Film capacitor as power-supply filter

The rest of the story...

For those who tuned in late, I'm replacing the power supply filters in a He athkit IT-11 capacitor checker. In the original circuit they are two 40-uF 350-V in series (to handle 600 V). I ordered a 20-uF 1-kV film capacitor to replace them.

Well, there just isn't room for that film capacitor under this chassis! So I had to resort to two 47-uF 450-V in series, like Heathkit's original.

There was debate in this thread about whether to add balancing resistors, a nd I didn't. John Larkin said "they take care of themselves." Indeed they do, sharing the voltage to within a few percent of equal.

I wonder if Mr. Larkin could explain further. Is there some effect that ma kes them equalize if they have different leakages to start with? I would e xpect the opposite, that the leakier one would carry a lower voltage and wo uld form less, causing the difference in leakage to increase -- but I might be wrong.

I am going to look back at the bleeder network and see if there's a good wa y to split it in half.

Reply to
mc
Loading thread data ...

Heathkit IT-11 capacitor checker. In the original circuit they are two 40- uF 350-V in series (to handle 600 V). I ordered a 20-uF 1-kV film capacito r to replace them.

So I had to resort to two 47-uF 450-V in series, like Heathkit's original.

and I didn't. John Larkin said "they take care of themselves." Indeed th ey do, sharing the voltage to within a few percent of equal.

makes them equalize if they have different leakages to start with? I would expect the opposite, that the leakier one would carry a lower voltage and would form less, causing the difference in leakage to increase -- but I mig ht be wrong.

Leakage currents have a tendency to increase exponentially with applied vol tage, so you often don't need much voltage difference to compensate for sig nificant differences in initial leakage current.

Some manufacturer's application note that I read back in 1970 was a full bo ttle on the subject.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

---------------------------

** The mid point voltage will be offset such that the same leakage current flows in both caps. The total voltage of the caps should exceed the supply by 25% or more to allow this.

Adding bleeders increases the leakage through the leakier example.

Both caps will get more leaky when hot, IME more so for the leakier one.

I've been doing is for decades with no issues at all.

Usually with 2 x 350V or 2 x 400V caps in series on a 500V supply with up to 600V surges at switch on.

..... Phil

Reply to
Phil Allison

because they are in series the capacitors share the voltage in such a way that the leakage current is the same in each. thus the leakier (more prone to leaking) has a lower voltage, typically leakage increases with voltage, so the capacitors self-balance.

--
  When I tried casting out nines I made a hash of it.
Reply to
Jasen Betts

Thanks. If the increase is exponential (more than linear), that explains why they self-regulate.

Reply to
mc

The leakage current must be the same in series caps, no matter what the voltage.

If they are in series, the leakier one has less voltage drop so dissipates less power.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
jlarkin

The one with the highest I:V leakage curve gets a smaller share of the total voltage, and the other gets more voltage, automatic equilibrium. They don't self-equalize on voltage; that's the point. They must have the same leakage *current*

If the caps are similar, and leakage in each one increases radically with voltage, they will settle down with nearly equal voltage drops. If they don't, that's OK too. As Phil points out, forcing the series cap voltages to be equal breaks the equilibrium and increases cap leakage.

Forming is an electrochemical reaction that is driven by current, not voltage. And the current is necessarily equal in all caps of a series string.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
jlarkin

-----------------------------------------

** Really ?

** The actual leakage current with HV electros is normally in the micro-amp range so dissipation causes no discernible temp rise. Ambient temp and rises in same entirely swamp it.

Where series bleeder resistors are used ( eg 100,000 Fender tube amplifiers ) they pass about 1mA which is enough to reduce and nearly eliminate mid point voltage offset.

The main benefit however is in bleeding off charge so the caps do not present a hazard if the AC supply is removed while there is no load on the PSU.

There are a couple of ways this can happen.

See typical schem:

formatting link

..... Phil

Reply to
Phil Allison

Ole tube gear seldom had fans. It just cooked itself.

BLDC fans are great, small and quiet and cheap. The fans back when were big shaded-pole motor things.

Reply to
John Larkin

-------------------

** For an unexplained reason JL introduces a new topic here.

Keeping the wacky guy on-topic is a battle.

** Not one bit true - the cooking itself bit I mean.

** But could be very quiet.

Many tube scopes had fans fitted to the back panel and with a series dropping resistor could be made very quiet indeed.

Some famous tube instrument amps had AC fans as standard, but there was a big nasty waiting to catch any designer who was stupid enough.

Ampeg released a new version of their famous SVT, a 300W tube bass amp in the late 1990s. It had an AC fan that ran continuously.

The designer made two bad errors:

  1. He made the air flow go under the chassis, over the main PCB with all the output tube sockets and up though holes to cool each 6550 tube's glass envelope. A similar method is used in high power RF transmitters.

  1. He also fitted octal sockets that had 2 inch long pins - so the PCB could be remain deep under the chassis with the socket's flange secured to the metal deck as usual.

The result was that dust and fluff collected around all the long pins until a massive 700VDC arc-over occurred between an adjacent pair ( plate and heater ) causing extensive damage to the PCB, tubes and many other parts.

Took hours to clean up and fix.

Later production examples had plastic sleeving fitted to the long pins.

The PCB still became so badly contaminated it was no cure.

.... Phil

Reply to
Phil Allison

This is called a "discussion group." And isn't heat related to electrolytic cap life?

The ones in the Tek 500-series scopes were noisy.

I wonder why Tek used fans.

It didn't help thermally that some audio gear was literally upholstered.

Reply to
John Larkin

a Heathkit IT-11 capacitor checker. In the original circuit they are two 4

0-uF 350-V in series (to handle 600 V). I ordered a 20-uF 1-kV film capaci tor to replace them.

So I had to resort to two 47-uF 450-V in series, like Heathkit's original .

s, and I didn't. John Larkin said "they take care of themselves." Indeed they do, sharing the voltage to within a few percent of equal.

t makes them equalize if they have different leakages to start with? I wou ld expect the opposite, that the leakier one would carry a lower voltage an d would form less, causing the difference in leakage to increase -- but I m ight be wrong.

oltage, so you often don't need much voltage difference to compensate for s ignificant differences in initial leakage current.

bottle on the subject.

They do no such thing. For aluminum ecaps the leakage is k x sqrt(CV), maki ng the ratio of voltages proportional to C1/C2. That can be a pretty big di sparity. My guess is the manufacturer makes some attempt at matching C befo re installation. But what does your incredibly incapable and super ignorant f*ck nothing of a pedophiles tell you about it?

Reply to
bloggs.fredbloggs.fred

I think your equation is generic on leakage at rated voltage, across cap ratings. It implies a declining slope for a particular capacitor, which is not how they behave. Applied to single caps, it would make a series string unstable.

It's surprising how rare I/V curves are for electrolytics, practically not to be found.

formatting link

Fig 7 looks like semi-serious data, but ends at rated voltage. In my experience, leakage increases radically above rated voltage.

But even that curve implies series cap string voltage self-equalization.

Reply to
John Larkin

------------------------------------------------

** Fraid they do.

** Fraid that is not the formula for ACTUAL electro cap leakage at all.

One sees it sometimes on spec sheets describing the MAX leakage to be expected where "C" and "V" are the makers rated vales for a particular cap.

This link give a bit more detail to an non-simple situation.

** Absolute rubbish.

** Kinda undoes any possibility of a sane post - don't it?

.... Phil

Reply to
Phil Allison

------------------------

** Oops - see same link JL posted.

.... Phil

Reply to
Phil Allison

I might just measure a couple of electrolytic caps. I think the leakage jumps radically a bit past the reform voltage.

Here's some data on some other caps.

formatting link

formatting link

formatting link

They all curve up! The polymer leakage data is entangled with its time behavior, sort of a reform or dielectric absorption thing.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Here's possibly the only curve like this ever posted online:

formatting link

The data is crude, because there is reforming and dielectric absorption going on, and I don't have a month to play with this. But the curve is clearly radically upward. Not a zener, more like an MOV.

The current is very noisy above maybe 70 volts. I see what looks like brief high current spikes.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

I posted that link two days ago in my reply to Win on Mon, 07 Oct 2019.

Here is my post:

"When the capacitor is left with no supplied voltage, the film tends to dissociate and become thinner. Applying a voltage to the cap restores the film."

"An excellent description is given in"

"Leakage current properties of modern electrolytic capacitors"

formatting link

--------------------------------------------------------------------------- Also see

formatting link

This paper has lots of good information, such as fig. 4, leakage as a function of time, voltage and temperature. This precedes your later post showing leakage as a function of voltage. The leakage is exponential.

It also includes information on using resistors to balance the voltages in series capacitors.

One point that has not been discussed so far is the paper mentions the balancing resistors will discharge the electrolytics quickly when power is removed. This is an important safety factor in the case of very low leakage capacitors.

There is a huge amount of information on electrolytic capacitors on the web. For example, see Cornell Dublier, "Aluminum Electrolytic Capacitor Application Guide" at

formatting link

Also see Rubycon, at

formatting link

and so on.

Reply to
Steve Wilson

------------------

** Fig 4 is next to useless. Presents only the barest nominal, virtuual data.
** That tiny we comment is hard to find.

And you missed this post from me here two days ago

-------------------------------------------------------------------

** The actual leakage current with HV electros is normally in the micro-amp range so dissipation causes no discernible temp rise. Ambient temp and rises in same entirely swamp it.

Where series bleeder resistors are used ( eg 100,000 Fender tube amplifiers ) they pass about 1mA which is enough to reduce and nearly eliminate mid point voltage offset.

The main benefit however is in bleeding off charge so the caps do not present a hazard if the AC supply is removed while there is no load on the PSU.

There are a couple of ways this can happen.

See typical schem:

formatting link

----------------------------------------------------------------------

** Yawwnnnn .....

..... Phil

Reply to
Phil Allison

See Fig. 4(b) on Page 2 in

formatting link

The curves are exponential.

Reply to
Steve Wilson

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.