Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!uwm.edu!linac!att!cbnewsj!asd
From: asd@cbnewsj.att.com (Adam S. Denton)
Newsgroups: sci.electronics
Subject: Re: IR Reception - Modulation and Quality
Keywords: harmonics
Message-ID: <1991Apr17.201326.14990@cbnewsj.att.com>
Date: 17 Apr 91 20:13:26 GMT
References: <4729@osc.COM> <1991Apr10.235418.10741@cbnewsj.att.com> <4739@osc.COM>
Organization: AT&T
Lines: 67

In article <4739@osc.COM> jgk@osc.COM (Joe Keane) writes:
>In article <4729@osc.COM> i write:
>>Note that, especially with a large filter capacitor, the peak current can be a
>>lot higher than the average current if the rectifier only conducts during a
>>small portion of a cycle.  The choke should reduce this problem somewhat, but
>>you should make sure the choke is rated for the peak current.
>
>In article <1991Apr10.235418.10741@cbnewsj.att.com> asd@cbnewsj.att.com (Adam
>S. Denton) writes:
>>You DO NOT want to `filter' pure AC.  You'll just be introducing a voltage
>>loss.  You DO want to filter the AC component of DC.  Thus, the choke goes
>>after the diodes (and initial filter cap), not before.
>
>I stand by my original statement.  It's more important to filter the current
>before the capacitor.

Try it and see.  I have designed and built many power supplies and switchers.
You are right technically, in that a series inductor will do some filtering,
but that same inductor will be much for effective in a pi filter.
Try it.  If it doesn't turn out that way, I'll eat my hat.

>The key phrase here is `harmonic current'.  A rectifier and capacitor
>connected directly to the mains would produce large harmonic currents.
>Somewhere in this path you need some impedance to smooth out the current.
>Like i said before, cheap power supplies rely on the resistance and leakage
>inductance of the transformer.  With reasonable filtering before the
>rectifier, the high harmonics aren't generated in the first place.  This is a
>lot easier than trying to filter them out later.

Oh...I thought your goal was the most noise-free DC output from which a
circuit could be powered.  If your goal is the slowest-changing currents
thru the rectifiers, then of course you'd want the L there.  But
if you want the smallest ripple (at any freq) at the supply output,
which is usually what you want, then move the L to the DC portion.

>With large rectifiers, series inductors are mandatory.  Large AC-DC converters
>have elaborate filtering schemes, with series inductors and multiple tuned
>shunts, and these are all on the line side.  The power company would have a
>bird if you hooked up a large rectifier and capacitor directly to the power
>line.  I don't know what specifically `large' would be though.

Actually the rectifer would release its smoke, and thus cease to work. :-)
The L's and RC filters in switchers etc. in series with the diodes are there
for very different reasons than the need you implied in your 1st post.
They are there to limit absolute voltages, absolute currents, and
to limit the derivates of both with respect to time.  These constraints
are seldom necessary in a low-frequency (120Hz), low-voltage supply.
Most importantly, they are not there for `filtering' per se.  They are
there to ensure the V, I, dV/dt, and dI/dt ratings of the rectifiers
and othere components are not exceeded.  Otherwise, device failure will occur.

>Assumedly, the voltage on the filter capacitor is subsequently regulated.  A
>regulator can deal with some 120 Hz hum, and if that's all there is, very
>little will get through to the output.  That's because it can respond well to
>slowly changing voltages, and 120 Hz is fairly slow.  However, high-frequency
>transients will shoot right through, or around, by capacitive coupling.

This is true.  Especially troublesome are the RF components radiated
directly by the diodes themselves.  Often diodes in, e.g., TV main supplies
have series ferrite cores (i.e. inductance) and parallel capacitors
to limit the slewing of the current.  Yes, you can probably do without
a series choke if there is a post-regulator for LF.  But put that choke in,
and you'll see the HF disappear also.  Try it!

Adam Denton
asd@mtqua.att.com
Disclaimer: Presently, I am *not* wearing a hat.