Tell me About Caps Across AV Mains

[Roy Batty]

So long, and thanks for all the fish.

[tggzzz]

I have a couple of noob questions about capacitors used as filters on AC mains.

I just got my old HP 3300A function generator running. It had two issues. The one which is relevant to my questions was a short in a filtering capacitor. There was a dual cap on the AC lines on the way in. One leg went to ground, and the others went to neutral and hot. This dual cap was right at the power receptacle, before the transformer.

One side of the cap had shorted, so I replaced it with two new caps.

Since it is across the mains, I do hope the replacements had the same X/Y rating. The specification for those ratings defines what should not happen when they fail.

I've had them fail, and the results weren't pretty - but at least they didn't start a fire.

[Tandy]

Yes the X & Y Caps are there to reduce noise injected into the mains circuit.

[Vtile]

Other use for "AC mains capacitors" are capacitive reactance creation to neutralise the effect of the inductive reactance of the motor or tranformer in other words power factor correction capacitors for inductive loads like motors, transformers etc.

[Gyro]

I'm not sure about the infancy of digitalia, he must have been around in the days of DC mains!   

Usually there is a large capacity, maybe .33 Mf or larger if space allows, and also a .0001 ceramic disk across it, from each wire to ground. The .33 should be NP (non polarized) but I have seen folks using electrolytics or tantalum both of which are polarized (the latter known to catch fire when they fail). If there is a coil in the can, it is inline and has a ferrite core, that would buck RF as well, I think it is only caps. Make your own filter, there is no reall critical issue.

Please don't follow that advice, I shudder to think of the effect of putting those sort of value electrolytics or Tantalums from line(s) to ground.... and as for there being no real critical issues!   

As you have seen from previous replies, the type and rating of X and Y caps is absolutely critical to safety, both fire and shock risk. He is right that in-line inductance improves filtering effect, common mode and single chokes are common, but again need to have suitable insulation and flamability rating.

If you have the space in the HP3300A then retrofitting an integrated IEC mains filter [Edit: as a replacement for the existing mains inlet connector] might be your best option as it includes choke, X and Y caps in a safe encapsulated form.

[Seekonk]

It used to be that the power mains was a pretty low impedance with all those incandescent lamps.  Now it is more of an unterminated transmission line, the noise just bounces around. And now there are even more devices throwing trash on the lines.  I found many of our clients resolved noise problems using a RC filter like 10 ohms and .47uF or higher across the line.  Certainly not a one solution for all, but keep this in mind.

[Gyro]

His advice was not very clearly written, for instance

maybe .33 Mf or larger if space allows, and also a .0001 ceramic disk across it, from each wire to ground

. This implies paralleled .330nF and 100pF caps from each line to ground. What I hope he actually meant was 330nF across the line and 100pF capacitors from each line to ground (I think he stuck in an extra 0 in there, 1nF -2n2 would be more expected).

Yes I did note that he indicated NP Electrolytics (there are no non polar Tantalums, or indeed Tantalums rated to mains voltage at DC ). The only mains application I know of for NP electrolytics is as 'Motor Start' capacitors for some induction motors, these are intended for very short term operation (until the centrifugal switch opens) and never for continuous operation. 'Motor Run' capacitors for induction motors, which are continuously in circuit are invariably self-healing MKP or MKT Film types.

As I said, type and rating of X and Y capacitors is critical for your safety and others - You will never find an X rated NP Electrolytic (they are not self-healing for a start!). You need to use properly approved parts (you will find the rating and VDE/SA/UL stamps on the package), you simply can't use unspecified capacitors as he suggests. An across the line capacitor of 330-470nF X capacitor would not be unreasonable, or as Seekonk suggests an safely constructed RC Snubber (X rated capacitor + fusible resistor).

To sumarize I was resopnding to the advice that component choice / filter construction was non-critical and the suggestion that NP Electrolytics (let alone the Tantalums anecdote), would ever be seen in a mains filter application. Please be safe.

[Seekonk]

There are large elctrolytics used in one type of filter.  A large fast diode feeds one set of electrolytic capacitors and a diode in reverse feeds another set of electrolytics.  The capacitors charge up to the peak voltage of both polarities. A resistor bleeds off a small amount of charge.  Any power spike above the voltage of the capacitors gets shunted to them. This always tracks the maximum voltage of the power line unlike a MOV which is fixed.  These also contain the normal filter components.

[FlyingHacker]

I am certainly not suggesting doing this for connecting mains to ground, but perhaps he meant using a pair of polarized caps (tantalum/electrolytic) back to back to make them non-polarized.

[tggzzz]

I am certainly not suggesting doing this for connecting mains to ground, but perhaps he meant using a pair of polarized caps (tantalum/electrolytic) back to back to make them non-polarized.

Think about the chemistry that distinguishes polarised capacitors, and think what would happen.

[FlyingHacker]

Not sure what you are referring to, tggzzz.

Again, I am not suggesting you do this, but I mean where you use two polarized caps with negative leads connected to make one non-polarized cap (where you then use the two positive leads as *the* leads). Of course they have to be the right voltage rating, and there are lots of reasons not to use an electrolytic for this purpose (ESR, fails shorts,frequency response, etc.).

[tggzzz]

What causes the dielectric to form, and are the conditions necessary in the configuration you propose?

[Vtile]

I have understood that the whole polarisation thing on electrolytics is because of mechanical construction that makes one of the leggs less isolated (from the container/outer shell) and that is the reason of the so called polarisation.
It seems I have been badly informed. I still like to note that physical structure can be also a reason for some polarity restrictions.

[Seekonk]

Time for the electrolidifiles to start a new thread.

[FlyingHacker]

What causes the dielectric to form, and are the conditions necessary in the configuration you propose?

You talking about the lack of a constant DC voltage across the capacitors in that example?

[tggzzz]

Sigh. Google is everybody's friend.

The first search result for "electrolytic capacitor" leads to  https://en.wikipedia.org/wiki/Electrolytic_capacitor . There are better descriptions around, but these two paragraphs will suffice to indicate the issues.

Standard electrolytic capacitors are polarized components due to their asymmetrical construction, and may only be operated with a higher voltage (ie, more positive) on the anode than on the cathode at all times. Voltages with reverse polarity, or voltage or ripple current higher than specified (as little as 1 or 1.5 volts may suffice), can destroy the dielectric and thus the capacitor. The destruction of electrolytic capacitors can have catastrophic consequences (explosion, fire).
...
After forming a dielectric oxide on the rough anode structure, a counter electrode has to match the rough insulating oxide surface. This is accomplished by the electrolyte, which acts as the cathode electrode of an electrolytic capacitor. There are many different electrolytes in use. Generally they are distinguished into two species, “non-solid” and “solid” electrolytes. As a liquid medium which has ion conductivity caused by moving ions, non-solid electrolytes can easily fit the rough structures. Solid electrolytes which have electron conductivity can fit the rough structures with the help of special chemical processes like pyrolysis for manganese dioxide or polymerization for conducting polymers.

You can do further research on your own.

[FlyingHacker]

So, what I said... not a relatively constant voltage... I.e. too much ripple. Perhaps I should have phrased that better.

I was just curious what the specific point you were trying to make was. We do have access to the web here. 

[tggzzz]

So, what I said... not a relatively constant voltage... I.e. too much ripple. Perhaps I should have phrased that better.

I was just curious what the specific point you were trying to make was. We do have access to the web here. 

For successful operation within device limits, the voltage does not have to be "relatively constant", and the percentage ripple is only of secondary importance.

[FlyingHacker]

I give up. So what exactly is the point you are trying to make with all these cryptic questions?

Was there something you wanted to communicate to the readers of the thread?

[tggzzz]

Two points...
Operate within the limits of the data sheet.
Think.

... and it helps if you understand the chemistry and physics.

[FlyingHacker]

All true... which was certainly why I repeatedly said not to do what I described.

I was just throwing out ideas as to why someone would have suggested using electrolytics or tantalum electrolytics in the first place. It was merely an guess of how someone might think they would implement a polarized capacitor on an unbiased AC circuit.

[FlyingHacker]

All about circuits? The website? Only been there via google.

[KE5FX]

It definitely doesn't make sense to take chances with line filters.  Even the genuine Corcom filters used by HP, Tektronix, and other top-tier manufacturers can get a bit explodey after 20 or more years of service. 

You could make a non-polarized electrolytic by using two back-to-back capacitors with diodes to short out the opposite-polarity voltage seen by each.  Need to check the data sheet to make sure that the 0.6-0.7 volt diode drop is lower than the polarizing voltage.  Said polarizing voltage will most likely not be specified on the data sheet, which should be taken as a strong hint to use a capacitor that's rated for line filter service instead of hacking something together with electrolytics.

[FlyingHacker]

I just want to be clear that I did not, nor ever would, suggest that anyone actually make a line filter with electrolytics. That would be really stupid. I would use a Y rated cap.

 It was merely a thought experiment based on a earlier post that seemed to indicate that using electrolytics had been suggested.  I am sorry that it led to such tangents. 

[KE5FX]

Isn't allaboutcircuits just a BBS-style Usenet gateway?  If so, lack of social skills is putting it charitably...