[Boatanchors] RE: selenium rectifier... replacement strategies

[Barry L. Ornitz]

Well, it is pretty obvious that virtually no one on this
list knows any chemistry!

The result of a burned out or overheated selenium rectifier
is selenium dioxide, not selenium sulfide or hydrogen
selenide.  But people are not kidding when they talk about
the toxicity.

Selenium rectifiers do age, with their internal resistance
increasing with age.  They generally do not lose their
reverse voltage rating, however, until their forward
resistance makes them all but worthless.  From a safety
standpoint, they should always be replaced.  If you want an
old radio to look authentic, leave the selenium stack in
place, but hide the new silicon diodes beneath the
chassis.  Modern silicon diodes take extremely little
volume compared to the old selenium stacks for the same
current rating.

A power selenium rectifier, rated at the same operating
current, will generally have a higher forward voltage drop
than a silicon diode.  Since an individual selenium
rectifier is only rated to around 25 volts or so, to get
higher voltages you have to place a number in series.  So
each individual selenium rectifier will have a greater
voltage drop than a single silicon diode, and when you
stack several in series, you can get a sizeable voltage
increase if you replace the stack with silicon diode(s).

I won't name names here, but someone said the voltage drop
of s silicon diode is constant at 0.7 volt.  This is quite
incorrect, and I wish the ARRL would quit propagating this
myth in some of their publications and articles.  A real
silicon diode behaves almost identically to an ideal diode
in series with a resistance over an exceptionally large
range of currents (typically over 8 to 10 orders of
magnitude).  At low currents, the V-I characteristic of the
diode dominate.  The voltage drop across the diode is a
logarithmic function of the current through the diode.  But
at higher currents, the voltage drop is dominated by the
diode's internal resistance and the voltage drop becomes
linearly proportional to diode current.

A selenium rectifier actually behaves in much the same
way.  Based on my research, there are some interesting
differences between selenium diodes and most other
semiconductor diodes (there appears to be what might be
described as a double junction), but the same basic diode
equations can be applied with some changes in parameters.

I don't want to go into all the gory details because I have
not finished my research yet, but it appears that it should
be possible to predict selenium rectifier operation based
solely on the active area and the number of plates.

I will say, however, that at low currents, an individual
selenium rectifier (not a stack which is individual units
in series) can have a lower forward voltage drop than a
typical silicon diode.  But at currents at which they are
typically operated at in power supplies, their voltage drop
is far in excess of a similarly rated silicon diode.  This
is in agreement with the fact that many early televisions
used selenium video detector diodes over silicon diodes
because they were more sensitive.  But this is also in
agreement with the fact that selenium power rectifiers are
far less efficient than silicon diodes.  They are just
operating on different portions of the actual selenium
characteristic curve.

I wish I could post some of the curves here.  I have
comparisons between copper oxide and selenium rectifiers of
the same area, a 1N93 germanium diode, and 1N4148 and
1N4007 silicon diodes.  I have curves shown in both linear
(which is better for showing the diode internal resistance)
and logarithmic (better for showing low current operation)
scales.  I will be happy to email copies of these graphs in
Adobe Acrobat PDF format to anyone who is interested, but I
will warn you they are large files.

Most of my characteristic data was taken from manufacturers
specifications and other published data.  I have recently
discovered a local source for a number of never-used
selenium stacks, and I have collected a number of selenium
units at hamfests too.  These will all go on my Tektronix
semiconductor curve tracer before I complete my work.  I
have a Sarkes-Tarzian engineering handbook on selenium
units, but I would appreciate any other information the
members of this list might have, especially data from
Federal or International Rectifier.  If Ed is willing to
loan a few units for non-destructive testing, I would like
to talk to him offline.  I also need to make a trip to the
University of Tennessee's library to copy about a dozen
articles from some engineering journals of the 1940's on
selenium units.  [I sure wish there was an engineering
school closer to Kingsport than Knoxville!]

As I noted earlier, I eventually hope to be able to
provide some simple equations that describe selenium diode
behavior.  These could be used to calculate the required
series resistance to be added to a silicon diode when
replacing a selenium unit.  Several readers of this group
have detailed the experimental method of getting the
correct resistor value which is a heck of a lot simpler
than what I am doing.  But those that know me (like Ed
Tanton), know that I like doing this sort of research for
my own fun.

I am including with this post (already far too long!) an
article I wrote several years ago on selenium and selenium
rectifiers (even longer!).  It has been updated a little
recently.  The voltage drop estimations are crude in this
post, but they should help you get started.  If you do want
to see the preliminary graphs, drop me private email.  If
you have technical data on commercial selenium rectifiers
of years ago, I would sure appreciate a copy.

So before the article, let me summarize...

Selenium rectifiers do produce poisonous gases when
overheated.

Selenium rectifiers do age with increasing forward
resistance with age.

In power supply applications, when replacing a selenium
rectifier or stack with a properly sized silicon diode, you
will obtain a higher output voltage, often high enough to
overstress downstream circuits.  A properly sized series
resistor can reduce this output voltage to the desired
value.

        73, Dr. Barry L. Ornitz     WA4VZQ     ornitz@tricon.net

Now on to the article...

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

                   Selenium Rectifier Replacement
                    (c) 1997, 2002  B. L. Ornitz

Several people have recently asked about the toxicity of selenium and
its compounds since selenium stacks were common as rectifiers in many
Boatanchors.  Like others here, I too have learned the hard way just
how bad a selenium rectifier can smell when it is overheated or
shorted.  One of my first electronic projects had a small selenium
stack that I wired in backwards.  Naturally I didn't plug it into the
wall socket at my workbench - I carried it into my bedroom to test it!
I couldn't sleep in there for three days!    ;-)   Murphy Rules!

Selenium, a p-type semiconductor, is coated on steel plates in a thin
layer to produce a selenium diode.  Normally seen as a black or gray
shiny coating, the metal is a close relative to sulfur (in fact, I
have made homebrew copper sulfide rectifiers).  It was named after the
Greek word for the moon, Selene, by Berzelius in 1817 because he found
it associated with tellurium which is named for the Latin word for
earth.  If I remember my chemical history, his housekeeper accused him
of eating loads of garlic, when he had not.  This garlic odor is
characteristic of many selenium compounds.  Selenium metal is, in
itself, only modertely toxic.  Its compounds are, however, even to the
point of making some plants toxic to animals when grown in soil rich
in selenium.  [Hopefully we don't have any mega-supplement health food
nuts taking massive doses of selenium here.  Some effects of chronic
exposure include depression, lassitude, fatigue, liver and spleen
damage, yellow skin, garlic breath, giddiness and emotional
instability - and reproductive effects which nature gladly provides to
prevent the stupidity gene from being passed on.]  ;-)

Selenium dioxide is the major compound produced when a selenium
rectifier is overheated.  It can cause severe burns to the mucous
membranes and severe respiratory tract, skin, and eye irritation.  It
is also a dermal sensitizer in that it can promote allergic reactions.
Fortunately it is not considered a carcinogen.  Another fortunate
thing
is the BAD smell.  When I say BAD, I mean really, really, _really_
_BAD_.  Our odor threshold for selenium dioxide is 0.0002 mg/m3.

The allowed exposure for selenium and it compounds (expressed as
selenium) is:

   0.2 mg/m3 OSHA TWA
   0.2 mg/m3 ACGIH TWA
   0.2 mg/m3 NIOSH recommended 10 hour TWA
   0.1  mg/m3 DFG MAK TWA (total dust);
   1 mg/m3 DFG MAK 30 minute peak, average value, once per shift

Note that the odor threshold is _far_ below these.  If you smell
something really rotten, like decaying onions and garlic, coming from
your equipment, it is best to leave the area immediately, opening some
windows on the way out.  Allow the selenium dioxide vapors to
dissipate for several days before you go back.  If you have ever
smelled this odor - believe me - you will not want to go back very
soon anyway!  The odor is very distinctive to say the least.

I almost always replace selenium rectifiers as standard practice in
BA gear. If you have the schematic and know the voltages, choosing the
proper silicon diode is easy.  With no other information, a reasonable
rule of thumb is that a plate in a stack handles about 25 volts RMS.
Thus
the typical stack for rectifying something off the power line would be
five plates.  Another "eyeball" approximation is that the area of each
plate in the stack is about a square inch per 300 milliamps.  For most
boatanchor applications where small stacks are used, common 1 amp
diodes are fine.  Remember that selenium rectifiers have a much
greater
forward voltage drop with current than do silicon diodes.  Thus when
you
replace a selenium diode with a silicon diode, expect a higher voltage
out of your circuit. For most selenium stacks, the DC output is about
2
volts less than the RMS input voltage multiplied by the number of
plates.
This implies a considerable internal resistance.  Silicon diodes
generally
have about a volt of forward drop associated with each diode, but the
internal resistance is very low.  A series resistor may be added if
necessary to drop this voltage.  For applications like the T/R relay
rectifier in a military R-390A, this series resistor is hardly
necessary.
However, for large low-voltage DC supplies using selenium stacks
(often
only one or two plates per leg, but all four legs of a bridge are
usually
mounted in one assembly), a series resistor may be necessary when a
modern silicon bridge is substituted.  Many such supplies included
multiple taps on the power transformer, fortunately, so investigate
this
before wiring in a power-wasting resistor.

>From the interest of safety, I believe it is always wise to replace
selenium stacks with modern diodes. Some restorers leave the original
selenium stacks in older gear to make it look authentic.  Since modern
silicon diodes generate so little heat in comparison, they may often
be
hidden in the circuitry replacing the selenium units while not being
noticed.

   73,  Barry L. Ornitz  WA4VZQ  ornitz@tricon.net   (c) 1997, 2002