It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

"gojamo"

It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance.


** Not common at all - only ever done by the most gullible audiophools
and shonkiest snake oil merchants.


Is there any proof that this is beneficial?


** If you mean scientific proof - there is none at all.


Is there a rule of thumb as to the size of such
capacitors? Thanks


** Depends only on the size of your wallet and how gullible you are.

Nothing to do with the size of your thumb whatever.




............ Phil

"gojamo" wrote in message
ups.com...
It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

Large capacitors inherently exhibit a small amount of inductance (L) mostly
because of their spiral windings. I always thought that if you paralleled a
small film capacitor that typically has a low "L" across the electrolytic,
high frequency current would be easier to pass. Remember that frequency is
inversely proportionate to the Capacitive reactance. Xc = 1/2pi * f * c
while Inductive reactance is proportioate to frequency XL = 2pi * f * L.
So at least, I hope that this proves some benefit, at least mathematically.
Practically (hearing wise), it may be like picking fly sh-t out of pepper,
but that never stopped a true audiophile from pursuing the Absolute Sound.
Cordially,
west

The ESL and ESR of modern electrolytics is probably lower than that for the
older types, so it isn't as crucial as it was before. I think the old rule
of thumb was 1% (or perhaps 10%, can't remember). In fact, you will end up
with a parallel resonant LC circuit at HF when the electrolytic goes
inductive if you try to bypass it.

IMO it is mainly audophool drivel to sell overpriced caps. But of course
some will disagree.

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

Jason



"gojamo" wrote in message
ups.com...
It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

"west"


Large capacitors inherently exhibit a small amount of inductance (L)
mostly
because of their spiral windings.


** Massive, bloody stupid LIE !!!!!!

There is no damn spiral - all the edges are joined !!!!


I always thought that if you paralleled a
small film capacitor that typically has a low "L" across the electrolytic,


** Film caps are wound too - dickhead !!

But also in a way that in not inductive.


high frequency current would be easier to pass. Remember that frequency is
inversely proportionate to the Capacitive reactance. Xc = 1/2pi * f * c
while Inductive reactance is proportioate to frequency XL = 2pi * f * L.
So at least, I hope that this proves some benefit, at least
mathematically.


** The simplest test or analysis shows it does not and cannot work.



................. Phil

Rule of thumb is if it's good enough to decouple the supply from the audio
floating on it, it's good enough for that amp.

Personally, I always use active regulation for all low-level stages,
therefore making type and size of capacitors almost irrelevant.


--
Gregg "t3h g33k"
http://geek.scorpiorising.ca
*Ratings are for transistors, tubes have guidelines*

"gojamo" wrote

It's common to parallel electrolytic capacitors in a power supply
or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof
that
this is beneficial? Is there a rule of thumb as to the size of
such
capacitors? Thanks


The rule of thumb used to be to bypass each cap with one of 100th
its value.

The HF behaviour of many electrolytics is good enough now to obviate
the need.

Another issue is distortion. Cap manufacturers don't publish test
results for this, and I think it is worth buying on reputation.
Electrolytics, particularly with high bias voltage, can produce
marginally significant harmonic and IM distortion. Quality of
manufacture is important. Best to avoid them wherever possible.

Proof of benefit is hard because benefit is subjective. Proof of ESR
is easy to find in specs (either direct or in the form of quoted max
ripple voltage). ESL, or variation of capacitance with frequency,
isn't often quoted though AFAIK.

cheers, Ian

gojamo wrote:

It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

Say tou have a 40 year old tube amp with a 16 uF electro cap.
As F rises, the cap becomes inductive, unless we shunt it with
something that remains a capacitor up to a far higher F.
So folks might use a 0.47 polypropylene cap strapped across the old
electro.

But with modern electros made for SMPS, their performance is much better,
and they remain a purer capacitance up to a higher F.
I still place a plastic 0.47 uF across 3 x 470uF caps in parallel
in a supply to a CT on an OPT.
I also add plastic caps, say 2 uF, or whatever I have
to the B+ of early amp stages and place these caps near the tube and
with short leads.
Sometimes higher Gm tubes and cahtode followers can oscillate
at RF if the rail voltage isn't bypassed well enough at RF.

Patrick Turner.

"Ian Iveson" = a colossal pommy fool and spiteful little ****


The rule of thumb used to be to bypass each cap with one of 100th its
value.


** A 10 uF 350 volt electro of 1950s vintage benefited from a 0.1uF paper
fitted cap in parallel. The simple reason was that the electro had an ESR
of 5 or 10 ohms - the paper cap bypassed that resistance above 500 kHz.

This sometimes mattered in valve radios that could become unstable
otherwise.

Note the ratio is 1% .

This is where the "rule" comes from.



The HF behaviour of many electrolytics is good enough now to obviate the
need.


** Low value, high voltage electros can still have high ESRs - ie several
ohms or more.


Proof of benefit is hard because benefit is subjective.


** Bunkum - it is simply non existent in the audio band.


Proof of ESR is easy to find in specs (either direct or in the form of
quoted max ripple voltage).


** Simple to measure - just do an impedance test at 100kHz and you have a
reliable measure of electro ESR.


ESL,


** ESL for any cap is very nearly the same as a length of wire the same
size as the cap body plus any leads.

A 1 inch length of wire has a L of about 25 nH.

Keeping leads short and circuit connections direct matters more than cap
type.




........... Phil

On Tue, 08 Mar 2005 00:10:27 +1100, Patrick Turner
wrote:



gojamo wrote:

It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

Say tou have a 40 year old tube amp with a 16 uF electro cap.
As F rises, the cap becomes inductive, unless we shunt it with
something that remains a capacitor up to a far higher F.
So folks might use a 0.47 polypropylene cap strapped across the old
electro.

But with modern electros made for SMPS, their performance is much better,
and they remain a purer capacitance up to a higher F.
I still place a plastic 0.47 uF across 3 x 470uF caps in parallel
in a supply to a CT on an OPT.
I also add plastic caps, say 2 uF, or whatever I have
to the B+ of early amp stages and place these caps near the tube and
with short leads.
Sometimes higher Gm tubes and cahtode followers can oscillate
at RF if the rail voltage isn't bypassed well enough at RF.

Patrick Turner.



Or you can go all film with these:

http://www.eci-capacitors.com/
http://www.ecicaps.com/ECI_ul30_data.htm




Regards,

Boris Mohar

Got Knock? - see:
Viatrack Printed Circuit Designs (among other things) http://www.viatrack.ca

Phil Allison wrote:

"west"


Large capacitors inherently exhibit a small amount of inductance (L)
mostly
because of their spiral windings.

** Massive, bloody stupid LIE !!!!!!

There is no damn spiral - all the edges are joined !!!!

I always thought that if you paralleled a
small film capacitor that typically has a low "L" across the electrolytic,

** Film caps are wound too - dickhead !!

But also in a way that in not inductive.

high frequency current would be easier to pass. Remember that frequency is
inversely proportionate to the Capacitive reactance. Xc = 1/2pi * f * c
while Inductive reactance is proportioate to frequency XL = 2pi * f * L.
So at least, I hope that this proves some benefit, at least
mathematically.

** The simplest test or analysis shows it does not and cannot work.

................ Phil

In a 300 watt amp I built in 1996 using mosfet outputs,
I decided to check the bypassing of the rails.
I had 100,000 uF Sprague computor grade caps in there
with thick leads about 200 mm long.
Then I applied a fast square wave, and watched the rail signal
near the drains increase as F rises, probably a function of the lead's L and
capacitor L
whatever that might be.
I added 1,000 uF caps closer between the drains and 0V, and the
the supression of HF part of the square wave improved.
Then I fitted 0.47uF, got more suppression, then 0.01 at each drain to 0V,
and finally streched the bandwidth for full suppression of the square wave
HF content out to probably 30 MHz+.
Just to stop any more funny business in the mosfets, I put 33 pF between
gates and drains, in addition to the 22o ohm gate resistors with a
ferrite bead.

In an SS discrete component sub woofer filter unit I built last month,
I had large value electros on the rails and still the emitter followers
oscillated at RF, and the only thing that really worked to stop parasitic
RF oscillations was to have 1 or 2 uF from the collector supply to 0V
and at each collector, and to have separate decoupled rails, so the stages were
RF isolated with RC filters, and with a very thick wire 0V rail.

I think it good practice to use plastic caps in addition to electros
for bypassing rails adequately.

The maths and calcs don't matter;
you should build it so it works properly, period, and that means investigative
testing.

Assume that anything you build will oscillate at RF until you proove it cannot.
All too often it is found to oscillate, because the audio circuit
resembles a HF oscillator at RF due to stray C and L components that
have no effect at AF, but a lousy and unintended effect at RF.

Usually by intelligent de-bugging of circuits with careful
RF bypassing placed strategically, one can get audio circuits to not
become RF oscillators, ever.
RF sometimes isn't easily noticed; something might be
oscillating badly at 100 MHz, but the CRO
only goes to 20 MHz, so we see nothing.
Usually by simply using a screw driver to touch existing screen bypass caps or
grounded grid bypasses
or plate bypass caps there will be a give away click in a speaker,
even though we have the input to the audio amp grounded.
It shows there is a click where none should be, and its the starting or stopping

of RF oscillations, also affected by the screw driver affecting the VHF
oscillation.
Judiciously placed 0.1 uF ceramics have their place.


Patrick Turner.

I think most readers usually discount what you say, not necessary because
your conclusions are incorrect, but with the manner in which they are
delivered.
Phil,
You don't know who I am, yet you recklessly hurl those vulgar, and ad
hominem diatribes at me.
I feel sorry for you more than I am angry at you. You seem to relish in this
isolation. I'm not an expert in psy by no stretch of the imagination. I did,
however, minor in it. I can show you the pages in my text books under the
abnormal psychology section. 24 out of 25 people will give a resounding
endorsement, saying "That's Phil." You're not very different than many at
large. Yours is a classic example of a disorder. You really should get help.
Eventually, it will all cave in on you. In the meanwhile, all I can do is
keep you in my prayers and pray that your tortured soul will eventually find
some peace.
Cordially,
west

"Jason R." wrote in message
u...
The ESL and ESR of modern electrolytics is probably lower than that for
the
older types, so it isn't as crucial as it was before. I think the old rule
of thumb was 1% (or perhaps 10%, can't remember). In fact, you will end up
with a parallel resonant LC circuit at HF when the electrolytic goes
inductive if you try to bypass it.

IMO it is mainly audophool drivel to sell overpriced caps. But of course
some will disagree.

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

Jason



"gojamo" wrote in message
ups.com...
It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

"Patrick Turner" = illiterate, autistic, stupid ****head
Phil Allison
"west"


Large capacitors inherently exhibit a small amount of inductance (L)
mostly because of their spiral windings.

** Massive, bloody stupid LIE !!!!!!

There is no damn spiral - all the edges are joined !!!!

I always thought that if you paralleled a
small film capacitor that typically has a low "L" across the
electrolytic,

** Film caps are wound too - dickhead !!

But also in a way that in not inductive.

high frequency current would be easier to pass. Remember that frequency
is
inversely proportionate to the Capacitive reactance. Xc = 1/2pi * f * c
while Inductive reactance is proportioate to frequency XL = 2pi * f *
L.
So at least, I hope that this proves some benefit, at least
mathematically.

** The simplest test or analysis shows it does not and cannot work.


[ snip mucho Turneroid rambling verbal diarrhoea ]


I think it good practice to use plastic caps in addition to electros
for bypassing rails adequately.


** Never helps when the film cap is simply placed ACROSS the electro (
1950s radios notwithstanding) .

Supply decoupling at the end of significant lengths of wiring is another
matter entirely - and was not the issue raised by the OP.



The maths and calcs don't matter;


** So says a total ****wit with no clue how to use them.

See how the Turneroid idiot screwed up all his transformer calcs.

Wouldn't know a " µ " from a ****ing moo cow.



.............. Phil

"west" = another top posting, PITA ******


I think most readers usually discount what you say, not necessary because
your conclusions are incorrect, but with the manner in which they are
delivered.


** Good - I would not want any folk here to upset themselves by
changing a life long habit of believing UTTER BULL**** just because it was
presented politely.


Phil,
You don't know who I am, yet you recklessly hurl those vulgar, and ad
hominem diatribes at me.


** You must be totally delusional as well as top posting fool.

You just replied to Jason R !!!!!!!



I feel sorry for you more than I am angry at you.


** No - please get angry and then ****ING go learn something !!!!!


You seem to relish in this isolation.


** I have no wish to be accepted as a member of RAT.

Only complete fools and arseholes are accepted an RAT.


I'm not an expert in psy by no stretch of the imagination.


** Certainly - you are just another proselytising, audiophool, PITA
hell bent wasting folk's time and money by posting absolute ****e.



I can show you the pages in my text books under the
abnormal psychology section. 24 out of 25 people will give a resounding
endorsement, saying "That's Phil." You're not very different than many at
large. Yours is a classic example of a disorder. You really should get
help.
Eventually, it will all cave in on you. In the meanwhile, all I can do is
keep you in my prayers and pray that your tortured soul will eventually
find
some peace.
Cordially,


** Cordially west - go **** yourself .




............... Phil





"Jason R." wrote in message
u...
The ESL and ESR of modern electrolytics is probably lower than that for
the
older types, so it isn't as crucial as it was before. I think the old
rule
of thumb was 1% (or perhaps 10%, can't remember). In fact, you will end
up
with a parallel resonant LC circuit at HF when the electrolytic goes
inductive if you try to bypass it.

IMO it is mainly audophool drivel to sell overpriced caps. But of course
some will disagree.

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

Jason



"gojamo" wrote in message
ups.com...
It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

I rest my case.

"Phil Allison" wrote in message
...

"west" = another top posting, PITA ******


I think most readers usually discount what you say, not necessary because
your conclusions are incorrect, but with the manner in which they are
delivered.


** Good - I would not want any folk here to upset themselves by
changing a life long habit of believing UTTER BULL**** just because it
was
presented politely.


Phil,
You don't know who I am, yet you recklessly hurl those vulgar, and ad
hominem diatribes at me.


** You must be totally delusional as well as top posting fool.

You just replied to Jason R !!!!!!!



I feel sorry for you more than I am angry at you.


** No - please get angry and then ****ING go learn something !!!!!


You seem to relish in this isolation.


** I have no wish to be accepted as a member of RAT.

Only complete fools and arseholes are accepted an RAT.


I'm not an expert in psy by no stretch of the imagination.


** Certainly - you are just another proselytising, audiophool, PITA
hell bent wasting folk's time and money by posting absolute ****e.



I can show you the pages in my text books under the
abnormal psychology section. 24 out of 25 people will give a resounding
endorsement, saying "That's Phil." You're not very different than many
at
large. Yours is a classic example of a disorder. You really should get
help.
Eventually, it will all cave in on you. In the meanwhile, all I can do
is
keep you in my prayers and pray that your tortured soul will eventually
find
some peace.
Cordially,


** Cordially west - go **** yourself .




.............. Phil





"Jason R." wrote in message
u...
The ESL and ESR of modern electrolytics is probably lower than that for
the
older types, so it isn't as crucial as it was before. I think the old
rule
of thumb was 1% (or perhaps 10%, can't remember). In fact, you will end
up
with a parallel resonant LC circuit at HF when the electrolytic goes
inductive if you try to bypass it.

IMO it is mainly audophool drivel to sell overpriced caps. But of
course
some will disagree.

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

Jason



"gojamo" wrote in message
ups.com...
It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof
that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks

"west"

I rest my case.


** YOU are a ****ing head case - awake or not.

God save us all from such top posting, audiophool assholes.



................. Phil

"Jason R." wrote in message
u...
IMO it is mainly audophool drivel to sell overpriced caps.

Mainly, he says!

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

I'm still waiting on Frank on dA to give me any examples where circuit
impedance is within 10% of ESR/L. :^)

Tim

--
"California is the breakfast state: fruits, nuts and flakes."
Website: http://webpages.charter.net/dawill/tmoranwms

Patrick Turner said:

In a 300 watt amp I built in 1996 using mosfet outputs,
I decided to check the bypassing of the rails.
I had 100,000 uF Sprague computor grade caps in there
with thick leads about 200 mm long.
Then I applied a fast square wave, and watched the rail signal
near the drains increase as F rises, probably a function of the lead's L and
capacitor L
whatever that might be.
I added 1,000 uF caps closer between the drains and 0V, and the
the supression of HF part of the square wave improved.
Then I fitted 0.47uF, got more suppression, then 0.01 at each drain to 0V,
and finally streched the bandwidth for full suppression of the square wave
HF content out to probably 30 MHz+.


What bearing did it have on the perceived sound?


Just to stop any more funny business in the mosfets, I put 33 pF between
gates and drains, in addition to the 22o ohm gate resistors with a
ferrite bead.


Thus adding more capacity to the gate-source cap that is large
already?

--
Sander de Waal
" SOA of a KT88? Sufficient. "

in article , west at
wrote on 3/7/05 9:07 AM:

I think most readers usually discount what you say, not necessary because
your conclusions are incorrect, but with the manner in which they are
delivered.
Phil,
You don't know who I am, yet you recklessly hurl those vulgar, and ad
hominem diatribes at me.
I feel sorry for you more than I am angry at you. You seem to relish in this
isolation. I'm not an expert in psy by no stretch of the imagination. I did,
however, minor in it. I can show you the pages in my text books under the
abnormal psychology section. 24 out of 25 people will give a resounding
endorsement, saying "That's Phil." You're not very different than many at
large. Yours is a classic example of a disorder. You really should get help.
Eventually, it will all cave in on you. In the meanwhile, all I can do is
keep you in my prayers and pray that your tortured soul will eventually find
some peace.
Cordially,
west


Heck West, all Phil's tortured soul wants to do is torture YOUR ass.

You should have learned in psych class or in the school of life that it's
pointless to argue with the mentally ill. Anosognosia is typical of mental
illness . . . .

BTW, how would you "classify" your good buddy, anyway?

Jon

François Yves Le Gal said:


Thus adding more capacity to the gate-source cap that is large
already?

Gate - drain gate - source.


Yep, I saw my mistake after hitting "send".

--
Sander de Waal
" SOA of a KT88? Sufficient. "

On Mon, 07 Mar 2005 14:07:18 GMT, "west"
wrote:

You don't know who I am, yet you recklessly hurl those vulgar, and ad
hominem diatribes at me.

Kill-file the asshole troll and be done with it. The troll feeds on
attention. Don't reward bad behavior by responding.

On Tue, 08 Mar 2005 12:46:03 +1000, Wayne McDermott
wrote:

Hey Phil,
You should wander over to sci.math, you have a soul mate there ! Though
James tends to be more urbane than you I'm you could both benefit by
comparing medications !!

Why feed the troll?

Phil Allison wrote:
"Patrick Turner" = illiterate, autistic, stupid ****head

Phil Allison
"west"


Large capacitors inherently exhibit a small amount of inductance (L)
mostly because of their spiral windings.

** Massive, bloody stupid LIE !!!!!!

There is no damn spiral - all the edges are joined !!!!


I always thought that if you paralleled a
small film capacitor that typically has a low "L" across the
electrolytic,

** Film caps are wound too - dickhead !!

But also in a way that in not inductive.


high frequency current would be easier to pass. Remember that frequency
is
inversely proportionate to the Capacitive reactance. Xc = 1/2pi * f * c
while Inductive reactance is proportioate to frequency XL = 2pi * f *
L.
So at least, I hope that this proves some benefit, at least
mathematically.

** The simplest test or analysis shows it does not and cannot work.



[ snip mucho Turneroid rambling verbal diarrhoea ]



I think it good practice to use plastic caps in addition to electros
for bypassing rails adequately.



** Never helps when the film cap is simply placed ACROSS the electro (
1950s radios notwithstanding) .

Supply decoupling at the end of significant lengths of wiring is another
matter entirely - and was not the issue raised by the OP.




The maths and calcs don't matter;



** So says a total ****wit with no clue how to use them.

See how the Turneroid idiot screwed up all his transformer calcs.

Wouldn't know a " µ " from a ****ing moo cow.



............. Phil


Hey Phil,
You should wander over to sci.math, you have a soul mate there ! Though
James tends to be more urbane than you I'm you could both benefit by
comparing medications !!

Wayne McDermott

"Wayne McDermott"
Hey Phil,


** Hey - **** you !


You should wander over to sci.math, you have a soul mate there !


** You should go straight to hell - there are thousands of assholes *just
like you* there.





............... Phil

dizzy wrote:
On Tue, 08 Mar 2005 12:46:03 +1000, Wayne McDermott
wrote:


Hey Phil,
You should wander over to sci.math, you have a soul mate there ! Though
James tends to be more urbane than you I'm you could both benefit by
comparing medications !!


Why feed the troll?

Hi dizzy,

I know that's what it looks like but I was motivated by a desire to
help...honest. It could be they *would* help each other.

OK OK, you're right, I'm rattling the cage. It's cheaper than visiting
the zoo. Sorry.

Wayne

"Tim Williams" wrote in message
...
"Jason R." wrote in message
u...

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

I'm still waiting on Frank on dA to give me any examples where circuit
impedance is within 10% of ESR/L. :^)

Tim

--
"California is the breakfast state: fruits, nuts and flakes."
Website: http://webpages.charter.net/dawill/tmoranwms



The closest he'll come is probably something like a cathode bypass cap for
some out-there valve like a 7242 (µ=9, ra=82ohm), hence rk=ra/(µ+1)=8.2ohm.
It'll be a big cap, too. Of course, these types of valves were designed for
series regulator service, and thus cathode bypass caps aren't necessary.

Jason

"Jason R." wrote in message
u...

"Tim Williams" wrote in message
...
"Jason R." wrote in message
u...

In any case, have you ever seen a valve circuit operating at impedances
comparable to that of the ESR of an electrolytic? Didn't think so.

I'm still waiting on Frank on dA to give me any examples where circuit
impedance is within 10% of ESR/L. :^)

Tim

--
"California is the breakfast state: fruits, nuts and flakes."
Website: http://webpages.charter.net/dawill/tmoranwms



The closest he'll come is probably something like a cathode bypass cap for
some out-there valve like a 7242 (µ=9, ra=82ohm), hence
rk=ra/(µ+1)=8.2ohm. It'll be a big cap, too. Of course, these types of
valves were designed for series regulator service, and thus cathode bypass
caps aren't necessary.

Jason

Gee, the omission of a few words there totally changed the meaning... of
course what I meant is that when used as a the pass valve in a series
regulator, cathode bypass caps aren't necessary.

Jason

Sander deWaal wrote:

Patrick Turner said:

In a 300 watt amp I built in 1996 using mosfet outputs,
I decided to check the bypassing of the rails.
I had 100,000 uF Sprague computor grade caps in there
with thick leads about 200 mm long.
Then I applied a fast square wave, and watched the rail signal
near the drains increase as F rises, probably a function of the lead's L and
capacitor L
whatever that might be.
I added 1,000 uF caps closer between the drains and 0V, and the
the supression of HF part of the square wave improved.
Then I fitted 0.47uF, got more suppression, then 0.01 at each drain to 0V,
and finally streched the bandwidth for full suppression of the square wave
HF content out to probably 30 MHz+.

What bearing did it have on the perceived sound?

Absolutely sweet farnarkle adamsky.

I arranged all these bypassing measures as natural good practice
well before I even tested it with Motzartian signals.

But I did fool a group of audiophiles by switching from a class A 50 watt tube
amp
to the 300 watter without telling them.
Nodody noticed a thing until I announced I had changed something.
Then I had then guess what the dickens I had changed.
I said it was a major change.
After a few ninutes, they gave up, so I owned up to my horrible tricks.
Then one of them said "well now that you mention it, this mosfet amp
sounds a bit thin in the top end..."
That is the usual criticism of mosfet amps.
OK, I didn't argue.

But now I have far better speakers than I had for that test,
the outcome might be different.
The SEAS based designs were a vast improvement over the
cheap asian drivered speakers I had used,
despite having laboured so long and hard to get the response to be flat
via tweaks to the crossovers.




Just to stop any more funny business in the mosfets, I put 33 pF between
gates and drains, in addition to the 22o ohm gate resistors with a
ferrite bead.

Thus adding more capacity to the gate-source cap that is large
already?

There is a large amount of C between gate and source, but its effectively
reduced by the source follower connection.
The current ability of the transistor VAS stage is quite ample to
provide more than enough current to give a fast enough rise time
to the G-S C.
The Drain to Gate C is far less, and the drain circuit can act like an HF tank
circuit
if parasitics are not suppressed.
The use of source follower means there is no voltage gain, so no miller effect ,
so the 33pF has very little effect on the C already being powered by the VAS
stage which effectively is a very high output impedance voltage source
equal to the collector resistance of the BF469/470 complementary driver
stage I happen to be using.
When the error signal is examined with a square wave present, the
amp easly compensates for the droop in open loop HF
response above 5 kHz with a slightly raised leading ege of the square wave,
free of ring oscillations, so all is well.

The mosfet amp I made was far easier to stabilise compared to a
BJT amp I recently cobbled together, and did not suffer from
latch up cross conduction at above 20 kHz..
It has a simpler circuit because it does not need two stage buffereing between
the VAS
and the output devices.

I doubt I will ever build another bjt power amp.

Patrick Turner.








--
Sander de Waal
" SOA of a KT88? Sufficient. "

Sander deWaal wrote:

François Yves Le Gal said:

Thus adding more capacity to the gate-source cap that is large
already?

Gate - drain gate - source.

Yep, I saw my mistake after hitting "send".

It still makes sense what you typed.

IN my 300 watter, there are 3 mosfets each side of a
complementary follower pair.
G-S C is about 400 pF per mosfet.
But because of the follower action and that
only about 1/20 of the output voltage existis across the G-S,
the capacitance seen by the driver stage is 1/20 less than 400 pF.
So for 6 mosfets, there is effectively about only 150 pF, or not much
more than a few triodes.
Adding 33 pF from G to *drain* makes only a slight difference to
the audio open loop gain and phase shift, but prevents the HF
oscillations.


Patrick Turner.



--
Sander de Waal
" SOA of a KT88? Sufficient. "

On 6 Mar 2005 21:21:21 -0800, "gojamo" wrote:

It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks


i've seen that done a lot, but if you think about it, it is probably the wrong
place to place them. first, it is a high frequency problem, therefore the lead
to the main filter would be an inductor, and would block the high frequency
impulses you so much want to remove! you might even make things worse with all
these stray components.

you need to filter the high frequency 'errors' on your supply rails at the point
they are causing the problem, and possibly you will have to pay a lot of
attention to where you ground out these caps.

it's hard to calculate the size of the caps, it's probably easier to just use
'standard' bypass types and see if they work. if you do want to calculate, you
need to think of the amount of energy you need in the cap to squash a possible
pulse! i would pay more attention to the type of cap and its placement, and
just grab a .1 or something smaller. higher power stages require bigger caps to
supply local impulses.

with audio systems of course the requirement is only for oscillation problems,
the bandwidth of audio is well within the capability of today's ele. caps.

i've worked on equipment that had supply problems, and have actually removed
these caps as being worthless to the problem, which is usually the ground wiring
scheme. and yes i've seen factory wired supply's that failed every rule in the
book!

"Mister" wrote in message
...
On 6 Mar 2005 21:21:21 -0800, "gojamo" wrote:

It's common to parallel electrolytic capacitors in a power supply or
coupling with polypropolene or other non electrolytic capacitor to
presumably improve high frequency performance. Is there any proof that
this is beneficial? Is there a rule of thumb as to the size of such
capacitors? Thanks


i've seen that done a lot, but if you think about it, it is probably the
wrong
place to place them. first, it is a high frequency problem, therefore the
lead
to the main filter would be an inductor, and would block the high
frequency
impulses you so much want to remove! you might even make things worse with
all
these stray components.

you need to filter the high frequency 'errors' on your supply rails at the
point
they are causing the problem, and possibly you will have to pay a lot of
attention to where you ground out these caps.

it's hard to calculate the size of the caps, it's probably easier to just
use
'standard' bypass types and see if they work. if you do want to calculate,
you
need to think of the amount of energy you need in the cap to squash a
possible
pulse! i would pay more attention to the type of cap and its placement,
and
just grab a .1 or something smaller. higher power stages require bigger
caps to
supply local impulses.

with audio systems of course the requirement is only for oscillation
problems,
the bandwidth of audio is well within the capability of today's ele. caps.

i've worked on equipment that had supply problems, and have actually
removed
these caps as being worthless to the problem, which is usually the ground
wiring
scheme. and yes i've seen factory wired supply's that failed every rule in
the
book!

OK, everyone's preoccupied with HF oscillation, inductance, etc. - it's a
hell of a lot simpler than that.
Yeah, new 'lytics are much "faster" than old (in a model of a ladder of very
small C's linked together by small R's, the R's are have lower value, but
they're still there. Don't believe me? Short out a large HV 'lytic
*momentarily* (watch out, the new small ones from switchers, incapable of
any ripple current, will burn out or explode - indication of quality,
actually - *real* small R component, so just brush the shorting wire past
the pos. terminal to get a spark). Have a VTVM or good high-R multitester
across the terminals reading the voltage. You'll see the voltage go down to
0, and then ... quickly climb back up to a substantial portion of the
original charge voltage. That's the C's at the end of the ladder charging
the C's at the terminals through a series of R's (using our little model)).
You can repeat, and see the voltage climb again.
So. The huge 'lytic hangin' around your HT power supply does, in fact,
store a huge charge, and willmake up for an underrated transformer during a
short series of high-current peaks, but it *will* sag, 'cos the C your
current drain's *really* seeing is the first C in the ladder model, which
gwts depleted *real* fast.
Further, it's not too practical to hang a huge 'lytic *right* at the
current soak - you have to use real - life *wire*, which,, too, has R.
Yeah, you can use giantsilver planks for the HT rail, but ... not too
practical, is it?
So, we take a smaller cap, which doesn't need to handle the high-ripple
current of the filter cap, and doesn't need to be a chassis away from the
current soak. It could be stuck *right on* the current drain, by very short
leads (it's much smaller than the 'lytic, it can hang right off a toob
socket), thus eliminating the need for hefty wire. "rule of the thumb" (you
can't beat your spouse with any stick thicker than your thumb - either US or
British law - can't remember) is 10% 10% etc., how ever long you want to
stretch it. Pretty arbitrary, as far as I could tell. As an aside, the C a
rect. toob can see, according to it's data sheet, assumes ? an ideal cap?
Or a cap with plenty of R? In other words, with an ideal C at the rect tube,
any simulation should fail at the turn-on point (the silly standby switch
being hit) - the current rating of the tube will be exceeded, before it even
reaches the (whatever %) duty cycle it will see with thecap fully charged &
associated drain pulling its power). You can stick in a *huge* slow 'lytic,
and the rect toob won't even *see* most of it at 120Hz - in other words, you
should be able to stick in a bigger 'lytic than an oil cap - sort'a like
putting a resistor in front of the cap.
Finally, bypassing caps is a practice used not only by audiophools, but by
good engineers in *every* field, not jjust the computer /rf crew. You'll
see it done everywhere, just look for it. From switcher PS's to laser PS.
A transient is a transient. Put a scope probe on the CT of a PP amp - biase
it to C. Watch the ripple as you drive it hard, Now, put a smallish (10%)
bypass cap right where the probe's connected - see? Neat, huh? Now, take
the B+ wire, and replace it with a real thin chunk. Whoa - with the bypass
in place, there's no difference. Without - horrors!
It's simple, it's practical, it allows you to use cheaper caps, it allows
you to use *smaller* caps. And no, you don't bypass across the 'lytic's
terminals - that's silly. And bypassing the bypass gives you progressively
diminishing gains - after 2, you might as well stop.
-dim