Hello Gang;
I did some more work to help me understand more about this topic. I
decided
to take some capacitors from my junk box, and measure leakage currents
versus applied voltage, and then series up a pair to see how they behave.
First I had two Sprague D76588 47 Uf - 450 vdc date code 7904L. These
had
sealed ends on the + end using some sort of reddish glue. My equipment
consisted of an ultra hi-tech Eico RC bridge (the one with a 1629 magic eye
tube), a Beckman DVM (10 meg ohm DVM input), and a Radio Shack VOM (for
current measurements). I ran the caps up to 450 volts for 5 minutes, to
"form" them. This wasn't really necessary because the charge current dropped
under 100 Ua in just a few minutes. I recall doing this to old NOS
"Can-type" capacitors where it took hours to get to the rated voltage...and
many never got there.
After the 5 minute soak at rated voltage, I reduced the voltage to 250,
and
worked up in 50 volt increments. Here is how it went, both caps were pretty
similar, so I will only show one of them here:
250V 3 Ua
300V 5 Ua
350V 7 Ua
400V 12 Ua
450V 45 Ua
475V 105 Ua
Next I put them in series with a 10 meg-ohm resistor across one cap (C1),
and my DVM (10 meg input) across the other capacitor (C2). I also set the
voltage adjust knob on the Eico by loosening the set screw, and moving the
knob for a "dead on" reading at 450 volts. Since I was one DVM short (or is
that bricks), this would have to do. The knob accuracy was very close down
to 250 volts. The Eico folk did a decent job there! So here is how the
capacitors balanced:
V-Applied: leakage current: Volts across C2
250 v 11 ua 118 v
300 13 141
350 15 165
400 20 195
450 25 221
The voltages balanced very well. Turns out that the currents seen are
basically the current through 20 meg ohms. At 450 volts, I = 450/20e6 = 22.5
ua. So in this test the data is skewed because the resistances used across
the caps (10 meg across each) still swamped out the leakage current through
the capacitors. I needed to go higher in voltage in order for the capacitors
to get leaky, and possibly miss-behave. With my Eico, 475 volts was all she
wrote! Back to the drawing board.
After doing this, I measured the capacitance with my Beckman which has C & L
measurement capability. C1 was 35 uf, and C2 was 36 Uf. Both low, but well
matched.
The second set of capacitors were Sprague 8 TVA 1344 60 Uf - 100 vdc. I did
the same test, this time from 50 volts up to 250 volts. I had to make a
chart of Eico knob setting VS actual voltage. The Eico folk got sloppy at
the low end calibration! Here is how it went:
V-Applied: leakage current: Volts across C2: Calculated voltage
across C1:
50 v 0 ua 27 v 23 v
100 0 50 50
150 3 74 76
200 10* 97 103
250 350 116 134
* Bleeder current at 200 volts: I = 200/20e6 = 10ua
The equalizing resistors again held the voltage even until we got into the
"surge voltage" rating of the capacitors where the leakage currents rise
very rapidly. It is interesting that with a 25% over voltage (250/200), the
imbalance was only 15% (134/116). I still need a better way to measure these
voltages with minimum circuit disruption. Any thoughts guys?
I did one more test where I paralleled C2 with a 220 K resistor with 200
volts V-Applied. The thinking here was to simulate extra leakage in one of
the capacitors, but not the other. This brought C2 down to 75 volts, and C1
rose to 125 volts. As expected, the leakage current rose significantly to
350 ua. The power dumped into C2 increased, but 350 ua * 125 v = 0.04 watts.
This is still a workable imbalance. Now leave C2 at 125 volts for a year, an
d see if the leakage current changes... Big question.
It does seem to appear that series capacitors without equalizing
resistors
will work when the there is sufficient headroom should one capacitor take a
disproportionate share of the total. If the leakage current is still small,
then this is of little consequence. In my case with my 20A, I have two 250
volt electrolytics in series, each rated at 300 volts surge, so with 450
volts across the series combo, there is room for them to move around some.
More to come, unless you guys shoot me first!
Regards,
Jim candela
WD5JKO
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] Behalf Of Mark Foltarz
> Sent: Tuesday, February 25, 2003 11:27 AM
> To: [email protected]
> Subject: Re: [AMRadio] Series capacitor equalizing resistors
>
>
>
> >
> > I realize that this debate over equalizing resistors with
> series >capacitors
> is controversial
>
> Jim,
>
> I am at a loss. What controversy? It has always been
> engineering practice to
> include equalizing resistors in series cpacitor filters. Even with the new
> capacitors initial low leakage these caps will eventually start
> to leak. Yes,
> the new caps are much better than the old ones but its not the same as
> semiconductor technology.
>
> Can anyone point out ANY commercial gear that uses series
> filter capacitors
> without equalizing resistors?
>
> Well, without the use of equalizing resistors in your series capacitor
> filter bank you'll be the one cleaning up all the goop that blew
> out of the
> electrolytics when one of them starts to leak and the divider goes out of
> balance. Won't happen today. Won't happen tomorrow.... But it will happen.
>
> Maybe with luck the worst you'll see is a blown fuse. Or perhaps
> these are'nt
> necessary either? HI!
>
> 73
> de KA4JVY
> Mark
>
>
> --- Jim Candela <[EMAIL PROTECTED]> wrote:
> >
> >
> > Hay all,
> >
> > I have been experimenting with series electrolytic
> capacitors, and when the
> > capacitors are high grade, and of equal capacitance, they seem
> to drop the
> > DC voltage pretty evenly when used without equalizing resistors.
> >
> > I have two Central electronic 20a's, and the HV 40-40 Uf @
> 450 wvdc can
> > electrolytics needed to be replaced. This brings up another
> issue where with
> > today's line voltages, these filter capacitors often see higher
> than rated
> > voltage during warm-up when the 5U4G heats up faster than the indirectly
> > heated tubes in the radio. We often need 500+ volt
> electrolytics for bullet
> > proof reliability. Digging through my junk box I found eight
> NOS Sprague 100
> > uf 250V radial leaded capacitors date code 82 something (figure
> 20+ years
> > old). These were unused.
> >
> > I took these capacitors and with a series resistor of about
> 10k (5 watt WW)
> > I charged them to 250 volts. leakage currents were nil, and later (after
> > discharging!) I measured the capacitance. All 8 were the same on my 'C'
> > meter at 109 Uf. I found it strange that they were so close. I kept
> > remembering tolerances like +80%, -20%, but not in my case.
> >
> > I then fiddled around with some clip leads, and series two
> of the caps up,
> > and placed them across 450 vdc with no resistors. With my 10
> meg input DVM,
> > I measured 200 volts across one of the caps, and then 200 volts
> across the
> > other. Wait a minute! 200 + 200 does not equal 450. Apparently
> the 10 meg
> > DVM impedance was influencing my reading by increasing the
> leakage current
> > across the capacitor being measured. Since I did not have two identical
> > DVM's I conclude that the capacitors were sharing the 450 volts
> (225 + 225)
> > pretty evenly. So it looks like I made two 50 Uf 500 Vdc
> capacitors out of 4
> > pieces rated 100 uf, 250 vdc.
> >
> > These capacitors are now in both my 20a's with no problems.
> One change was
> > that even though one of the can electrolytics was still working, after
> > replacing it with my 4 caps ( pi filter ) the HV B+ rose some,
> and the 120
> > hz ripple voltage diminished considerably. I guess the can capacitor,
> > although still working had higher ESR, and maybe diminished
> capacitance. My
> > 'C' meter showed over 40 Uf capacitance, so I guess it was the higer
> > effective serial resistance (ESR) that degraded over the past 50 years.
> >
> > I realize that this debate over equalizing resistors with
> series capacitors
> > is controversial, and to some extent is similar to equalizing resistors
> > across series up HV diodes. In the later case, it is now
> becoming accepted
> > that with modern high quality diodes, that these resistors don't help
> > protect the diodes. I am beginning to think that the same is true with
> > series up modern high quality capacitors. I am talking about
> modern computer
> > grade capacitors that are small, cheap, and reliable up to
> voltages of 250
> > vdc. I am not referring to no-name 450 volt units from various
> suppliers.
> > Mouser & Digikey have a wide selection of good parts that I am
> referring to.
> > It would be interesting to see how the voltages would imbalance if the
> > capacitors were poorly matched in terms of capacitance (assume
> equal leakage
> > currents).... Any thoughts?
> >
> > Regards,
> > Jim Candela
> > WD5JKO
> >
> >
> >
> > > -----Original Message-----
> > > From: [EMAIL PROTECTED]
> > > [mailto:[EMAIL PROTECTED] Behalf Of [EMAIL PROTECTED]
> > > Sent: Saturday, February 22, 2003 5:52 AM
> > > To: [email protected]
> > > Subject: Re: [AMRadio] DRIVER XFMR UPDATE
> > >
> > >
> > > In a message dated 2/21/03 9:23:48 PM Eastern Standard Time,
> > > [EMAIL PROTECTED]
> > > writes:
> > >
> > >
> > > > If it's impossible to find 125 mfd's (which I doubt...they can
> > > > be found), then you can put two, say 65's or 75 mfd, in series
> > > to replace
> > > > each
> > > > 125mfd's.
> > > >
> > >
> > > Nope... capacitors are not like resistors. putting them in
> > > series will halve
> > > the total value, not double them. To get the equivalent
> value of two in
> > > series, you have to use caps twice the value. In this case,
> > > you'd put (2)
> > > 250 ufd caps in series to get the equivalent of 125 ufd. It's
> > > advisable to
> > > also put equalizing resistors across each cap to make sure that
> > > the voltage
> > > division is equal to prevent one cap from having a
> > > higher-than-rated voltage
> > > across it.
> > >
> > > Eric, W2CAU
> > >
> >
> > _______________________________________________
> > AMRadio mailing list
> > [email protected]
> > http://mailman.qth.net/mailman/listinfo/amradio
>
>
> __________________________________________________
> Do you Yahoo!?
> Yahoo! Tax Center - forms, calculators, tips, more
> http://taxes.yahoo.com/
> _______________________________________________
> AMRadio mailing list
> [email protected]
> http://mailman.qth.net/mailman/listinfo/amradio
