tube rectifiers, HexFreds, and the Big Deal

[Bill Layer]

Hi Larry/All,
 
>OK, Bill.   You've got me thinking again (dangerous).  Your explanations
>are easy enough to follow and appreciated.  Now, tell us how the tube
>rectifier can help us avoid this noisy peak-only rectifier conduction.

Well, the tube rectifier acts like sort of a shock aborber, becuase it
doesn't have the purveyance of an SS dioe. They are still a nonlinear
resistance, but the drop across a tube diode can be anywhere from several
volts to many tens. And like most resistors, the drop varies with current.
 
 >You've lost me there.  However, it would seem to me that much of the sine
>wave distortion would appear as internal voltage drop across the
>transformer impedance.  I would think that very little would appear back on
>the AC main line.  After all, when considering the source impedance (all
>the way back to the generator) of the AC main to your rectifier/filter
>capacitor, the lion's share is the final step down transformer.  Most of
>the voltage drop caused by the high current flow must appear across most of
>the impedance.

Get a transformer manufactuer's application guide for DC supplies. You'll
notice that the xfmr is only good for it's full ratings with one particular
style of filter design (inductive). They can probably give a more
authorative explanation than I can. But a good example of this philosophy
is found in most tube- regulated power supplies. Though such supplies may
boast a total ripple of .01% or less at full ouput, they usually have no
more than 2-10uF of filter capacitance, in a C-L-C config!

 
>I think the size of the supply relative to source impedance is the key
>issue.  I might buy that argument if the 10X ovefiltered supply was sized
>for that 1000 watt sub-woffer amp you just bought.  However, on that +/- 15
>VDC synth module drawing 50 ma, I really wonder if it is significant.

You know it. It really enforces that notion that there is no such thing as
a 'generic' supply, and that all supplies can and should be optimsed for a
particular circuit. There's no point flogging a poor little xfmr with 10X
the filter it needs, when a two stage filter (even C-R-C) with a carefully
chosen resistor would do as well or better. 

>One other thought just popped into my head.  If we are concerned about the
>harmonic noise generated back to the AC from the partial wave conduction,
>why couldn't we identify the particular harmonics of a particular power
>supply and use a filter back on the AC side of the supply to protect the
>line.

Well sure, but that energy is still lost as heat in the filter, and is not
being used productively. We'd be better off not generating it at all.
>
>Am I crazy? I think Harmonic filtering is not totally uncommon on high
>voltage systems where DC varible drives and the like create significant
>harmonic distortion.

Oh, I think you're right. The power company requires large industrial
consumers to correct their own power factor, at their facility. The reason
is twofold: 1) distortion products affect the rest of the system 2) the
consumer would else be ripping of the power co, as the distorted sine wave
no longer reads full value on the electric meter... This works at home too.
When you run inductive or electronic loads, you are getting slightly more
than you pay for!
> 
Is this just barely synth DIY?


Bill Layer
Sales Technician
<b.layer at vikingelectronics.com>

----------------------------------
Viking Electronics, Inc.
1531 Industrial St.
Hudson, WI. 54016 - U.S.A
715.386.8861 ext. 210
<http://www.vikingelectronics.com>
----------------------------------

"Telecommunications Solutions for the 21st Century"