Selenium Rectifier Replacement
Barry L. Ornitz
both her and in the antiques group. It is always a good idea to do this
because of the toxicity of the selenium compounds produced when a rectifier
burns out. These rectifiers also age by increasing their internal
resistance which will lower the voltage output. Replacing them with
silicon diodes will usually produce a higher voltage than the original
selenium unit. This is discussed in the paper below. This paper was
written for the Boatanchors mailing list (tube type radios), but it applies
to audio applications too. Most applications for selenium rectifiers in
audio applications involved bias supplies. If these are adjustable, the
higher voltage after replacement is usually no problem as the bias
adjustment will compensate.
It has been pointed out to me that several people here have suggested fast
recovery diodes for this application. This is an expensive, but worthless,
option. Such diodes, in fact, can produce more noise from the power supply
than ordinary silicon diodes. It never ceases to amaze me how the
audiophile crowd can be tricked into expensive exotic components without
any evidence that they can really do anything better than the cheaper
components. All it takes is for someone to say, "It sounds better." I
suspect that some of these people are selling these $3 diodes for $12 to
the audio crowd.
The smart audiophile should go to a regular electronics house and get 10
cent silicon diodes instead.
If anyone wants to discuss these fast recovery diodes, I will be glad to
listen to them - but only if they can postulate a scientific reason why
they should sound better. The audible differences of tube versus solid
state rectification are easily measured and explained scientifically (the
tubes have limited emission and clip softly adding to the "preferred"
distortion). But an ordinary silicon diode versus a fast recovery diode...
why?
Selenium Rectifier Replacement
(c) 1997 B. L. Ornitz
Several BA-folks 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 has 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, not terribly 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 consider 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 always replace selenium rectifiers as standard practice with 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 single stack handles about 25 volts PIV. 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.
Barry L. Ornitz orn...@dpnet.net
Paul Croft
Paul
Barry L. Ornitz <orn...@dpnet.net> wrote in article
<N9n92.3804$td3....@news12.ispnews.com>...
> I have seen a number of posts recently about replacing selenium
rectifiers,
(Snip...)
>
> Barry L. Ornitz orn...@dpnet.net
>
>
>
>
Mr. James C. Mcshane
Practices, there was considerable discussion about diode behavior. It
had more to do with behavior at turn-off than "speed", but I might
suggest that the respective authors had something to say that has
merit.
I'm sure someone else remembers the articles by Rick Miller and John
Camille. I might suggest you review those before arbitrarily dismissing
the idea.
I must confess I have difficulty agreeing with any blanket
pronouncements, and I think it never ceases to amaze me that some
people make blanket statements without a full review of the work in the
field.
No flame war intended, but I confess to thinking there is a difference
between certain diode types.
Jim McShane
Barry L. Ornitz
Mr. James C. Mcshane wrote in message
<747fhu$16b8$2...@newssvr03-int.news.prodigy.com>...
>In an article in Glass Audio, and also a similar article in Sound
>Practices, there was considerable discussion about diode behavior. It
>had more to do with behavior at turn-off than "speed", but I might
>suggest that the respective authors had something to say that has
>merit.
>
>I'm sure someone else remembers the articles by Rick Miller and John
>Camille. I might suggest you review those before arbitrarily dismissing
>the idea.
I mainly dismiss the idea from a technical standpoint. Glass Audio is one
of the last places I would expect to find accurate science when it comes to
electrical engineering. The behavior at turn-off of a fast recovery diode
is somewhat like that of what was once called a snap diode. These were
used as harmonic generators. Starting with an audio frequency signal, a
snap diode could produce harmonics well into the low microwave region. I
have two HP vector voltmeters that use such diodes for exactly this
purpose. From a noise standpoint, then, these diodes should be worse than
softer turn-off diodes. Fast recovery diodes certainly have their place in
switching power supplies and such operating well above the audio range, but
at 60 Hz - why waste your money?
>I must confess I have difficulty agreeing with any blanket
>pronouncements, and I think it never ceases to amaze me that some
>people make blanket statements without a full review of the work in the
>field.
I prefer to not do this either, but the "work in the field", as you call
it, is old hat - most having been done in the 1930's. Like I said, I would
like to discuss this with someone who can postulate a _scientific_
explanation of why such diodes should be better. If so, it can be
measured. As an engineer, I firmly believe it is possible to measure
anything that you can hear. But, and this is a big BUT, we cannot always
predict how the human brain will respond to what is being measured. People
LIKE the sound of tubes, but the total distortion of a tube amplifier is
virtually always in excess of a good solid-state amplifier. That
distortion is easily measurable, both in the time and frequency domains.
What is not so apparent is how the brain perceives certain types of
distortion as pleasurable and other types as punishment. This is more in
the realm of psychoacoustics (as my old friend "jj" might say), than in the
realm of engineering.
Having worked on polymer dielectrics for many years, I am quite familiar
with the subtle effects of dielectric absorption. So I can grant some of
the differences in capacitors and their effect on sound.
But again, most of these effects have been studied and known for years. I
do not dismiss capacitor effects because of this.
If you have _scientific_ articles on the subject of how fast recovery
versus conventional silicon diodes affect the sound, I would love to see
them. If they are like the pseudoscience and handwaving normally
associated with the "audiofool cable" market, they are not worth looking at
however.
>No flame war intended, but I confess to thinking there is a difference
>between certain diode types.
>Jim McShane
Of course there are very big differences, but are you going to see their
effects in a 60 Hz power supply? Germanium diodes have a lower forward
voltage drop than do silicon diodes, selenium diodes have much higher
internal resistance than do silicon diodes, vacuum tube diodes have even
more resistance and it is nonlinear (which is a blessing or curse depending
on your viewpoint). These are well-known properties of their construction
and their effect on audio is fairly easily described mathematically.
Whether you consider them a blessing or curse depends entirely on how well
you like imperfections and what they do to the sound.
I do not want to get in a flame war either, and I wish to thank you for
replying. My intent is to not denigrate the desire for pleasurable
sounding equipment. I could care less if someone wants to buy special
audiophile components. I do hate to see my friends, however, waste their
money on things that are overkill. My real intent is to get the readers
thinking about what they read here and learning "why" things sound as they
do, and not take pronouncements from "golden-eared" experts as gospel.
From what information I have on hand, fast recovery diodes would appear to
offer no advantages in the standard 60 Hz power supply. No one has, in
this group or by private email, yet offered any scientific explanation
otherwise. If they can, and the explanation appears valid and
scientifically correct, I will gladly retract my statement. If there are
articles that do so, I would hope the readers can give me full references.
Thanks again, Dr. Barry L. Ornitz orn...@dpnet.net
stewart ono
of a fast recovery diode. Why not try it and see what it does? FR diodes are
not very expensive, at least not in the current Digi Key catalogs. The
hexfred types, fast turn on, slow turn off, have dropped significantly in
price since the intro of the IXYS brand (again in the Digi Key catalog).
I've tried them, and I like them, especially the hexfreds. I like tube
rectifiers better but with both silicon and tube rectifiers the regulating
circuit is probably of greater importance, or at least it has clouded some
the comparisons I have done for myself. It seems to me the role of a
scientist is to take human perception and then to correlate that experience
to the natural world. The great problem for engineers is the challenge of
differing human perceptions from individual to individual and the fact that
machinery often cannot compete with the human senses. There was once an
electronics company who advertised that human sense of smell could detect
one part per trillion (determined through spectrographic analysis) and they
asked what does this translate to in terms of human hearing. If human
hearing is equivalent to smell, what is one part per trillion equivalent in
distortion for hearing? Rather arguing
try it. Digikey hexfreds are only a few dollars apiece. A couple of bridges
will only set you back $40. Expensive compared to a surplus 5 cent 1N4007
but hardly enough to break anyone's pocketbook.
stu
Richard Matthews
>It has been pointed out to me that several people here have suggested fast
>recovery diodes for this application. This is an expensive, but worthless,
>option. Such diodes, in fact, can produce more noise from the power supply
>than ordinary silicon diodes. It never ceases to amaze me how the
>audiophile crowd can be tricked into expensive exotic components without
>any evidence that they can really do anything better than the cheaper
>components. All it takes is for someone to say, "It sounds better." I
>suspect that some of these people are selling these $3 diodes for $12 to
>the audio crowd.
>The smart audiophile should go to a regular electronics house and get 10
>cent silicon diodes instead.
I fully AGREE. "Fast Recover" is just another thing in the audiophile tweek
vocabulary. If you were o ask people who say that it "makes a difference"
some would attribute sonic benifits exactly the OPPOSITE of what other FR
diode users/audiophiles say.
They are a waste of money.
The big problem is that most do not understand how powersupplies work! If
they did they would realise that FR diodes in audio freq. circuits is Snake
Oil.
Even if diodes were PERFECT the RF generated by cap input PS is so huge that
it swamp artifacts caused by reverse recovery.
>
>If anyone wants to discuss these fast recovery diodes, I will be glad to
>listen to them - but only if they can postulate a scientific reason why
>they should sound better. The audible differences of tube versus solid
>state rectification are easily measured and explained scientifically (the
>tubes have limited emission and clip softly adding to the "preferred"
>distortion). But an ordinary silicon diode versus a fast recovery diode...
>why?
>
Again, I couldn't agree more. One thing for certain is that safety is much
more important than being able to brag that "...my.... is original" and the
reliability benefits make replacement of selenium rec. a closed issue.
>From the interest of safety, I believe it is always wise to replace
>selenium stacks with modern diodes. Some restorers leave the original
>selenium stacks in older gear to make it look authentic. Since modern
>silicon diodes generate so little heat in comparison, they may often be
>hidden in the circuitry replacing the selenium units while not being
>noticed.
>
> Barry L. Ornitz orn...@dpnet.net
>
>
Thank you Barry for being sane.....I was starting to lose hope.
Richard
Bob Fitzgerald
<3lK92.4206$td3....@news12.ispnews.com>...
>From what information I have on hand, fast recovery diodes
would appear to
>offer no advantages in the standard 60 Hz power supply. No
one has, in
>this group or by private email, yet offered any scientific
explanation
>otherwise. If they can, and the explanation appears valid
and
>scientifically correct, I will gladly retract my statement.
If there are
>articles that do so, I would hope the readers can give me
full references.
>
> Thanks again, Dr. Barry L. Ornitz
orn...@dpnet.net
>
relate my
experience. First, the real issue is radiated and conducted
EMI.
The desireable change in switching diodes is to lower it.
The new
diode should have a soft recovery or turn off, and it turns
out
some of the fast diodes are better at this aspect than
others, the
Hexfred design for instance.
The reduction in EMI is measureable, and is easily found in
IR
application notes on Hexfreds, in The Audio Amateur articles
in
DAC960 Pooge (Rick Millers Spectrum Analyzer screen plots of
different diodes in the same circuit). I have also made
comparison
measurements using a HP 8590 SA and am convinced the soft
recovery
diodes reduce the conducted and radiated emission from the
diode.
The noise I measured was actually lower in frequency than I
would
have thought - in the 10kHz-10MHz range.
If anyone would like to conduct their own experiments here,
I
suggest they use an AM radio and a loop antenna as a probe
to
"listen" to different diodes under load in real circuits. A
small
portable AM unit held close to diodes in a power amp would
work
fine as well.
The connection between a local audio circuit being bombarded
with
EMI noise, and our ability to hear it, I'll leave for
someone else
to discuss. This is also pertinent to digital audio, and has
been
given much air time on other newsgroups already.
Bob
LGeoCole
could get into an audio signal path. A power supply is
supposed to be shielded and filtered enough that only DC
gets out. Radiated noise should be blocked by physical
shielding. Conducted noise should be removed by the
filter network. Noise transferred by either path implies
a poor layout and design.
Sticking an AM radio into a power supply will show noise
is present, but not its source. There is a multitude of noises
coming in on the AC line. These can be radiated or conducted
just as well as any noise generated by a power supply. And
most of them are synchronized with the 60 (or 50) Hz line.
The transformer keeps a lot of them out of the supply, but
they are still coming in.
Besides putting a cap across a cheap diode is still cheaper
than a high priced diode. The cap also makes life a little
easier for a diode in a high transient setting (extreme line
pulses, or switched loads).
Let rationality rule,
George C.
Mjzuccaro
I hereby appoint you group moderator.Nice to see an analytical,engineering
viewpoint making it presence known.Be brave!
Best,
Mike Zuccaro
Tim Reese
Barry L. Ornitz <orn...@dpnet.net> wrote:
>If anyone wants to discuss these fast recovery diodes, I will be glad to
>listen to them - but only if they can postulate a scientific reason why
>they should sound better. The audible differences of tube versus solid
>state rectification are easily measured and explained scientifically (the
>tubes have limited emission and clip softly adding to the "preferred"
>distortion). But an ordinary silicon diode versus a fast recovery diode...
>why?
Well, I didn't think that this was as controversial a topic as Ornitz makes
it out to be - and the generic fast recovery / soft recovery diodes, like the
GI 851, are quite reasonably priced. The data published in the Audio Amateur
by Walt Jung et al. appears to demonstrate the advantage of fr/sr diodes in
an unambiguous way - the noise power spectrum for fr/sr diodes contains a
smaller amount of RF hash than the generic 1N400X diodes.
Indeed, the advantages that Ornitz points to in the case of vacuum diodes,
and which I agree with, are limited emission and soft clipping. As I
understand it, fr/sr diodes were developed for switching power supplies with
these superior characteristics (eg low RF emission, soft clipping) in mind.
There are likely supporting articles in EDN or such publications if the
AA article is not convincing.
JMTCW tr
--
----------------------------------------------------------------------
Tim Reese, MGH NMR Center re...@nmr.MGH.harvard.edu
macfa...@hotmail.com
"Richard Matthews" <rich...@worldnet.att.net> wrote:
> I fully AGREE. "Fast Recover" is just another thing in the audiophile tweek
> vocabulary. If you were o ask people who say that it "makes a difference"
> some would attribute sonic benifits exactly the OPPOSITE of what other FR
> diode users/audiophiles say.
> They are a waste of money.
>
I hate to hear that. I´ve just bought some FREDs from VTV to upgrade my MC275
power supply. Well, what the heck I´ve got nothing to lose (except money!). I
will give them a shot anyway. Do Fast Recovery diodes have different voltage
drop than silicons. Do I need to add a series resistor or something?
Regards,
Nestor
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
Bob Fitzgerald
LGeoCole wrote in message
<19981204123912...@ng141.aol.com>...
>I've often wondered how any silicon diode generated noise
> could get into an audio signal path. A power supply is
> supposed to be shielded and filtered enough that only DC
> gets out. Radiated noise should be blocked by physical
> shielding. Conducted noise should be removed by the
> filter network. Noise transferred by either path implies
> a poor layout and design.
>
solid state diodes in the power supply that has a separate
metal shielded enclosure around diodes? Fisher, Eico, and
many other others including HK Citation II had diodes in the
main supply with no metal shield enclosure around them.
>Sticking an AM radio into a power supply will show noise
> is present, but not its source. There is a multitude of
noises
> coming in on the AC line. These can be radiated or
conducted
> just as well as any noise generated by a power supply. And
> most of them are synchronized with the 60 (or 50) Hz line.
> The transformer keeps a lot of them out of the supply, but
> they are still coming in.
What was said was measure(listen to AM radio) with a normal
cheap diode. Change to a soft recovery type, and then
measure (listen to AM radio) again. This is a 2 part or more
test.
Please, if you have tried this and want to criticize it,
fair. But don't do so without trying it first, using
theoretical postulation of results in place of actual
experiment.
>
>Besides putting a cap across a cheap diode is still cheaper
> than a high priced diode. The cap also makes life a little
> easier for a diode in a high transient setting (extreme
line
> pulses, or switched loads).
A while back I posted the references to a lengthy article on
this subject by an author that developed a spice model to
calculate the neccesary passive components to optimize the
snubber action of a circuit across a diode. It turned out to
be fairly complex analysis, and optimized it included a
combination of 3 or 4 resistors/capacitors to effectively
kill the transient noise generated. My feeling is it may be
more elegant a solution to never generate the noise, versus
generating it and then trying to kill it.
>
>Let rationality rule,
>George C.
Ever wonder why International Rectifier invented Hexfred
diodes, and who buys them?
Most (much >99%) are used in non - audio gear. If they did
not provide any benefit, in reducing EMI, why are design
engineers specifying them in new designs? If EEs could get
the same benefit, by using cheap diodes with caps across
them, don't you think they would and save the $ ?
A Dougla6
Nestor asked:
>Do Fast Recovery diodes have different voltage drop than silicons.
>Do I need to add a series resistor or something?
No, but watch the peak reverse voltage rating. Hexfreds only go to
1200V (or did, the last I checked) and there is no "headroom" in the
spec: they really go into reverse breakdown just above that.
73, Alan
John Byrns
<bf...@capital.net> wrote:
> Ever wonder why International Rectifier invented Hexfred
> diodes, and who buys them?
> Most (much >99%) are used in non - audio gear. If they did
> not provide any benefit, in reducing EMI, why are design
> engineers specifying them in new designs? If EEs could get
> the same benefit, by using cheap diodes with caps across
> them, don't you think they would and save the $ ?
Hi Bob,
I'm not a diode or power supply expert, but I suspect one reason for there
use in switching power supplies might be to provide increased efficiency
of operation. Also possibly to prevent overheating of the diode resulting
in self destruction. These are issues at the high frequencies used in
switching supplies, but not in regular passive supplies operating at 50-60
Hz.
Regards,
John Byrns
Mr. James C. Mcshane
arguments. Quite frankly, I think the "not invented here" syndrome
applies to people who, as you put it, want to criticize it without
trying it.
Tim Reese's post was also excellent. Do Messrs. Cole, etc. wish to say
that Walt Jung isn't qualified to pass judgment on this topic? Or John
Camille?
Jim McShane
MLJ...@prodigy.com
macfa...@hotmail.com
Thanks for your reply.
But isn´t 1200 PIV rating enough for the typical power supply?
My MC275 has very early metal encased Sarkes Tarzian silicon diodes. Any idea
of these diodes actual rating?
TubeGuy
The Audio community is being barraged with far too much "snake oil."
I think that if a power supply and the amplifier stages are adequately bypassed, there is no
need for anything other than plain old silicon diodes (or tube rectifiers, of course).
TubeGuy
Mr. James C. Mcshane
You might find there is more than "snake oil" to diode selection.
Also, keep in mind that many "experts" think tubes are just "snake
oil". And in 1980 before the Jung and Marsh series in Audio magazine
"Picking Capacitors" who cared about cap construction, materials, etc.?
My ears tell me there is a difference in the applications I'm familiar
with. Also, the earth, despite the pronouncements of early scientists,
is not flat.
My .02 worth.
Jim McShane
MLJ...@prodigy.com
A Dougla6
Nestor asked:
>But isn´t 1200 PIV rating enough for the typical power supply?
>My MC275 has very early metal encased Sarkes Tarzian silicon diodes.
>Any idea of these diodes actual rating?
I was thinking of ordinary full-wave supplies.Does the 275 have a doubler?
Probably all right there. Do those diodes fit into fuse clips, or are they
even earlier? I remember Sarkes Tarzian M500 diodes in ceramic cases
with metal ends; I have some and could put them on a curve tracer to see
what their real PIV ratings are. By the way, if anyone cares, Tarzian was an
engineer at Atwater Kent in the mid to late 1920s.
Cheers, Alan
SBench
OK, I've stayed out of this long enough. Jim, you've said
it well! Before pronouncing changing out the diode type
snake oil, please review the threads going back quite a
while in RAT. You can MEASURE the difference in fast
vs non fast recovery types quite readily. The turn OFF
characteristics are quite different. Normal 1N4007 etc types
do produce *some* level of RF hash that is lowered by use of
fast recovery diodes. How that manifests itself acoustically
is in the residual noise level - which, since the issue is
"construction" related may or may not have any effect in
*your* particular case. If the noise level is indeed lowered,
it's (not its) going to be more acoustically pleasing. The same
result may be achieved in a number of ways, using fast recovery
diodes is one expediency. There are others.
Best Regards,
Steve
Check my web page .. <A
HREF="http://members.aol.com/sbench101/">http://members.aol.com/sbench101</A>
Remove the .gov to EMail me
macfa...@hotmail.com
adou...@aol.com (A Dougla6) wrote:
> I was thinking of ordinary full-wave supplies.Does the 275 have a doubler?
> Probably all right there.
No, it is a full wave rectifier bridge.
The diodes are conventional looking top hats but metal instead of plastic
encased. Actually my "junker" MC275, of slightly newer vintage, uses the more
common plastic encased variety. The diodes are permanently soldered in a small
separate bakelite board along with a clip on 3 amp slo-blo fuse for B+ and the
bias filter/rectifier. The rectifiers are quite large, about double a 1N4007.
>By the way, if anyone cares, Tarzian was an
> engineer at Atwater Kent in the mid to late 1920s.
Wow!, quite interesting.
I wonder what happened with Sarkes-Tarzian. The company was immensely popular
in the 50s and 60s. I think their strongest market was Tube TV tuner modules.
Thanks again,
Nestor
>
> Cheers, Alan
macfa...@hotmail.com
I“ve done a little research and here is what I found:
Most early solid state rectified tube gear was of the voltage doubler type.
Full wave bridge supplies were quite rare. Even the mighty Citation II used a
voltage doubler. Most McIntosh models used them also. The MC275 was the
exception and used a full wave rectifier circuit. The rectifiers used must
have been extremely expensive for the time-given the rather high peak inverse
voltages they had to withstand.
The MC275 schematic shows an B+ of 453 volts. As a general rule the peak
inverse voltage rating of the diode should be roughly two times the DC B+
voltage. That would give 906 volts. A standard 1N4007 rectifier has a 1000
PIV rating. The IXYS FRED rectifiers I am getting are rated at 1200 PIV. That
means that unless there is an EXTREMELY high line voltage surge (very
unlikely given the stringent outboard AC regulation and MOVs I am using) I
should be okay.
I also found out that the Sonic Frontiers "Power" series are equipped with
(HEX) FREDs. The Power 3, for instance, gives a B+ of 540 volts and has 1200
PIV HEXFREDS. I haven“t heard of any problems with the power supply in these
amps.
I plan to measure B+ in my MC275 before and after the rectifier replacement
and will use heatsinks for the TO220 encased diodes. I think it will turn out
to be pretty neat. My MC275 is sounding excellent and quite frankly it seems
to be hard to improve. However, I am a tweak and I love to work on tube
equipment. Who knows? perhaps I am in for a major surprise!
Regards,
Nestor
LGeoCole
>> Conducted noise should be removed by the
>> filter network. Noise transferred by either path implies
>> a poor layout and design.
>>
>Can you give me an example of any vintage tube amp using
>solid state diodes in the power supply that has a separate
>metal shielded enclosure around diodes? Fisher, Eico, and
>many other others including HK Citation II had diodes in the
>main supply with no metal shield enclosure around them.
construction generally wasn't bothered by small noises. Hard
core hi-fi nuts agreed at the time that a separate power
supply chassis was the way to go.
>>Sticking an AM radio into a power supply will show noise
>> is present, but not its source.
>What was said was measure(listen to AM radio) with a normal
>cheap diode. Change to a soft recovery type, and then
>measure (listen to AM radio) again.
Oops, sorry. Missed that part. I will have to try it.
>>Besides putting a cap across a cheap diode is still cheaper
>> than a high priced diode. The cap also makes life a little
>> easier for a diode in a high transient setting (extreme
>> line pulses, or switched loads).
>A while back I posted the references to a lengthy article on
>this subject by an author that developed a spice model to
>calculate the neccesary passive components to optimize the
>snubber action of a circuit across a diode. It turned out to
>be fairly complex analysis, and optimized it included a
>combination of 3 or 4 resistors/capacitors to effectively
>kill the transient noise generated. My feeling is it may be
>more elegant a solution to never generate the noise, versus
>generating it and then trying to kill it.
How do I find this post? I admit that I wasn't going for the
best possible network here. But any is better than none.
Maybe I am still hung up in sixties technology in worrying
more about protecting the diode than anything else.
>Ever wonder why International Rectifier invented Hexfred
>diodes, and who buys them?
>Most (much >99%) are used in non - audio gear. If they did
>not provide any benefit, in reducing EMI, why are design
>engineers specifying them in new designs? If EEs could get
>the same benefit, by using cheap diodes with caps across
>them, don't you think they would and save the $ ?
These fast turn off diodes are very popular with the high speed
power switching people, mainly power supply designers. Since
there is less power dissapated during turn off, smaller heat
sinks can be used. Smaller power supplies sell better. Also,
in these power supplies, diode noise is not much of an issue.
The oscillator makes more than enough to worry about.
Jim McShane said:
>Do Messrs. Cole, etc. wish to say that Walt Jung isn't
> qualified to pass judgment on this topic? Or John Camille?
I will show my ignorance here. Who?
Oh boy,
George C.
A Dougla6
George Cole wondered:
>>Do Messrs. Cole, etc. wish to say that Walt Jung isn't
>> qualified to pass judgment on this topic? Or John Camille?
>I will show my ignorance here. Who?
This is the Sound Practices article referred to, in the Fall 94 issue #7,
which deals with filtering, shielding, and use of RF-construction
techniques to reduce diode RFI. HexFREDs are not mentioned,
nor replacement of diodes in existing amplifiers.
73, Alan
Mr. James C. Mcshane
(which I don't use, I use the cheaper GI fast/soft recovery diodes),
but only to point out that some solid work in the area of diode turn
off behavior exists.
BTW, since Dr. Ornitz now regards Jung as irrelevant, does that mean we
should now discount his early work as well?
Jim McShane
MLJ...@prodigy.com
TubeGuy
> You might find there is more than "snake oil" to diode selection.
>
> Also, keep in mind that many "experts" think tubes are just "snake
> oil". And in 1980 before the Jung and Marsh series in Audio magazine
> "Picking Capacitors" who cared about cap construction, materials, etc.?
>
> My ears tell me there is a difference in the applications I'm familiar
> with. Also, the earth, despite the pronouncements of early scientists,
> is not flat.
>
background tells me otherwise (provided the PS is adequately bypassed).
Remember, were talking about low Z DC voltage (the end result).
I would attribute any difference one might hear to "the placebo effect" or
another weak point in the overall system, that could be affected by the diode
change.
I'll try to get the Sound Practices and read it, I am open to anything that can
improve the end result.
TubeGuy
Bob Weiss
fast-recovery diode) is as a rectifier in high-frequency switching power
supplies. These devices have a faster switching characteristic, which is
required for maximum efficiency when rectifying high frequencies. Some
of these supplies switch at frequencies FAR above the audio range. These
devices are also sometimes used as snubbers across switching transistors
in motor drives. The fast recovery characteristic helps absorb the high
voltages created by high dv/dt through the motor windings.
I have yet to see a scientific explanation of how these devices would
provide ANY sonic benefit when used in a 60 Hz bias supply.As far as
generating EMI is concerned, these devices would be (theoretically)
worse than regular 1N400x devices, precisely because of the faster
switching characteristic. Diode junctions can and do create RF
harmonics, and specialized diodes are employed for this purpose in
various pieces of test and measurement gear. Any residual noise and
ripple on a bias supply will be removed by the filter network, assuming
that it is properly designed. Using fast recovery diodes to replace a
selenium stack in old tube gear is a waste of money at best. Any claimed
"improvement" would be in the same category as exotic solder alloys,
green CD markers, phase-aligned digital clocks, or $500 "oxygen free,
monocrystalline copper" AC line cords. The well known effect of "If it
costs more, it MUST be better!" , which is all too common in the
high-end audio field.
Bob Weiss N2IXK
Bob Fitzgerald wrote:
> LGeoCole wrote in message
> <19981204123912...@ng141.aol.com>...
> >I've often wondered how any silicon diode generated noise
> > could get into an audio signal path. A power supply is
> > supposed to be shielded and filtered enough that only DC
> > gets out. Radiated noise should be blocked by physical
> > shielding. Conducted noise should be removed by the
> > filter network. Noise transferred by either path implies
> > a poor layout and design.
> >
> Can you give me an example of any vintage tube amp using
> solid state diodes in the power supply that has a separate
> metal shielded enclosure around diodes? Fisher, Eico, and
> many other others including HK Citation II had diodes in the
> main supply with no metal shield enclosure around them.
>
> >Sticking an AM radio into a power supply will show noise
> > is present, but not its source. There is a multitude of
> noises
> > coming in on the AC line. These can be radiated or
> conducted
> > just as well as any noise generated by a power supply. And
> > most of them are synchronized with the 60 (or 50) Hz line.
> > The transformer keeps a lot of them out of the supply, but
> > they are still coming in.
>
> What was said was measure(listen to AM radio) with a normal
> cheap diode. Change to a soft recovery type, and then
> measure (listen to AM radio) again. This is a 2 part or more
> test.
> Please, if you have tried this and want to criticize it,
> fair. But don't do so without trying it first, using
> theoretical postulation of results in place of actual
> experiment.
>
> >
> >Besides putting a cap across a cheap diode is still cheaper
> > than a high priced diode. The cap also makes life a little
> > easier for a diode in a high transient setting (extreme
> line
> > pulses, or switched loads).
>
> A while back I posted the references to a lengthy article on
> this subject by an author that developed a spice model to
> calculate the neccesary passive components to optimize the
> snubber action of a circuit across a diode. It turned out to
> be fairly complex analysis, and optimized it included a
> combination of 3 or 4 resistors/capacitors to effectively
> kill the transient noise generated. My feeling is it may be
> more elegant a solution to never generate the noise, versus
> generating it and then trying to kill it.
>
> >
> >Let rationality rule,
> >George C.
>
Tim Reese
LGeoCole <lgeo...@aol.com> wrote:
>Jim McShane said:
>>Do Messrs. Cole, etc. wish to say that Walt Jung isn't
>> qualified to pass judgment on this topic? Or John Camille?
>I will show my ignorance here. Who?
Jung is well-known to homebrew builders who accept that op-amps might be
useful for some audio purposes ... he's written a few nice books (IC
Op-Amp Cookbook, Audio IC Op-Amp Applications, unfortunately out of print,
others) and probably some industry app notes, as well as many popular
articles. His credentials are linked to industry design and application
jobs at Analog Devices most recently (AFAIK).
The article I referred to, "Measured RFI differences between rectifier
diodes in simple capacitor-input power supplies" by Rick Miller (The Audio
Amateur, 1:94) mentions Jung's participation in the tests leading up to
the article.
I have some more specific biographical info on John "Buddha" Camille, "a
broad and extensive technical background in military electronic, test and
transmitting equipment and ... an ex-fighter pilot and "Top Gun" flight
instructor" (Sound Practices #4, p.3). He's published various articles in
Sound Practices (including the ref'd diode noise article, #7 p.39) and is
most recently affiliated with the self-termed "ultra-fi team" from VSAC
and VALVE magazine. Apparently a very quantitative, no-nonsense kind of
guy with boatloads of expensive test equipment.
No comment intended - FYI. tr
Sheldon D. Stokes
re...@larmor.nmr.mgh.harvard.edu (Tim Reese) wrote:
> Well, I didn't think that this was as controversial a topic as Ornitz makes
> it out to be - and the generic fast recovery / soft recovery diodes, like the
> GI 851, are quite reasonably priced. The data published in the Audio Amateur
> by Walt Jung et al. appears to demonstrate the advantage of fr/sr diodes in
> an unambiguous way - the noise power spectrum for fr/sr diodes contains a
> smaller amount of RF hash than the generic 1N400X diodes.
I didn't see the article in question, but I do believe that Jung knows his
stuff (not nessesarily so for other TAA authors). In this study, did they
put snubbers around the diode to see how much high frequency energy
escapes? I can beleive that there are big differences with a plain diode,
but I'd be very suprised if there was much of a difference with the
appropriate small cap and resistor placed in parallel with the diode.
> Indeed, the advantages that Ornitz points to in the case of vacuum diodes,
> and which I agree with, are limited emission and soft clipping. As I
> understand it, fr/sr diodes were developed for switching power supplies with
> these superior characteristics (eg low RF emission, soft clipping) in mind.
yes, but there's serious problems with vacuum diodes in that you are
limited by instantanious forward voltage, and thus cap size after the tube
rectifier. This tends to make softer supplies. That can be a very
musicly satisfying sound, but from an objective standpoint, it's not ideal
for a supply.
Sheldon
--
"...from Genghis Kahn to the Fuller brush man, they're just a bunch
of losers like me" -Dave VanRonk
Remove SPAM_BE_GONE from my address to reply to me.
Sheldon D. Stokes
> Can you give me an example of any vintage tube amp using
> solid state diodes in the power supply that has a separate
> metal shielded enclosure around diodes? Fisher, Eico, and
> many other others including HK Citation II had diodes in the
> main supply with no metal shield enclosure around them.
actually the diodes in the citation 2 were metal, so you could say that if
the anode had a low impedance path to ground at high frequencies, it is
sor tof sheilded. Now because the Cit has a doubler supply, one of the
two anodes is sort of grounded. The other one is in the middle of the
string, and my citation had a ceramic cap across that one. Interesting?
>
> What was said was measure(listen to AM radio) with a normal
> cheap diode. Change to a soft recovery type, and then
> measure (listen to AM radio) again. This is a 2 part or more
> test.
> Please, if you have tried this and want to criticize it,
> fair. But don't do so without trying it first, using
> theoretical postulation of results in place of actual
> experiment.
I've done things like that before. That doesn't work for digital circuits
though. All that digital hash will make that am radio sound like it's
going to melt down.
> A while back I posted the references to a lengthy article on
> this subject by an author that developed a spice model to
> calculate the neccesary passive components to optimize the
> snubber action of a circuit across a diode. It turned out to
> be fairly complex analysis, and optimized it included a
> combination of 3 or 4 resistors/capacitors to effectively
> kill the transient noise generated. My feeling is it may be
> more elegant a solution to never generate the noise, versus
> generating it and then trying to kill it.
That's true. But because the soft recovery diode isn't a perfect device
either, I'd like to see how a standard cheapie diode and ceramic cap
compares to a soft recovery diode.
> Ever wonder why International Rectifier invented Hexfred
> diodes, and who buys them?
> Most (much >99%) are used in non - audio gear. If they did
> not provide any benefit, in reducing EMI, why are design
> engineers specifying them in new designs? If EEs could get
> the same benefit, by using cheap diodes with caps across
> them, don't you think they would and save the $ ?
I'd bet they are widley used (no pun intended) in sensitive wideband gear,
like o'scopes and such. Ever wonder how tek. can make a DC to light
o'scope that has a sensitivity of liek a millivolt or less? Those circuit
designers have big brains.
I'm of the belief that using a device that never produces any RF hash is a
better idea than trying to snub it out. However, I haven't done the
comparison tests yet, but it's on the list.
Damon Hill
>In article <74ahne$3nb$1...@Usenet.Logical.NET>, "Bob Fitzgerald"
><bf...@capital.net> wrote:
>
>> A while back I posted the references to a lengthy article on
>> this subject by an author that developed a spice model to
>> calculate the neccesary passive components to optimize the
>> snubber action of a circuit across a diode. It turned out to
>> be fairly complex analysis, and optimized it included a
>> combination of 3 or 4 resistors/capacitors to effectively
>> kill the transient noise generated. My feeling is it may be
>> more elegant a solution to never generate the noise, versus
>> generating it and then trying to kill it.
>
>
>That's true. But because the soft recovery diode isn't a perfect device
>either, I'd like to see how a standard cheapie diode and ceramic cap
>compares to a soft recovery diode.
I've been doing some experiments with quieting rectifier noise with snubbers,
although I think mostly what I accomplished was to minimize the ringing of
the transformer's secondary inductance. As a starting point, I recommend
a .47uF and 100 ohm RC snubber across the secondary; some 'tuning' is
necessary of both values to minimize the ringing I could observe on my old
Heathkit 'scope (only 5 Mhz bandwidth, so I may be missing things).
Walt Jung had recommended in email correspondence to use similar snubbers
across each rectifier, and this also had some improvement.
I've even tried a choke-input filter consisting of a largish ferrite toroid
wound common-mode fashion with 16 gauge wire, which introduces new
kinds of ringing problems, of course. I had hoped to create a low-pass
filter, but for all I know it's just merrily radiating away. However, RC
snubbers do appear necessary to reduce what I assume is inductive ringing
caused by the rectifier switching transients, plain capacitors had little
visible effect on the problem
Measurements remain problematic for me. I plan to continue experiments
with soft-recovery rectifiers, MUR-series fast recovery rectifiers, Schottky
power rectifiers, and ordinary silicon rectifiers, just to compare the
differences.
I don't really expect to >hear< differences as my audio system is not
very 'high-end'; I just hope to reduce a source of noise with relatively
simple fixes.
My 'test beds' are a Hafler SE-100 preamp (replaced 1N4004s with
Schottky rectifiers and snubbers) and a hefty power amp based on
Marshall Leach's low-TIM designs (dual mono power supplies with
25 amp bridges that I'm currently trying to quiet with snubbers and
LC filters, to be replaced with various perfboard modules containing
above-mentioned rectifier types.) My Rotel RCD-970 CD player's
audio output board has a separate power supply with bypassed
1N400x rectifiers, but it's under warranty and I can't fool with it
other than to note the designer's practices.
One thing I want to attempt to measure is the amount of noise
in the power amp with varying load levels--does higher current produce more
noise?
>> Ever wonder why International Rectifier invented Hexfred
>> diodes, and who buys them?
>> Most (much >99%) are used in non - audio gear. If they did
>> not provide any benefit, in reducing EMI, why are design
>> engineers specifying them in new designs? If EEs could get
>> the same benefit, by using cheap diodes with caps across
>> them, don't you think they would and save the $ ?
I occasionally see a capacitor across either the bridge output or the
secondary winding, and in some audio equipment I've sometimes seen
small value disc or film caps across each rectifier in a bridge. That seems
to indicate that some designers think it is a problem
>I'd bet they are widley used (no pun intended) in sensitive wideband gear,
>like o'scopes and such. Ever wonder how tek. can make a DC to light
>o'scope that has a sensitivity of liek a millivolt or less? Those circuit
>designers have big brains.
>
>
>I'm of the belief that using a device that never produces any RF hash is a
>better idea than trying to snub it out. However, I haven't done the
>comparison tests yet, but it's on the list.
That's my belief also, but it may be a while before I can test the
various types I mentioned above. As I say, measurement is a problem
for me given my test equipment limitations. The suggestion of using
a cheap AM radio is one I will also try.
I have little idea of what to expect in the way of results; so far however,
I feel that RC snubbers can provide effective noise reduction by helping
to control ringing in the typical LC tank formed by capacitor input
power supplies.
--Damon