap, resulting in a higher voltage, as the esr is in series with the discharge current.

It could be as simple as a 1 ohm resistor load, with the load time controlled by the micro, based on the measured capacitance.

Sort of like the old-timey way, charge a cap, momentarily short it, count to three and short it again. If the second spark is almost as large as the first one, the esr sucks.

Either way, that looks like a cool gizmo, if you do get one, let us know what you think of it.

Gene

 
Gene Pink said:
Either way, that looks like a cool gizmo, if you do get one, let us know what you think of it.

Gene

For the price I think I will get it, it would be useful for something for sure.

I was just trying to understand how I would interpret the "voltage loss after a load pulse" measurment"

and also I would like to get a Capacitor Leakage Meter, so I was interested in some cheaper alternatives just to have fun around
 
for leakage just stick your DMM in series with the cap, set it on ma DC and apply voltage, read meter,

internal leakage on a lytic will cause it's voltage to drop after it is dis-connected from the DC power supply,

the faster the voltage drop, the more leakage, a healthy cap should take a long time to discharge, note that your volt  meter might have an internal impedance that will act as leakage,

you can make your own capacitor checker with a spreadsheet, volt meter and power supply,

can you count out 5 seconds +/-  10%?

1 Time Constant = the time it takes to charge a cap to 70.7% of its rated voltage with a resistor in series with the cap, 

so if you had a 100 uf cap at 25 volts with a 10 K resistor in series, it would take T=RC  seconds

T = 10,000 x .0001 = 1 second 

Capacitance has to be converted to Farads for this formula.

so in one second, if you applied 25 volts thru the 10 K resistor, your voltage across the cap should be .707 x 25 = 17.7

if you used half the resistance, the time to charge the cap to 1 time constant would be .0001 x 5000 = .5 seconds.

this is all a cap checker does, only it is limited by your battery voltage unless it is a bench model with ad. voltage.

so now you have a leakage tester and a cap checker for the grand total of...

zero dollars (considering that you have a meter and supply and a stash of resistors)

here is how you  can test a cap and get a ball park figure by doing some math, picking the right resistor and reading a voltage after say, 5 seconds>

 
Thank you so much CJ,
yes I have adjustable  bench PSU and a pretty good stash of resistors.

I will test it that way.

Any chance you might share your beautiful spreadsheet?
 
Gene Pink said:
Just a guess, possibly they charge the cap, and dump the charge into a fixed resistor for a short period of time. Then they measure voltage remaining.
======

Sort of like the old-timey way, charge a cap, momentarily short it, count to three and short it again. If the second spark is almost as large as the first one, the esr sucks.

Gene
That phenomenon is called "soakage" or DA (dielectric absorption). If you visualize the internal structure of a large capacitor there is a resistance associated with the connection to the distributed internal capacitance. This is modeled like a number of small capacitors in parallel with small resistances between them.

If you apply a short directly across the capacitor terminals, these internal Rs limit how fast that distributed internal capacitance can discharge. A momentary short will not fully discharge all the internal charge and after the removal of the short the terminal voltage will return to a fraction of the original voltage.

Logically a low ESR cap will have lower internal resistance so discharge faster when shorted.

Back in the '70s/'80s some audiophiles made a big deal about DA in DC blocking caps in audio paths but in my judgement they made much too big of a deal about this. DA is most obvious in the shorted cap example, but in the real world, a sample and hold circuit, comes closest to that. It quickly charges a capacitor to a nominal voltage through a low impedance, then buffers that cap to hold that voltage.  A cap with high DA will move around during the hold time based on previous conditions and residual charge.

The typical DC blocking audio application loads the capacitor with a constant (high) impedance. As long as the pole frequency is tuned octaves below the audio passband I don't expect any audible artifacts (if the terminal voltage doesn't change while passing signal and internal currents are tiny, DA won't express) . High DA caps are not good for use in audio band filters, and high current paths like passive loudspeaker crossovers.

JR
 
Back
Top