Larry,
Thanks for the constructive input. I am beginning to lean towards a less
simplistic approach on this topic. I think that equalizing resistors passing
10 ma (typical) is pretty overkill, unless you need a bleeder resistor
anyway. Also hot equalizing resistors are often attached to capacitors
directly heating them up from the inside out. In this instance, the
equalizing resistors contribute to the early demise of the capacitors for
thermal reasons. I have been leaning towards using equalizing resistors but
reducing the bleeder current to 1 ma. Then today I was looking at HV
Panasonic electrolytics in Digikey, and they were rating some of the 450 vdc
units at 1.6 ma leakage current. Other brands of the same Voltage &
capacitance are rated at 100 Ua max. I forget the brand, but Digikey dropped
the line, and mainly carries Panasonics now. Mouser has a better selection.
A good analog / power Engineer gave me a different perspective on series
electrolytics:
-----Original Message-----
From: cttaylor [mailto:[EMAIL PROTECTED]
Sent: Thursday, February 27, 2003 10:34 AM
To: Candela, Jim;
Subject: Re: [AMRadio] Series capacitor equalizing resistors
Hi Jim,
Good stuff, and I agree as far as it went. My problem is with the
transient conditions that can occur during turn on/off and accidental short
circuit of one (or both) of the capacitors. Remember my background in high
voltage and transient induced failure prevention.
OK, the quick answer is to put a 400+ volt, 20A pulse rated (1/2 60Hz
line cycle) diode backward across each capacitor in the string. I won't
even try to to describe all the bad stuff that goes on during a high current
short, but basically if you prevent the current shoved backward through the
capacitor from reverse biasing it, you greatly reduce the possibility of
punching a hole through the oxide insulation on the foil. Remember that
during a trasient, having the capacitances balanced is not enough, the ESR
of the capacitor also comes into play. Again, more complex than I like to
think about, but the diode also works to prevent this by preventing the
reverse current from reverse biasing the cap.
That's my nickel's worth. Ya'll be carefull out there.
Craig
I then answered craig with this email:
Craig,
Thanks for the input. I see your point about putting suitable diodes
across each capacitor to provide a better defined path for discharge
currents in the event of an inadvertent short circuit. With series
capacitors in normal power up, power down, the lack of diodes, or equalizing
resistors doesn't seem to be a big issue as long as the capacitors are
similar in characteristics. When powering down, the bottom (closest to
ground) cap has to discharge through the top capacitor in the reverse
direction. When the discharge time is 100 ms, or more, the reverse current
is not high (compared to a fault condition), and maybe that is OK.
For short circuits after the caps are charged, the diodes should
help. Would it also be good to add a series resistor with the capacitor
bank, say a few ohms in a high impedance tube type power supply where a few
ohms increase in ESR won't degrade the capacitor filtering effect (120 hz
ripple smoothing) noticeably? That way the peak current is defined precisely
as i=e/r, and something the diodes can tolerate? For example, with 450 volts
, a series resistor of 22 ohms will limit the peak current to 20 amps at the
instant the short occurred. Beefier diodes with higher surge ratings could
maybe get the resistance under 10 ohms. Any thoughts as to what kind of
resistor would be a good choice?
As for equalizing resistors, past practice has often led to about 10
ma bleed current through the divider. This will do a great job equalizing
the voltage across the capacitors, and provide a bleeder current should one
be necessary. With a 450 volt B+, 10 ma will generate 4.5 watts of heat
(distributed through two or more resistors), or in a linear tube amplifier
with say 2500 volts, that's 25 watts to warm up the capacitor bank. The
equalizing resistors if attached directly to the capacitors will warm up the
caps from heat conducted through the copper wires, and into the capacitor
terminals. Even with 10 ma bleeder current, a fault condition could still
lead to upper capacitor dielectric punch through from high reverse current.
In many instances a safety bleeder is separate from the capacitor
equalizing resistors. In this case, when using modern capacitors (where
leakage currents at rated voltage are < 100 ua), maybe drop the equalizing
current to 1 ma, and use diodes as you describe. Maybe a little series R as
I suggest also. Would this be a good recommendation, and if there is no
external bleeder circuit, increase the bleeder current to 10 ma? For safety
reasons the capacitors should be discharged as quickly as possible
(preferably in less than 1 minute)..
Regards,
Jim
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] Behalf Of Larry Szendrei
> Sent: Wednesday, February 26, 2003 9:49 PM
> To: amradio@mailman.qth.net
> Subject: Re: [AMRadio] Series capacitor equalizing resistors
>
>
> The voltage would divide according to the capacitance values (assuming 0
> leakage in both caps). Using the formula C=Q/V, where C is capacitance
> in Farads, Q is charge in Coulombs, and V is voltage (in volts, of
> course!), Q necessarily has to be the same for both caps in a series
> circuit once both are charged. Therefore, Q = C1/V1 = C2/V2, and the
> magic of algebra yields V1/V2 = C1/C2.
>
> 73,
> -Larry/NE1S
>
> Jim Candela wrote:
>
> > 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
>
> _______________________________________________
> AMRadio mailing list
> AMRadio@mailman.qth.net
> http://mailman.qth.net/mailman/listinfo/amradio
