Electrolytic cap voltage ratings

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Ahhh but zeners dont have that ghostly etherial glow to signal out to you ,
of course your right all kinds of down sides to gas discharge tubes , they can go noisey or not maintain the intended voltage when they get worn out . Maybe a reclaimed neon from the disposable flash camera could serve as indicator the zeners are starting to conduct.

As far as the veer against disposable cameras goes , Im quite happy to take a bag of 20 fuji units for dismantling any day I happen to call by the store , thats 20 near new caps of the best quality , they even leave plenty of leg room so you can simply clip the component from the pcb and put into the tray marked 82uf 300v . Im swapping out all the garbage caps in my Chinese mics supplies for these asap.

Oh yes by the way , be warey the capacitor can remain charged in the flash unit for a longtime , it'll give you a nice little zap if your not careful. If your unlucky enough to make contact with positive and negative terminals with left and right hand it could very well stop your heart ,my advice is crow bar the cap with an insulated screwdriver tip before handling the pcb . the cap appears to have a memory effect where even after discharge through the zenon lamp voltage reappears .
all the other components of the camera go to plastic recycling even though there is a few tiny metal springs you could also make a collection of if you wanted ,
 
Last edited:
I dug out a nice multitap 20w core transformer for tomorrows bench tests ,
Im going to re-house everything on a bamboo chopping board I got for a few euros at the thrift store,
I'll mock up the circuit in PSUD for clarity and add annotations later today .
 
I have seen a few electrolytics run to their ragged edge, like a 15v circuit with 16v electros. But for the sake of simplicity, I usually stock 50v units, and where there is P48 in play, 63v. Makes sense to me. But anyone with an opinion of using 50v caps in a 10-15v application? I’ve read in articles long past that the actual value is somewhat dependent on voltage? Any potential issues? Thanks.
 
be interesting to take a closer look at the forming process caps go through ,maybe experiment with frequencies other than 50/60hz and waveforms besides sine and see how leakage current is affected . And maybe making the unit capable of 110% of the max cap voltage isnt such a bad idea, after all it is for bench purposes ,

You could do that if it's of sufficient interest. Or you just derate enough to give a long expected lifetime.
I know it's for 'bench' use but it sort of depends whether your priority is to get something that works and is reliable on your bench. Or investigate electrolytic capacitor design and performance characteristics.
 
I have seen a few electrolytics run to their ragged edge, like a 15v circuit with 16v electros. But for the sake of simplicity, I usually stock 50v units, and where there is P48 in play, 63v. Makes sense to me. But anyone with an opinion of using 50v caps in a 10-15v application? I’ve read in articles long past that the actual value is somewhat dependent on voltage? Any potential issues? Thanks.

Having too high a rated voltage isn't a 'thing'.
Actual value varying with voltage ? You mean varying value of capacitance ? ie Voltage Coefficient
That is a 'thing' with ceramic caps - gets worse as you go to dielectrics that have increasing capacitance per volume eg X7R, Y5V.
I don't know if it applies to electrolytics. I can't recall it ever coming up as a design issue but never say never...
But really electrolytics are generally loosely spec'd eg +/-20% and value decreases with usage then any voltage coefficient is likely just "noise" ?
 
Having too high a rated voltage isn't a 'thing'.
Actual value varying with voltage ? You mean varying value of capacitance ? ie Voltage Coefficient
That is a 'thing' with ceramic caps - gets worse as you go to dielectrics that have increasing capacitance per volume eg X7R, Y5V.
I don't know if it applies to electrolytics. I can't recall it ever coming up as a design issue but never say never...
But really electrolytics are generally loosely spec'd eg +/-20% and value decreases with usage then any voltage coefficient is likely just "noise" ?
Voltage coefficient, namely capacitance changing with terminal voltage requires significant changing terminal voltage. This is mainly an issue when using lousy dielectric caps in audio frequency filters where the terminal voltage is changing with the audio signal (this can be measured as THD in circuit).

Electrolytic caps are generally used in DC blocking applications so terminal voltage is relatively stable even when passing audio. The exception would be electrolytic caps used in passive loudspeaker crossover filters where they see the changing terminal voltage with audio. If you can afford the size/cost, film caps are better for passive loudspeaker crossovers. Active speaker crossovers completely eliminate that issue by scaling down the current/voltage of the filters so quality film caps can be used, or DSP eliminates the need for even those small film caps.

JR
 
Yes. I don't know how much it applies (or not) to electrolytics when there is a varying voltage across the cap.
But, as you say, in ac coupling operation there should be minimal voltage across the cap in any case and that minimises any distortion issue.
+1 on film caps for crossovers (and anywhere else you have the budget/space for them in place of electrolytics).
 
Yes. I don't know how much it applies (or not) to electrolytics when there is a varying voltage across the cap.
But, as you say, in ac coupling operation there should be minimal voltage across the cap in any case and that minimises any distortion issue.
+1 on film caps for crossovers (and anywhere else you have the budget/space for them in place of electrolytics).
I do not recall seeing voltage coefficient specified on typical electrolytic capacitor data sheets, suggesting to me that most are not intended for use that way...

There may be some specialized series designed for use inside loud speakers (they are cheaper), but film is generally better.

JR
 
I do not recall seeing voltage coefficient specified on typical electrolytic capacitor data sheets, suggesting to me that most are not intended for use that way...

There may be some specialized series designed for use inside loud speakers (they are cheaper), but film is generally better.

JR

Nor me. Ever. In either audio or instrumentation applications. Mind you I once insisted to someone that he meant "Tantalum Capacitors" when he was mentioning "Tantalum Resistors". I was wrong on that one 🙄
 
Last edited:
Tantalum Resistors

Hmmm...that was a new one for me.
Fortunately Google has sorted the available information with most relevant at top:
"These exotic resistors from Japan, sound less dry than the metal films and they do not emphasise any particular frequency range. "
:unsure:
 
Should be OK that far down the filter, ripple v is almost non existent,

When you measure DC with a meter, it is averaging any ripple v, so if you put a scope on the first cap in a filter, you might see peaks that go above what your DC meter is reading.

In other words, the cap rating is being hit momentarily with that saw tooth wave.

I have seen a filter cap being used at over voltage, possibly in a fender. Music master Bass which is a CBS creation.

Go ahead and use the caps, the preceding resistors will drop the v as soon as the caps start to leak DC,
Thus creating an ohms law protection circuit,

Ah yes, the old 300 volt cap getting 295 plus or minus line voltage drift trick, at the worst possible place ripple wise,so that cap is probably seeing peaks of maybe 330 volts,
1659680463791.png
 
Last edited:
Hmmm...that was a new one for me.
Fortunately Google has sorted the available information with most relevant at top:
"These exotic resistors from Japan, sound less dry than the metal films and they do not emphasise any particular frequency range. "
:unsure:

yeah - not to be cryptic on this - related to Audionote - Hi-Fi Valve kit and components company. A technical clerk type guy who worked there used to go to a pub that a couple of my friends ran. Her was 'complaining' about the tedium of going through the BOMs and changing to tantalum resistors.
I just had a look and see that they are now also looking at Niobium resistors:

"Niobium has never been used in a resistor as far as we know. We looked at its properties and felt it was worth testing, and discovered that it is finer and more even sounding than Tantalum."

Yeah - okay 🙄
tbf a friend had/has one of their amplifiers and it did sound good in his system.
Obvs would be better now with the extra finesse of Niobium balanced against the err, whatever, of Tantalum. Or not...
 
I presume to make a cap the insulating layer has be formed , the right chemicals are subject to a voltage and the barrier layer is built up to the required strenght , I wonder what happens over the longer term to a capacitor which never gets anywhere near its maximum voltage rating in usage. Lets say we put in a higher voltage cap in a circuit than needed ,just on the off chance of user error , so this cap lives its entire life at say 1/10 its maximum voltage , then one day clever clogs comes along and and hooks up phantom power the wrong way , how effective is our 63 v cap after 20 years service at 6 volts ?

CJ's experience seems to mirror mine regarding tube amps , where the first filter cap see's the majority of the wear ,as we said the voltage at the screen is often not far below the plate supply , so its the strong ripple current causes the first cap to age.

Ive been experimenting with different transformers as step up , smaller cores of a few VA work best as they seem to present a better load to the Arb generator , looking into the drive side of the transformer I'm using with the LCR meter I read around 600ohms @100hz , (thats with bridge and filtering components in place the far side ) , dc resistance of the coil is around 50 ohms .In larger transformers where the LV winding measures a few ohms it loads the output of the generator down and you get next to nothing out at the far end .
The simulation below works out accurate only I'm feeding in a 100hz square wave from both channels on the arb gen set 180 degrees out of phase . sine wave ouput gives much lower voltages .


I made up a probe cable with a series cap and 10M resistor to ground after , used it to link up with a high Z audio usb input.
If I cranked the drive signal all the way up into the hundreads or khz region I could hear some breakthrough , as you'd imagine with square waves it was slightly more gritty sounding than sine ,subjective anoyance factor increasing with frequency .
The audio interface quotes 1M input with -101db noise at minimum gain , patching in the dc blocked output from the PSU running at 100hz had no impact whatsoever on either measured (REW) or percieved noise (headphones) although the resonances were visible on FFT .

Interestingly, I got a better reduction in noise from the interface itself when I made contact from the metal casework to a large section of metal shelves isolated from ground in the wooden workshed, only the arb gen runs off mains power ,doesnt share a common ground / has the galvanic isolation of the transformer , with computer and USB interface running on internal battery . My interface noise levels are still way to high to see whats really going on .

1659748240573.jpeg
1659748292556.jpeg
 

Attachments

  • PSUII..JPG
    PSUII..JPG
    36.7 KB · Views: 0
  • PSUII voltages.JPG
    PSUII voltages.JPG
    14.5 KB · Views: 0
Last edited:
If you restore vintage amps/equipment then reforming e-caps is a common task, especially if the e-cap is from a good manufacturer and not toooo old. I reform to the rated e-cap voltage (eg. 450V for a 450V rated e-cap from the 1960's) and if the leakage current subsides to a reasonable level, and the cap shows ok capacitance and ESR and visual integrity then I'm happy to keep it and not replace it. Sure, it's easy to observe that some e-caps have lived their entire life at a relatively low B+ level, and it can take a while for the oxide interface to reform as that low B+ level is transitioned through on the way up to rated voltage.

There can be a few ways for a soundcard measurement system to show up poor noise floor. Certainly having only one connection to the mains AC is a great start. Another aspect can be how the soundcard output passes in to the DUT, and how the signal from the DUT returns to the soundcard input. If the soundcard's output and input cables connect to different parts of the DUT grounding system, then noise on the DUT ground system can couple in to the REW measurement. It's also worthwhile confirming that your measurement setup shows a suitably low noise floor when all connected up, but the mains AC lead is (a) disconnected, and (b) connected but AC power at the DUT is turned off.
 
Although the caps I used in this PSU arent very old there was definately a settling in period , output voltage slowly drifting upwards.
After around a week I was able to set the output of the gen to give 130.00 vdc at the psu output , this morning its still right on the money to two decimal places . I have the gen output settings pre programmed as default at start up, so its easy to cut the power to the gen to check what contribution its making to noise , when you switch it back on it takes a few minutes for it to top back up to a stable voltage again .

The gen has a programable ramp up/down feature which might be useful , you can set the period for 999 minutes maximum and specify upper and lower output voltages .

The noise of my interface at -100db isnt good enough, Im only seeing the peaks of what the generator is adding to the noise floor , I might try an active DI box which claims -110db noise and see if the results are any better.



1659796781174.png
 
Are you using REW's capability for FFT and averaging settings for general noise floor assessment, and for locking to the soundcard generated frequency and coherent averaging to better discern harmonics of the test signal?
 

Latest posts

Back
Top