Why doesn't AC destroy Polarized Electrolytics?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ruckus328

Well-known member
Joined
Apr 14, 2010
Messages
906
Location
Philadelphia, USA
I've been talking with a bunch of Engineers at my plant about this.  They all keep pointing out that if using a polarized Electrolytic in the AC audio path with a split rail supply it will kill the cap, as when the AC swings negative it is reverse biasing the electrolytic, which yea, is true. 

But obviously, this is done all the time, for decades in countless designs I keep pointing out, so trying to understand why the caps survive?  It's common to see 25V polarized caps on +/-18V supplies, which means approx 32V peak to peak can potentially go through the cap.

Only explanation I have is the impedance after the cap is the only saving grace, as with say a 10K impedance the cap barely sees any of the voltage if I'm understanding correctly.  If that is true though with lower impedances the cap will be stressed and potential failure mode could ensue if value/voltage rating are too low.  Is this correct?
 
The answer is the only important factor is the voltage across the capacitor. With no signal the cap has no volts across it. When the signal goes positive, both ends of the cap swing positive taking the load voltage upwards. If the cap is sized correctly, then no matter what the frequency, the voltage across the cap will be almost zero. Similarly when the signal swings negative, again, both ends of the cap swing negative taking the load voltage with it and the actual voltage across the cap is practically zero.

Cheers

Ian
 
If most of the signal gets through, then nearly-no signal is ACROSS the cap.

Simply make sure the cap will "pass" the lowest frequency fed to it.

> it will kill the cap

No. Not in small-signal audio.

Reverse-voltage kills by excess leakage current causing heat.

What's the worst can happen? Some sub-sonic lingers at -10V. Cap acts near-short. "Infinite" current would flow.... but our circuits don't deliver infinite currents. You are likely to stall at 40mA (NE5532) at about 2V reverse. This is 0.08 Watts. Most such caps are larger than 0.08W resistors, more like 0.5 Watt resistors or larger. The cap would stabilize at about 0.08/0.5 or 1/6th the temerature rise of a 0.5W resistor dissipating its full 0.5W. Hardly-warm, NOT "hot", not hot enough to burst or even die soon.

Further, if it is a subsonic and not a DC fault, half the time it will be correct polarity and near zero heat.

Also the load will limit the fault current. Typical 10K load on 15V supply, only 0.0015mA can flow, heat will be undetectable.

> talking with a bunch of Engineers

Listen, but don't believe that everything they "know" is true.

It may, however, be instructive to TRY IT. Got a 12V-36V supply around? Can you wire a 5532 so it can be either an oscillator or follow a voltage? Put 5uFd cap and 2K (16Hz) dummy-load on the end. Worst-case, DC-bias the opamp so its output is slammed to the rail, and connect the cap in reverse voltage. Wait for the explosion. Keep looking at your watch. See the well-educated engineers move away from the bench. They may be very frustrated waiting for nothing to happen.

Or with 1KHz even 100Hz full-swing oscillations, the usual <20Hz couping cap and resistor will show "no" voltage across the cap. If the shop runs to a differential probe, you can confirm 10V audio flowing in, 9.4V getting through, 0.6V across the cap. Consult large-system schematics, caps are sized for <2Hz bass-cut (so a lot of cascade cuts don't dent 20Hz much). 0.1V across cap.
 
I recall as a very junior technician being sure that AC doesn't hurt capacitors (because I'd seen some across line cords before) and to prove it I plugged a small electrolytic cap into a 120v mains outlet, as a handful of senior technicians stood around and goaded me on...  :eek: The result was explosive.

It's all about power.. if the capacitor heats up due to terminal voltage, something will give...

JR

PS: Don't do that at home... I was a professional  ;D



 
PRR said:
It may, however, be instructive to TRY IT. Got a 12V-36V supply around?

You might also notice that the leakage is related to the rating of the cap and the size of the reverse voltage.  A constant -4v across a 50v cap will be very low leakage.  (no current, no heat, no instant death as others have pointed out.) -170v across a 120v cap is an explosion as John demo'ed in the past. 
 
> electrolytic cap into a 120v mains outlet
> -170v across a 120v cap is an explosion


That is a very different, and instructive, experiment, which should NOT be tried at home (do it where someone else will pay fire and hospital costs).

The question was about an assertion that "using a polarized Electrolytic in the AC audio path with a split rail supply it will kill the cap". Neither of your experiments are reasonable models of proper or normal "audio" design.

This would be a valid over-simplification:

tVbSx.gif


(Yes, I revised my number for max current out of 5532 and similar audio chips.)

 
Hey guys, sorry for the delayed response.  Thank you all very much for the input, gave me some nice ammo to stick in the engineer's faces.  Like I said, I knew it works just fine, as I can look at about a thousand schematics (including a bunch of my own), as well as quite a bit of gear in my studio and there's obviously no issues, but gets fustrating to have 3 senior engineer's telling me something when I know they are wrong (I just wasn't quite clear on the why's other than "well I know it works just fine").  They still seemed adamant to accept what you all had to say, in the end their closing response was... "shrug, well it might work for a while but it's still not a very good idea".  Seems it's all 1's and 0's these days with these guys, I often strike out when I present even simple analog questions to them.  Fustrating as they're very far from dumb.  For the most part they're helpful, but sometimes they leave me scratching my head more after I talk to them then I was before.
 
When you think you know everything, you'll find that you can learn nothing. Some engineers are great at what they do, but don't want to admit that they are unsure about something, so they call on their expert bullshitting skills to confuse you to the point that you have no will to continue debating... I knew an old timer audio engineer that "taught" me a bunch of stuff. Years later I slowly realized that he was talking out his ass... he was very far from dumb as well.
 
I wouldn't single out engineers alone for this affectation, I believe we are genetically wired to think we understand more than we do, so we don't go into mental melt down over all the things around us that we don't understand.. So we can see the saber tooth tiger coming. Imagine what life would be like for most people if they truly appreciated how little they know. 

I too have known many engineers or engineer types who were versant at repeating the operative buzz words with little real understanding...It's called "blinding them with science" (actually jargon, but I like Dolby's version better).  I recall being at one intra company presentation when one of the engineering directors literally ran out of BS mid sentence  ;D while trying to clarify something he said after being challenged by a member of the audience. He appealed to the other engineering directors in the audience to bail him out, and we just smiled at him... sorry about that. 8) The guy in the audience was right.

JR
 
@PRR
The current in your sketch goes in the opposite direction. ;)
@ruckus328
Take a look into the SSL schematics. You will find there bunch of bipolar elcos and DC biasing sheme for
the ordinary coupling elcos. So, be careful with your ammo.  ;)
Regards,
Milan
 
JohnRoberts said:
I wouldn't single out engineers alone for this affectation, I believe we are genetically wired to think we understand more than we do, so we don't go into mental melt down over all the things around us that we don't understand.. So we can see the saber tooth tiger coming. Imagine what life would be like for most people if they truly appreciated how little they know. 

I too have known many engineers or engineer types who were versant at repeating the operative buzz words with little real understanding...It's called "blinding them with science" (actually jargon, but I like Dolby's version better).  I recall being at one intra company presentation when one of the engineering directors literally ran out of BS mid sentence  ;D while trying to clarify something he said after being challenged by a member of the audience. He appealed to the other engineering directors in the audience to bail him out, and we just smiled at him... sorry about that. 8) The guy in the audience was right.

JR

Agreed, I said engineers but what I should have said was professionals.  I'm a highly trained automotive tech and I know there have been times I've BSed my way through a discussion with management knowing it was over their heads. True, it's in our nature. I bet if you separated the weakest one from the herd, you might get him / her to admit your validity.
 

Latest posts

Back
Top