40 years lytic bipolar that drifted up...

Hi,

Not having soundcard and software yet for parts measurement to impulse, I checked old bipolar lytics from a loudspeaker with a cheap capacitance tool.

My surprise : all have between 10 % to 15% than their marked value, same for 5% marked. And the speaker was never refurbished and ages a pretty 40 :eek: ... weekendly singing !!!! the pcb and caps are labelled from the speaker brand, factory soldering quality so never recaped.

Is it the dry electrolyt that gives too much frequency drifts in a lytic bipolar giving a bad result from my 20 usd capacitor tester ? Is it because it loads the non polar cap with a DC at a given frequency ? Or do I have the correct value (no capacitance drifts in 40 years, just crap caps 10 to 25% precision instead 5% marked for some ?

ESR became so bad and had affected the capacitance measurement ?

WTF ?

someone suggested there was no magic in it, however I still not understand why ?
 
A lytic with 5% tolerance is a rarity, even for a NP type, and I am not surprised they drifted somewhat more.

They could have developed a leakage current fooling your cheap tester, or the oxide might have partially dissolved.

You could try to apply a 50Hz signal of ~half their rated voltage for a few minutes and remeasure them to see if they have come back to normal, but the simplest solution is probably to replace them with modern, film type caps.
This is now possible unless the value is really very large
 
Thank you Elvee,


I was not aware it was possible for a npe to see its capacitance growing up. I simply believed they can just loose their capacitance but not drift the opposit way.
Your solution is very interressant to retest them as the current here is 50 Hz and I have several traffos to match with the secondary the Vac of the caps :)


Yes I will replace them but I want more precesion in the measurement because of the 3% to 5% of many film caps... just to try to pick up the best uF values !
 
Last edited:
Thanks for the tips guys...I,m very curious about that old npe will try a technic Lozjek member gave, 50 hz at half the rated voltage to wake up a little an old npe to see if the capacitance goes down closer from its marking...I imagine a secondary taking the peak to peak rating, circa 1,4 x the secondary could work. Just curious...
 
I was not aware it was possible for a npe to see its capacitance growing up. I simply believed they can just loose their capacitance but not drift the opposit way.
And it´s probably not going UP.
Just the displayed value is.

Problem is, cheap meters measure capacitance by placing it in serie with a low value resistor (think 100 or 10 ohm) , apply some AC voltage (often a cheap easy 50Hz or 400Hz squarewave) and measure millivolts across the resistor.

The higher the capacitance the higher the voltage, which is displayed as "X" uF.
Crude but it works, sort of, and it´s **cheeeappp**

Problem is, a lossy capacitor will also let more current through (it was suggested above to add a parallel resistor to it) so it will *display* a higher value.
 
You could try to apply a 50Hz signal of ~half their rated voltage for a few minutes and remeasure them to see if they have come back to normal

Aren't you afraid that may cause them to explode if the oxide is in bad shape? I'd rather slowly charge them to their rated DC voltage through a large resistor, 100 kohm or so, leave them biased for an hour and then do the same with the opposite polarity.
 
Aren't you afraid that may cause them to explode if the oxide is in bad shape?
Not really: 40 years old means made in ~'80, and normally these caps wake up gracefully.
I still have bags full of NOS Ecaps made in the early seventies, and when I fire them up, well I simply do it, without any particular precaution, sometimes at their ~full working voltage (OK, they are mostly Siemens and similar brands).

I am much more circumspect for tube gear from the sixties and before: the technology wasn't mature yet, and a brutal cold start after 50y of inactivity can have disastrous effects
I'd rather slowly charge them to their rated DC voltage through a large resistor, 100 kohm or so, leave them biased for an hour and then do the same with the opposite polarity.
That is of course the safest option, without possible contradiction.

I don't think it is generally required, especially for NP caps which are built more robustly than polar types, but it will do no harm.

An even safer and simpler option is to replace them with mylar caps: they now exist in large enough values and are vastly superior in all respects
 
:) ...I remember some cool ones... Vishay, or Mexico...

Yes the wize way is just to swap with new lytics...their esr is better today than the 80s...Mundorf, Jantzen have some with0.025 esr caps... btw surely the same manufacterer...ok mkt has more 0.00x esr...i like the k73_16 ...butfew values choices.
 
Btw, just being curious on the hisyoric equipment side for the culture, do some know who were the manufacter or inside caps comodities suppliers of 80s english speakers...Spendor, kef, tannoy, aberth, monitor audio, b&w... english caps makers and german comodities providers?

I apologize myself being off topic here...uh... I never understood why some speakers companies were buying caps branded with their own name as the diy was not so seen? ... was it for marketing or to hide the capacitance values with generics after their own sorting before filter assembly ? I read no one was able to recapture the BBC monitors...
 
Last edited:
And it´s probably not going UP.
Just the displayed value is.

Problem is, cheap meters measure capacitance by placing it in serie with a low value resistor (think 100 or 10 ohm) , apply some AC voltage (often a cheap easy 50Hz or 400Hz squarewave) and measure millivolts across the resistor.

The higher the capacitance the higher the voltage, which is displayed as "X" uF.
Crude but it works, sort of, and it´s **cheeeappp**

Problem is, a lossy capacitor will also let more current through (it was suggested above to add a parallel resistor to it) so it will *display* a higher value.
Thanks, this is good to know.
I am using a cheap DMM. The cap measurement is up to100μF only.
So no good to check most filter and decoupling capacitors.
And from what is explained here, it is misleading
prone to show good caps that are actually badly leaking.
When using a cheap DMM, is the ohm measurement (which is DC I presume) a valid way to measure leak of an electrolytic cap ?
Do you get a valid measurement of the IR (insulation resistance), using the DMM as
measuring a resistor ? Here, DA dielectric absorption is a possible trouble maker.
 
Not sure I understand here as we talk about non polar electrolytic! An ESR is always measured at a given frequency ? Ohm measurement on an NPE with a cheap will really display a result as the cap is blocking DC ?
Or will you see the NPE ESR through a // precision resistor then calculating the difference between the displayed value and the standalone resitor value?? What is the best precision of a resistor ? 0.001 %, so enough for a lytic ?

Anyway I would like wake uping the np for my understanding.

The serie npe suffering less than the shunted ones in the impedance correction, a gently wake up Elvee suggested maybe will give something near the initial capacitance value permitting a safer new cap purchase (as esr is definitly wasted after 40 years) ?

I have for instance 100 uF npe caps 50 vdc rated in serie as HPF in the passband filter for the mid, measuring 110 uF after 40 years of week-end listening sessions I assume. These ones having no tolerance marking on their bully;;; so it could be initialy a sorted value by the speaker manufacterer between -10% to +20, or + 50%.

Is it o to proceed as this please ? to choose the secondary of a power transformer to feed the npe with 50/60 Hz I assume I have to take the secondary value x circa 1.4 for having the peak voltage ? For illustration 16V secondary is giving almost 23Vac RMS, so near the Vac max of the npe which would be around 35 Vac. How should I close the circuit please ? I take 2 diodes to have a positive drift on a secondary wire, then a resitor in serie, then the serie cap and I shunt towards a further resistor ? (should I have to putt two diodes also on the second wire to have the negative drift took fo gnd reference ? How would I choose the wattage and ohmic values the two resitors ?

Sorry, very noob I am :eek: ... non polar cap is strange to me vs polarized

Edit : I also imagine such gadget could be good to burn in tool for any non polar caps, be it film or npe ?
 
Last edited:
A transformer giving you 23v ac rms is a good start.
To close the circuit, I would simply use a resistor to limit the current in the case of anything goes wrong, worst case being a short.
A 1000 ohm would limit the current at 0.23A drawing 1/2 Watt.
For instance, to test a 450V 16 μF AC cap, I used the 240V 50Hz mains power with a light bulb as a limiting resistor.
 
I made à mistake in post #17, on the resistor value.
Read:
A transformer giving you 23v ac rms is a good start.
To close the circuit, I would simply use a resistor to limit the current in the case of anything goes wrong, worst case being a short.
A 100 ohm ( Not 1000 ) would limit the current at 0.23A

The resistor is for safety, one could put no resistor at all.
For instance with a 16uF on 50 Hz, the reactance is 200 ohm ( from Z = 1/(16 E-6 x 2 x Pi x 50 )
So, with 23v ac rms the current is 0.13A rms, the cap should be able to handle it, the power in the cap is near 0 ( it is not 1/4 Watt as it would with a 200 ohm resistor ).
There is a non 0 power because of ESR not 0.
Assuming ESR = 0.025, the power is 0.025 x 0.13 x 0.13....some very small value.
 
Last edited:

PRR

Member
Joined 2003
Paid Member
...An ESR is always measured at a given frequency ?....

No.

Conceptually: measure Impedance (E/I) at frequency from DC to infinity. For a while the impedance just falls. Then it may level-out, tending toward a constant resistance, "ESR". In any case the impedance will then rise, because everything is an inductance.

In real parts the impedance *may* go to "zero" (less than true ESR) because of C+L resonance.

It is possible (not real easy) to measure ESR in the falling-impedance zone by careful measurement of phase angle.

For nearly any modern capacitor, the impedance at 20kHz is nearly pure C, ESR at 20kHz is much less than capacitive reactance. Look in datasheets. This isn't the 1950s any more.