[Amps] Filter Capacitors

[Carl]

And after all that the handbooks are still the place to go for a good 
understanding as well as the software I suggested in a private email.

Its not a subject that can be covered in a few paragraphs. Study the book 
and THEN ask questions here and Ill be glad to answer them in more depth.

Carl
KM1H


----- Original Message ----- 
From: "Manfred Mornhinweg" <manfred at ludens.cl>
To: <amps at contesting.com>
Sent: Thursday, April 12, 2012 10:13 PM
Subject: Re: [Amps] Filter Capacitors


> Jim,
>
>>> How do you conclude the proper value (both the voltage and capacitance)
>>> for a filter capacitor for a 1500 watt linear amplifier?
>
> I will go into a little more detail than Carl did! ;-)
>
> You need to decide how much ripple is acceptable, and then simply go by
> the definition of the units. One farad is the amount of capacitance that
> will gain or loose one volt in one second, if a current of one ampere
> charges or discharges it. As simple as that.
>
> Let me calculate it with an example: Suppose you have an amp that takes
> 3000V and 1A, and that you decide that 100V ripple is OK. Suppose too
> that you live in a country that has 60Hz line frequency, and that the
> power supply uses a bridge rectifier and a simple capacitive filter,
> without a choke.
>
> In that case, a half cycle lasts for 0.0083 seconds, and the filter cap
> will have to power the amp for ALMOST that entire half cycle, before
> being charged by the next AC peak. We can simplify that "almost" and say
> that your filter cap needs to power the amp for 0.008 seconds only, and
> in that time it's allowed to drop by 100V, when delivering 1A. So this
> turns to simple math:
>
> 1A * 0.008s / 100V = 0.00008F, or 80uF.
>
> Note that this assumes that the transformer and the power line can
> actually deliver the intense charge pulse, which would be several
> amperes! In a practical case, the pulse will hardly be more than 3 or
> 4A, and so it will be somewhat longer than the 0.0003s I assumed, and
> this results in the capacitor having to power the amp for less time, and
> thus less ripple. How much less, depends on the combined soruce
> resistance of the transformer, power line, and anything else in series
> with them!
>
> Carl's roughly 30uF would result in about 200V ripple, and a somewhat
> softer and longer charge pulse in each half cycle. It's probably quite
> workable.
>
> It should also be good to understand how the supply voltage ripple
> affects amps. Pentodes and tetrodes have a gain that's basically
> independent from the plate voltage (but the screen voltage must be well
> filtered to keep the gain from mecoming hum-modulated!). So a pentode or
> tetrode can do with quite a lot of ripple on the plate, and you would
> notice it in the output only if you drive the tube to peak power, as the
> peak capability would be modulated by the ripple. On the other hand, a
> triode's gain depends on the plate voltage, so any ripple in the supply
> will modulate the gain and thus the entire output, at all power levels.
> For that reason I would say that with a triode amp you should use a
> pretty well filtered HV supply, while a pentode or tetrode amp can be OK
> with less good filtering at the plate, but good filtering at the screen.
>
> About the voltage rating, well, it has to be high enough to make the cap
> survive! You need to consider the highest line voltage to be expected,
> calculate how much peak voltage the transformer will put out with that,
> and then select a capacitor voltage rating that has  a reasonable safety
> margin on top of that. How much safety margin, well, I would say at
> least 20%. If you are more generous, you can enjoy a longer capacitor
> life expectancy, but usually at much higher cost and bulk.
>
> If you have to use electrolytics, you will need several in series, and
> then you need to put in some higher safety margin, to account for uneven
> voltage distribution between them, according to the tolerances of the
> equalizing resistors, and possible uneven leakage in the caps.
>
> Manfred
>
> The 4CX1000 is dead. Long live the MRFE6VP61K25H!
>
>
> ========================
> Visit my hobby homepage!
> http://ludens.cl
> ========================
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 10.0.1424 / Virus Database: 2411/4934 - Release Date: 04/13/12
>