Sign in with
Sign up | Sign in
Your question

Tantalum VS regular electrolytics in Audio filters and pre..

Last response: in Home Audio
Share
Anonymous
September 22, 2005 4:19:50 AM

Archived from groups: rec.audio.pro (More info?)

Any pros or cons? White papers..

Do Tantalum caps increase distortion over regular electrolytics when used in
audio circuitry?

Thanks

jona
Anonymous
September 22, 2005 4:19:51 AM

Archived from groups: rec.audio.pro (More info?)

Jona Vark <noemail@all.com> wrote:
>Any pros or cons? White papers..

They are different.

>Do Tantalum caps increase distortion over regular electrolytics when used in
>audio circuitry?

Maybe, or maybe not, depending on what the circuit is and what the cap is
being used for and what kind of tantalums or aluminum electrolytics you
are talking about.

This is a very vague question. In general, if you want better ESR in a
smaller package, you have to go with tantalums. However, these days the
ESR and high frequency response on aluminums are getting much better.
Proper circuit design can minimize most of these issues.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."
September 22, 2005 4:19:51 AM

Archived from groups: rec.audio.pro (More info?)

Pooh Bear wrote:
> Jona Vark wrote:
>
> > Any pros or cons? White papers..
> >
> > Do Tantalum caps increase distortion over regular electrolytics when used in
> > audio circuitry?
>
> With or without DC bias ?
>
> Regular electrolytics actually have vanishingly small distortion even when
> zero-biased if used intelligently. So small that even an AP test set will only
> see anything of interest at very low frequencies and elevated level. I've done
> this test myself btw.
>
> Graham


I salute you for doing a measurement and putting some science back into
the field. I'm sure some so called the "golden ears" will disagree
with you based on what they think they can hear.

Mark
Related resources
Can't find your answer ? Ask !
Anonymous
September 22, 2005 7:10:57 AM

Archived from groups: rec.audio.pro (More info?)

Jona Vark wrote:

> Any pros or cons? White papers..
>
> Do Tantalum caps increase distortion over regular electrolytics when used in
> audio circuitry?

With or without DC bias ?

Regular electrolytics actually have vanishingly small distortion even when
zero-biased if used intelligently. So small that even an AP test set will only
see anything of interest at very low frequencies and elevated level. I've done
this test myself btw.

Graham
September 22, 2005 2:20:07 PM

Archived from groups: rec.audio.pro (More info?)

Years ago, an article by Dutch High-end researcher / editor
Peter van Willenswaard stated that even when used in
suplly filtering, tantaliums did introduce audible distortion.

Since reading this article i just never used them again and
instead, when low esr was required, put a 100nF foil capacitor
in parallel with an Aluminium-electrolytic.

When it comes to using capacitors in the audio-path, even
Aluminium-electrolytics are te be avoided.
I recently done a complete revision of a 40-channel DDA
mixing console that has been in use for 12 years, and was
shocked to see distortion and deviation in frequency-responce
all due to aged electrolytics in the signal-path.
So "recapping" is a good thing and should have been done
allready some years earlyer.

Now electrolytics are specified some 1000 hours life by
their manufacturers. There are of cause more expensive /
audio-grade ones that will perhaps do 2000 hours or 5000
but, especially due to the reverse polarity situation,
still will age and fail.

So it would be good to replace them by _bipolar_ electrolytics
although they have probably the same lifetime and a much
higher price.

Modifying a cirquit to avoid coupling capacitors will allow
a DC-current, caused by amplifier off-set, flowing through
potentiometers who than will be "cracking" within weeks.
This can be avoided a bit by replacing op-amps by types
that do have a very small off-set like replacing TL072 by
TL052.

Finally, I once visited "an evening with Rupert Neve" by
the Dutch section of the Audio Engineering Society where
the man spend one full hour talking about his family and
motorbikes. After a coffee-brake he told a few things about
audio of which one will go public here:

When applying a little off-set, like for example 1 Volt,
to all amplifiers in a cirquit, electrolytics are polarized
correctly most of the time ( no / small signal ) and
will have longer life and give lower distortion. At the
same time, output cirquits of (discrete) op-amps will remain
in a class-a state for these small signals.

He did not say if this technique was actually applied in
his consoles or apparatus.

Andre





Jona Vark wrote:
> Any pros or cons? White papers..
>
> Do Tantalum caps increase distortion over regular electrolytics when used in
> audio circuitry?
>
> Thanks
>
> jona
>
>
Anonymous
September 22, 2005 2:20:08 PM

Archived from groups: rec.audio.pro (More info?)

Andre wrote:

> Years ago, an article by Dutch High-end researcher / editor
> Peter van Willenswaard stated that even when used in
> suplly filtering, tantaliums did introduce audible distortion.

A circuit has to be complete rubbish to be that significantly influenced by the
power supply rail.

Sounds like complete hokum to me.

" High-end researcher / editor " types have a required need to make simple things
seem complicated. If the truth was known thety'd all be out of jobs.

Graham
Anonymous
September 22, 2005 2:20:08 PM

Archived from groups: rec.audio.pro (More info?)

"Andre" <ten0fingers@hotmail.com> wrote in message
news:43326937$0$2795$e4fe514c@dreader28.news.xs4all.nl

> Years ago, an article by Dutch High-end researcher /
> editor Peter van Willenswaard stated that even when used
> in suplly filtering, tantaliums did introduce audible
> distortion.

British researcher Peter Belt recommends a product called
"Real Foil". Here's what one of his customers says about
"Real Foil":

"I've applied strips of Real Foil to all of my stereo
equipment, including power cords and interconnects, and have
treated every piece of furniture in the room, plus numerous
other objects, all resulting in a significant improvement to
the sound. I've also applied Real Foil to many of my already
highly-treated (for me) CDs, which is what I'll be
concentrating on with my new order. Every time I apply a
strip to a CD I'm very impressed with the immediately
resulting improvement. It seems to me that the effect grows
as I listen, so that by the time I've reached the end of the
CD the sound is even better than when the strip was first
applied. It also seems to me that the effect of applying a
strip of Real Foil varies with the quality of the recording.
If it's a good quality recording to begin with, I notice a
bit of an increase in perceived volume together with an
increase in clarity and richness and a more natural overall
sound. On lesser quality recordings I find a decrease in
volume, which I compensate for with about an additional 2 dB
volume adjustment. With that adjustment I also get the
enhancements in sound that I get with the better quality
recordings. I think the decrease in volume is the result of
the Real Foil reducing some grit and edge which contributes
to the volume perceived by my ears. When that's gone, I need
to increase the volume setting to get the same perceived
volume that I was used to. Strange."


> When it comes to using capacitors in the audio-path, even
> Aluminium-electrolytics are te be avoided.

It's darn hard to do if you rely on commercial recordings.
The whole thing with electrolytics snaps into focus if you
try to recap anything but a trival piece of audio gear with
plastic film capacitors. If the money don't get you, the
size will.

> I recently done a complete revision of a 40-channel DDA
> mixing console that has been in use for 12 years, and was
> shocked to see distortion and deviation in
> frequency-responce all due to aged electrolytics in the
> signal-path.

If you're shocked to see off-spec performance in a 12 year
old piece of equipment, I'm oh so pleased that you finally
received a wake-up call. Nothing's forever!

> So "recapping" is a good thing and should have been done
> allready some years earlier.

Yes, but you're talking about replacing defective
components, which sheds zero light on issues related to new
components.

> Now electrolytics are specified some 1000 hours life by
> their manufacturers. There are of cause more expensive /
> audio-grade ones that will perhaps do 2000 hours or 5000
> but, especially due to the reverse polarity situation,
> still will age and fail.

Last time I looked, commodity electrolytics were rated for
2,000 - 10,000 hour life, under stressful conditions. Audio
coupling caps aren't under a lot of stress.
Anonymous
September 22, 2005 2:26:06 PM

Archived from groups: rec.audio.pro (More info?)

Mark <makolber@yahoo.com> wrote:
>Pooh Bear wrote:
>> Jona Vark wrote:
>>
>> > Any pros or cons? White papers..
>> >
>> > Do Tantalum caps increase distortion over regular electrolytics when used in
>> > audio circuitry?
>>
>> With or without DC bias ?
>>
>> Regular electrolytics actually have vanishingly small distortion even when
>> zero-biased if used intelligently. So small that even an AP test set will only
>> see anything of interest at very low frequencies and elevated level. I've done
>> this test myself btw.
>
>I salute you for doing a measurement and putting some science back into
>the field. I'm sure some so called the "golden ears" will disagree
>with you based on what they think they can hear.

Problem is that I'm willing to believe that high order junk that may be
below the level of the AP test set might be audible. And I'm also willing
to believe that there might be stuff seen on transient signals that won't
be so noticeable on a single-tone THD test or a two-tone IMD test.

But I do agree that intelligent use of electrolytics is the key... there
are a lot of circuits out there that do some pretty boneheaded things.
You can build surprisingly clean circuits with electrolytics as coupling
and decoupling caps, but not everybody does.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 22, 2005 3:01:11 PM

Archived from groups: rec.audio.pro (More info?)

Is it possible to summarize or point to good, practical, practice? I
think what I read is

- if space is not a problem, signal path caps should be plastic film
(is mica still best for small values?)

- bipolar electrolytics are preferred for zero-bias signal path caps
(what chemistry? Aluminum?) when space/cost is an issue

- aluminum caps preferred for biased applications

Are there some quantitative measurements that help in selecting
components? perhaps some effects are subtle enough that we don't have
other measurement devices than ears to find them; hopefully that is not
the only guide we have. Are we principally concerned with
non-linearity, or noise, or?

thanks

dhs
Anonymous
September 22, 2005 3:01:41 PM

Archived from groups: rec.audio.pro (More info?)

Pooh Bear <rabbitsfriendsandrelations@hotmail.com> wrote:
>Andre wrote:
>
>> Years ago, an article by Dutch High-end researcher / editor
>> Peter van Willenswaard stated that even when used in
>> suplly filtering, tantaliums did introduce audible distortion.
>
>A circuit has to be complete rubbish to be that significantly influenced by the
>power supply rail.

Any single-ended gain stage without a constant-current source is.

I suppose you could think of a single-ended gain stage as having the
power supply in series with the output, so the supply decoupling cap
is really a signal coupling cap in disguise.

>Sounds like complete hokum to me.

I don't think it sounds like hokum, but it DOES sound like a good reason
to use constant-current sources.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."
September 22, 2005 3:03:13 PM

Archived from groups: rec.audio.pro (More info?)

I agree witch you but took that mans article as a starter for
a "rule of thumb"

Andre



Pooh Bear wrote:
> Andre wrote:
&><
> A circuit has to be complete rubbish to be that significantly influenced by the
> power supply rail.
>
> Sounds like complete hokum to me.
>
> " High-end researcher / editor " types have a required need to make simple things
> seem complicated. If the truth was known thety'd all be out of jobs.
>
> Graham
>
Anonymous
September 22, 2005 3:03:14 PM

Archived from groups: rec.audio.pro (More info?)

Andre wrote:

> I agree witch you but took that mans article as a starter for
> a "rule of thumb"

I seriously wouldn't treat it as very scientific.

The influence of power supply rails is *very* low on well designed equipment ( if not
very, very, very low ! ).

The main effect that power rails have is ac power line frequency ripple ( hum ). It's a
sawtooth like waveform typically. The most important thing to do is to remove it. This
is why regulated power supplies are normally used for sensitive stages. The problem
simply goes away.

Graham

>
>
> Andre
>
> Pooh Bear wrote:
> > Andre wrote:
> &><
> > A circuit has to be complete rubbish to be that significantly influenced by the
> > power supply rail.
> >
> > Sounds like complete hokum to me.
> >
> > " High-end researcher / editor " types have a required need to make simple things
> > seem complicated. If the truth was known thety'd all be out of jobs.
> >
> > Graham
> >
Anonymous
September 22, 2005 4:15:57 PM

Archived from groups: rec.audio.pro (More info?)

Arny Krueger wrote:

> "Andre" <ten0fingers@hotmail.com> wrote in message
> news:43326937$0$2795$e4fe514c@dreader28.news.xs4all.nl
>
> > Now electrolytics are specified some 1000 hours life by
> > their manufacturers. There are of cause more expensive /
> > audio-grade ones that will perhaps do 2000 hours or 5000
> > but, especially due to the reverse polarity situation,
> > still will age and fail.
>
> Last time I looked, commodity electrolytics were rated for
> 2,000 - 10,000 hour life, under stressful conditions. Audio
> coupling caps aren't under a lot of stress.

The 'life' for most commercial product ( usually 1000hr or 2000 hour
btw ) relates to operation at the *full rated temperature* ( usually
85 or 100 C ) and the *maximum rated ripple current*. Quite a severe
test !

Even then, the 'life spec' only says that capacitance may have dropped
by 10% max typically. Other parameters typically are also only
warranted to 90% of spec at 'end of life'. Operating below rated temp
increases life 2x for every 10C IIRC. Coupling caps also pass almost
no current at all. No problem there ! It's the PSU reservoir caps
that'll be the first to cause trouble normally.

In light duty with normal temperatures, electrolytics will last almost
indefinitely. In the end the electrolyte will dry up and that will be
terminal.

Oh - and since coupling caps are often zero-biased it's effectively as
if they're never even switched on ! No volts to 'wear them out' at all
!

Graham
Anonymous
September 22, 2005 4:15:58 PM

Archived from groups: rec.audio.pro (More info?)

"Pooh Bear" <rabbitsfriendsandrelations@hotmail.com> wrote
in message news:4332926D.5B8473F9@hotmail.com
> Arny Krueger wrote:
>
>> "Andre" <ten0fingers@hotmail.com> wrote in message
>> news:43326937$0$2795$e4fe514c@dreader28.news.xs4all.nl
>>
>>> Now electrolytics are specified some 1000 hours life by
>>> their manufacturers. There are of cause more expensive
>>> / audio-grade ones that will perhaps do 2000 hours or
>>> 5000 but, especially due to the reverse polarity
>>> situation, still will age and fail.
>>
>> Last time I looked, commodity electrolytics were rated
>> for 2,000 - 10,000 hour life, under stressful
>> conditions. Audio coupling caps aren't under a lot of
>> stress.
>
> The 'life' for most commercial product ( usually 1000hr
> or 2000 hour btw ) relates to operation at the *full
> rated temperature* ( usually 85 or 100 C ) and the
> *maximum rated ripple current*. Quite a severe test !

Indeed.

The thing is that almost nobody runs electronics at 85 or
100C. I'd expect the stress on capacitors with temperature
to be at least linear, and perhaps more like exponential. If
a component is warm to the touch that's 105F. If its too hot
to touch it might only be 120F. This is well short of 100C
or 212F.

> Even then, the 'life spec' only says that capacitance may
> have dropped by 10% max typically.

Right. To put that 10% into perspective, ab electrolytic
often has an intial tolerance of -20 +80. The average part,
after losing 10% still probably has more than name plate
capacitance. Furthermore, the designer probably specd the
part far larger than it had to be for the equipment to meet
spec.

> Other parameters
> typically are also only warranted to 90% of spec at 'end
> of life'. Operating below rated temp increases life 2x
> for every 10C IIRC. Coupling caps also pass almost no
> current at all. No problem there ! It's the PSU reservoir
> caps that'll be the first to cause trouble normally.

The PS is where the stress is at. Its hot and there might
even be some real current in there.

> In light duty with normal temperatures, electrolytics
> will last almost indefinitely. In the end the electrolyte
> will dry up and that will be terminal.

That has been my experience over 50 years of working with
this stuff. BTW have electrolytics improved over that time
period or what?

> Oh - and since coupling caps are often zero-biased it's
> effectively as if they're never even switched on ! No
> volts to 'wear them out' at all !

Not to mention the vanishing amounts of signal current
passing through them.
Anonymous
September 22, 2005 6:31:30 PM

Archived from groups: rec.audio.pro (More info?)

<dhs@mit.edu> wrote:
>Is it possible to summarize or point to good, practical, practice? I
>think what I read is
>
>- if space is not a problem, signal path caps should be plastic film

Space is always a problem. So is price.

>(is mica still best for small values?)

Depends. Mica is pretty good, but there are ceramics today that are
just as good at high frequencies.

>- bipolar electrolytics are preferred for zero-bias signal path caps
>(what chemistry? Aluminum?) when space/cost is an issue

Yes, the nonpolarized electrolytics are aluminum types. In general,
though, this is a bad application and it is really, really to your
advantage to avoid ever having unbiased caps in the circuit.

>- aluminum caps preferred for biased applications

For the most part, I'd tend to agree.

>Are there some quantitative measurements that help in selecting
>components? perhaps some effects are subtle enough that we don't have
>other measurement devices than ears to find them; hopefully that is not
>the only guide we have. Are we principally concerned with
>non-linearity, or noise, or?

Well, it depends on the circuit. The HF numbers can be useful. ESR
is very useful for decoupling applications, but usually isn't that big
a deal for coupling caps. For distortion spectra, you won't see anything
on the data sheet so you have to use your ears.

And there _are_ a bunch of new chemistries out there, like the dry
electrolytics and the new super-long-life ones. I don't know anyone
who has seriously auditioned them.

When you buy aluminums, there are a lot of different grades, and the
ones with better HF numbers (usually intended for switching supplies)
almost always sound better to me.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 22, 2005 7:23:37 PM

Archived from groups: rec.audio.pro (More info?)

"Scott Dorsey" <kludge@panix.com> wrote in message
news:D gsuri$n0r$1@panix2.panix.com...
> Jona Vark <noemail@all.com> wrote:
> >Any pros or cons? White papers..
>
> They are different.
>
> >Do Tantalum caps increase distortion over regular electrolytics when used
in
> >audio circuitry?
>
> Maybe, or maybe not, depending on what the circuit is and what the cap is
> being used for and what kind of tantalums or aluminum electrolytics you
> are talking about.
>
> This is a very vague question.


yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
working with a new device and designing preamp/output amp stages for it and
I have very little space.

thanks.
Anonymous
September 22, 2005 7:23:38 PM

Archived from groups: rec.audio.pro (More info?)

Jona Vark <noemail@all.com> wrote:
>
>yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
>working with a new device and designing preamp/output amp stages for it and
>I have very little space.

Try some caps and see how they sound. Almost certainly you will hear
more coloration from tantalum types than anything else, and of course
dry-slug tantalums fail into shorts when they do fail, which is annoying.

But you'll get better density with the tantalums. Your goal might turn
out to be figuring out how to reduce values in order to get things to
fit.

You may want to look at some of the new dry electrolytic designs from
Sanyo, which give you better density than typical aluminums. I do not
know how they sound but I'd be curious to try them for coupling. I think
they also have the same unfortunate failure mode, though.

Also, of course, eliminating capacitors and DC-coupling things will tend
to be a win in a lot of designs if you can manage it.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 22, 2005 7:23:38 PM

Archived from groups: rec.audio.pro (More info?)

Jona Vark wrote:

> yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
> working with a new device and designing preamp/output amp stages for it and
> I have very little space.

In a blocking application there will be very little signal
voltage across the cap so linearity is much less important.
I don't think it matters much at all what you use.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
Anonymous
September 23, 2005 1:20:08 AM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey wrote:

> Pooh Bear <rabbitsfriendsandrelations@hotmail.com> wrote:
> >Andre wrote:
> >
> >> Years ago, an article by Dutch High-end researcher / editor
> >> Peter van Willenswaard stated that even when used in
> >> suplly filtering, tantaliums did introduce audible distortion.
> >
> >A circuit has to be complete rubbish to be that significantly influenced by the
> >power supply rail.
>
> Any single-ended gain stage without a constant-current source is.

It's been a long time since I've seen one of them in the flesh!


> I suppose you could think of a single-ended gain stage as having the
> power supply in series with the output, so the supply decoupling cap
> is really a signal coupling cap in disguise.

That's fair enough. But the audiophools aslo maintain that you get better sound by
substituting Black Gates in the *digital* and motor supplies too !

Btw, Rubycon are stopping making them I just read. The last orders have already been
taken.


> >Sounds like complete hokum to me.
>
> I don't think it sounds like hokum, but it DOES sound like a good reason
> to use constant-current sources.

Or even power supply regulation.

Graham
Anonymous
September 23, 2005 1:25:18 AM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey wrote:

> Mark <makolber@yahoo.com> wrote:
> >Pooh Bear wrote:
> >> Jona Vark wrote:
> >>
> >> > Any pros or cons? White papers..
> >> >
> >> > Do Tantalum caps increase distortion over regular electrolytics when used in
> >> > audio circuitry?
> >>
> >> With or without DC bias ?
> >>
> >> Regular electrolytics actually have vanishingly small distortion even when
> >> zero-biased if used intelligently. So small that even an AP test set will only
> >> see anything of interest at very low frequencies and elevated level. I've done
> >> this test myself btw.
> >
> >I salute you for doing a measurement and putting some science back into
> >the field. I'm sure some so called the "golden ears" will disagree
> >with you based on what they think they can hear.
>
> Problem is that I'm willing to believe that high order junk that may be
> below the level of the AP test set might be audible.

What sort of minus dB figure do you reckon is needed to ensure inaudibility ?

I'm quite interested in doing some more 'research' into this. What I've seen is quite
interesting. The trick if you want to look closely is to measure across the cap
itself - not just the signal path.

> And I'm also willing
> to believe that there might be stuff seen on transient signals that won't
> be so noticeable on a single-tone THD test or a two-tone IMD test.

Do you have a 'theory' behind that suggestion ?

> But I do agree that intelligent use of electrolytics is the key... there
> are a lot of circuits out there that do some pretty boneheaded things.
> You can build surprisingly clean circuits with electrolytics as coupling
> and decoupling caps, but not everybody does.

Size is everyting with electrolytics ! The bigger the better. ;-)

Graham
Anonymous
September 23, 2005 2:18:17 AM

Archived from groups: rec.audio.pro (More info?)

"Scott Dorsey" <kludge@panix.com> wrote in message
news:D guhqp$eg5$1@panix2.panix.com...
> Jona Vark <noemail@all.com> wrote:
> >
> >yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
> >working with a new device and designing preamp/output amp stages for it
and
> >I have very little space.
>
> Try some caps and see how they sound. Almost certainly you will hear
> more coloration from tantalum types than anything else, and of course
> dry-slug tantalums fail into shorts when they do fail, which is annoying.
>

I wish I could tell the difference between .001% distortion and .05% with my
ears.
Anonymous
September 23, 2005 2:18:18 AM

Archived from groups: rec.audio.pro (More info?)

Jona Vark <noemail@all.com> wrote:
>"Scott Dorsey" <kludge@panix.com> wrote in message
>news:D guhqp$eg5$1@panix2.panix.com...
>> Jona Vark <noemail@all.com> wrote:
>> >
>> >yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
>> >working with a new device and designing preamp/output amp stages for it
>and
>> >I have very little space.
>>
>> Try some caps and see how they sound. Almost certainly you will hear
>> more coloration from tantalum types than anything else, and of course
>> dry-slug tantalums fail into shorts when they do fail, which is annoying.
>
>I wish I could tell the difference between .001% distortion and .05% with my
>ears.

Easy to do when the dominant distortion is 8th order. Physically impossible
to do when the dominant distortion is 3rd order.

This is why the THD number isn't all thet useful. It doesn't tell the whole
story at all. You need to know the whole spectrum to judge audibility,
which is why we are seeing all these proposals like the Geddes-Lee papers
for a distortion metric that correlates well with perception.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 23, 2005 3:26:34 AM

Archived from groups: rec.audio.pro (More info?)

"Scott Dorsey" <kludge@panix.com> wrote in message
news:D gvdct$khj$1@panix2.panix.com...
> Jona Vark <noemail@all.com> wrote:
> >"Scott Dorsey" <kludge@panix.com> wrote in message
> >news:D guhqp$eg5$1@panix2.panix.com...
> >> Jona Vark <noemail@all.com> wrote:
> >> >
> >> >yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
> >> >working with a new device and designing preamp/output amp stages for
it
> >and
> >> >I have very little space.
> >>
> >> Try some caps and see how they sound. Almost certainly you will hear
> >> more coloration from tantalum types than anything else, and of course
> >> dry-slug tantalums fail into shorts when they do fail, which is
annoying.
> >
> >I wish I could tell the difference between .001% distortion and .05% with
my
> >ears.
>
> Easy to do when the dominant distortion is 8th order. Physically
impossible
> to do when the dominant distortion is 3rd order.

Unfortunately, while you may be able to 'hear' it you can't quantify it.


>
> This is why the THD number isn't all thet useful. It doesn't tell the
whole
> story at all. You need to know the whole spectrum to judge audibility,
> which is why we are seeing all these proposals like the Geddes-Lee papers
> for a distortion metric that correlates well with perception.


Most engineering staff still use the THD measurement as criteria in
selecting audio components.
So we have to know what it is and publish it. I am creating a development
board and IDE for a new ASIC and I want to keep the THD minimal on the way
in and out of the ASIC. I was curious about Tants..

> --scott
> --
> "C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 23, 2005 3:26:35 AM

Archived from groups: rec.audio.pro (More info?)

Jona Vark <noemail@all.com> wrote:
>"Scott Dorsey" <kludge@panix.com> wrote in message
>news:D gvdct$khj$1@panix2.panix.com...
>> Jona Vark <noemail@all.com> wrote:
>> >"Scott Dorsey" <kludge@panix.com> wrote in message
>> >news:D guhqp$eg5$1@panix2.panix.com...
>> >> Jona Vark <noemail@all.com> wrote:
>> >> >
>> >> >yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I am
>> >> >working with a new device and designing preamp/output amp stages for
>it
>> >and
>> >> >I have very little space.
>> >>
>> >> Try some caps and see how they sound. Almost certainly you will hear
>> >> more coloration from tantalum types than anything else, and of course
>> >> dry-slug tantalums fail into shorts when they do fail, which is
>annoying.
>> >
>> >I wish I could tell the difference between .001% distortion and .05% with
>my
>> >ears.
>>
>> Easy to do when the dominant distortion is 8th order. Physically
>impossible
>> to do when the dominant distortion is 3rd order.
>
>Unfortunately, while you may be able to 'hear' it you can't quantify it.

Sure I can! I have a Gen Rad 2615 analyzer. Easy to do.

>> This is why the THD number isn't all thet useful. It doesn't tell the
>whole
>> story at all. You need to know the whole spectrum to judge audibility,
>> which is why we are seeing all these proposals like the Geddes-Lee papers
>> for a distortion metric that correlates well with perception.
>
>Most engineering staff still use the THD measurement as criteria in
>selecting audio components.

THD is still useful for a lot of things, like comparing circuits of similar
topology when distortion levels are high, or as a diagnostic to compare a
malfunctioning circuit with a properly operating one. But as a criterion
for equipment selection, it sets a pretty low bar.

>So we have to know what it is and publish it. I am creating a development
>board and IDE for a new ASIC and I want to keep the THD minimal on the way
>in and out of the ASIC. I was curious about Tants..

Tantalum caps may not have any higher THD, but the distortion spectrum
is going to be more oriented toward higher order junk that sounds worse.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 23, 2005 6:25:25 AM

Archived from groups: rec.audio.pro (More info?)

"Scott Dorsey" <kludge@panix.com> wrote in message
news:D gvirg$7qv$1@panix2.panix.com...
> Jona Vark <noemail@all.com> wrote:
> >"Scott Dorsey" <kludge@panix.com> wrote in message
> >news:D gvdct$khj$1@panix2.panix.com...
> >> Jona Vark <noemail@all.com> wrote:
> >> >"Scott Dorsey" <kludge@panix.com> wrote in message
> >> >news:D guhqp$eg5$1@panix2.panix.com...
> >> >> Jona Vark <noemail@all.com> wrote:
> >> >> >
> >> >> >yes... it can appear vague.. as blocking caps mainly. 1 to 10uF. I
am
> >> >> >working with a new device and designing preamp/output amp stages
for
> >it
> >> >and
> >> >> >I have very little space.
> >> >>
> >> >> Try some caps and see how they sound. Almost certainly you will
hear
> >> >> more coloration from tantalum types than anything else, and of
course
> >> >> dry-slug tantalums fail into shorts when they do fail, which is
> >annoying.
> >> >
> >> >I wish I could tell the difference between .001% distortion and .05%
with
> >my
> >> >ears.
> >>
> >> Easy to do when the dominant distortion is 8th order. Physically
> >impossible
> >> to do when the dominant distortion is 3rd order.
> >
> >Unfortunately, while you may be able to 'hear' it you can't quantify it.
>
> Sure I can! I have a Gen Rad 2615 analyzer. Easy to do.
>
> >> This is why the THD number isn't all thet useful. It doesn't tell the
> >whole
> >> story at all. You need to know the whole spectrum to judge audibility,
> >> which is why we are seeing all these proposals like the Geddes-Lee
papers
> >> for a distortion metric that correlates well with perception.
> >
> >Most engineering staff still use the THD measurement as criteria in
> >selecting audio components.
>
> THD is still useful for a lot of things, like comparing circuits of
similar
> topology when distortion levels are high, or as a diagnostic to compare a
> malfunctioning circuit with a properly operating one. But as a criterion
> for equipment selection, it sets a pretty low bar.
>

Scott. No one is going to design our part into a product without some
knowledge of THD.
It is ONE of the criteria people use to judge the quality of an audio
component.




> >So we have to know what it is and publish it. I am creating a development
> >board and IDE for a new ASIC and I want to keep the THD minimal on the
way
> >in and out of the ASIC. I was curious about Tants..
>
> Tantalum caps may not have any higher THD, but the distortion spectrum
> is going to be more oriented toward higher order junk that sounds worse.

Thanks. I you've given me something to consider.

> --scott
> --
> "C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
September 23, 2005 9:55:07 AM

Archived from groups: rec.audio.pro (More info?)

"Scott Dorsey" <kludge@panix.com> wrote in message
news:D gvdct$khj$1@panix2.panix.com
> Jona Vark <noemail@all.com> wrote:
>> "Scott Dorsey" <kludge@panix.com> wrote in message
>> news:D guhqp$eg5$1@panix2.panix.com...
>>> Jona Vark <noemail@all.com> wrote:
>>>>
>>>> yes... it can appear vague.. as blocking caps mainly.
>>>> 1 to 10uF. I am working with a new device and
>>>> designing preamp/output amp stages for it and I have
>>>> very little space.
>>>
>>> Try some caps and see how they sound. Almost certainly
>>> you will hear more coloration from tantalum types than
>>> anything else, and of course dry-slug tantalums fail
>>> into shorts when they do fail, which is annoying.
>>
>> I wish I could tell the difference between .001%
>> distortion and .05% with my ears.

> Easy to do when the dominant distortion is 8th order.
> Physically impossible to do when the dominant distortion
> is 3rd order.

That sounds optimistic to me.

Furhtermore Graham reports, and I can confirm that the
nonlinear distortion from unbiased electrolytic coupling
caps can be well below 0.001%.

For example the LynxTWO audio interface has aluminum
electrolytic coupling caps that appear to be unbiased. It
delivers residuals in the 0.00015% range as shown by
http://www.pcavtech.com/soundcards/LynxTWO/index.htm .

Besides, I see no evidence that the nonlinear distortion
from an unbiased aluminum electrolytic coupling cap is
necessarily high order.

> This is why the THD number isn't all thet useful. It
> doesn't tell the whole story at all. You need to know
> the whole spectrum to judge audibility, which is why we
> are seeing all these proposals like the Geddes-Lee papers
> for a distortion metric that correlates well with
> perception. --scott

Now that I can whole-heartedly agree with.

There is no doubt that higher order nonlinear distortion is
more audible, if for no other reasons that it is far enough
from the fundamental to not be masked by the ear.

Furthermore, low order distortion can be quite audible at
relatively low levels when the source material is
multitones, which music always is.

However, audibility of nonlinear distortion products
below -60 dB (0.1%) is iffy because of the ear's lack of
instantaneous dynamic range, and below -100 dB (0.001%),
there's simply not a chance.
Anonymous
September 23, 2005 1:27:43 PM

Archived from groups: rec.audio.pro (More info?)

"Andre" <ten0fingers@hotmail.com> wrote in message
news:43326937$0$2795$e4fe514c@dreader28.news.xs4all.nl...
> Years ago, an article by Dutch High-end researcher / editor
> Peter van Willenswaard stated that even when used in
> suplly filtering, tantaliums did introduce audible distortion.

That's odd. That the place where onewould least expect artifacts. Did he
attempt to explain how this would be ?


> Finally, I once visited "an evening with Rupert Neve" by
> the Dutch section of the Audio Engineering Society where
> the man spend one full hour talking about his family and
> motorbikes. After a coffee-brake he told a few things about
> audio of which one will go public here:
>
> When applying a little off-set, like for example 1 Volt,
> to all amplifiers in a cirquit, electrolytics are polarized
> correctly most of the time ( no / small signal ) and
> will have longer life and give lower distortion. At the
> same time, output cirquits of (discrete) op-amps will remain
> in a class-a state for these small signals.
>
> He did not say if this technique was actually applied in
> his consoles or apparatus.

Dunno, but the CIB if stuffed full of tantalums ! (And some Dublier Al
electros). And 5532s.

geoff
September 23, 2005 3:26:20 PM

Archived from groups: rec.audio.pro (More info?)

Geoff@work wrote:
> "Andre" <ten0fingers@hotmail.com> wrote in message
> news:43326937$0$2795$e4fe514c@dreader28.news.xs4all.nl...
>
>>Years ago, an article by Dutch High-end researcher / editor
>>Peter van Willenswaard stated that even when used in
>>suplly filtering, tantaliums did introduce audible distortion.
>
>
> That's odd. That the place where onewould least expect artifacts. Did he
> attempt to explain how this would be ?

It was in those years when opamps were rarely used because
there existed only noise generators like uA741 and discrete
op-amps were just to expensive or space consuming.
Cirquits were 2 or 3 transistor pre-amps and thus very sensitive
to suplly ripple, noise or whatever.

I just kept avoiding tantals remembering this persons statement.

If you Google "Peter van Willenswaard" you can find him still
being into audio, beit valves and high-end... so high !

Andre

>>Finally, I once visited "an evening with Rupert Neve" by

&><

>>He did not say if this technique was actually applied in
>>his consoles or apparatus.
>
>
> Dunno, but the CIB if stuffed full of tantalums ! (And some Dublier Al
> electros). And 5532s.
>
> geoff
>
>
Anonymous
September 23, 2005 9:44:18 PM

Archived from groups: rec.audio.pro (More info?)

"Arny Krueger" <arnyk@hotpop.com> wrote in message
news:g7Sdnc6BqtXgTa7eRVn-2w@comcast.com...

> Furthermore, low order distortion can be quite audible at
> relatively low levels when the source material is
> multitones, which music always is.
>
> However, audibility of nonlinear distortion products
> below -60 dB (0.1%) is iffy because of the ear's lack of
> instantaneous dynamic range, and below -100 dB (0.001%),
> there's simply not a chance.

Perhaps not, but there's also no guarantee that a device which
produces -100dB distortion on a single tone will produce the same level of
distortion on multitones. I suspect the only really honest test is an
input-output comparator device. David Hafler tried to introduce this method
of evaluation, but his somewhat simplistic approach was sensitive to the
phase characteristics of the devices, so that any amplifier (for example)
with frequency response flat to 200kHz beat the pants off any amplifier
whose response rolled off at 50kHz, simply because the comparator was
picking up the phase shifts. (Not coincidentally, Hafler at the time was
making power amps with very wide bandwidths.) Somebody at the British QUAD
company introduced the "crooked-wire bypass", a variant on the Hafler
experiments. They compared amplifier outputs with the input signal after
running the latter through a passive network with the same frequency/phase
characteristics as the active device. Worked very nicely, but developing the
passive network for a particular amplifier took a *lot* of work, and you had
o fiddle with it a lot for each individual amplifer you tested.

Perhaps these days it would be easier. I suspect, in fact, that you could
develop a unit which would compare things like capacitors relatively easily,
using music as a test signal.

Peace,
Paul
Anonymous
September 23, 2005 9:44:19 PM

Archived from groups: rec.audio.pro (More info?)

"Paul Stamler" <pstamlerhell@pobox.com> wrote in message
news:S9XYe.301526$5N3.250179@bgtnsc05-news.ops.worldnet.att.net
> "Arny Krueger" <arnyk@hotpop.com> wrote in message
> news:g7Sdnc6BqtXgTa7eRVn-2w@comcast.com...

>> Furthermore, low order distortion can be quite audible at
>> relatively low levels when the source material is
>> multitones, which music always is.

>> However, audibility of nonlinear distortion products
>> below -60 dB (0.1%) is iffy because of the ear's lack of
>> instantaneous dynamic range, and below -100 dB (0.001%),
>> there's simply not a chance.

Note that "nonlinear distortion products" is intended to be
taken in the most general possible way. IOW a product
conforming to the indicated specs may be tested with *any*
combination of tones or other kinds of signal.

There's actually quite a bit of reasonably-priced audio gear
that seems to be impossible to coax into producing
distortion products that sum to more than -100 dB. Quite a
bit of it is digital audio interfaces. And of course, once
the signal is in the digital domain, distortion-free
processing is pretty easy.

Just to get back on topic, the audio gear I'm referring to
with -100 dB or less nonlinear distortion products has a
goodly number of aluminum electrolytic capacitors in the
signal path, as well as in bypass and power supply filter
applications.

> Perhaps not, but there's also no guarantee that a device
> which
> produces -100dB distortion on a single tone will produce
> the same level of distortion on multitones.

Paul, did you miss the part of the post of mine that you
responded to that said the following:

"Furthermore, low order distortion can be quite audible at
relatively low levels when the source material is
multitones, which music always is."

Again, "nonlinear distortion products" is intended to be
taken in the most general possible even vaguely reasonable
way.
Anonymous
September 24, 2005 12:55:41 AM

Archived from groups: rec.audio.pro (More info?)

"Arny Krueger" <arnyk@hotpop.com> wrote in message
news:W-ydnZSyCIcX3KneRVn-gw@comcast.com...
> > Perhaps not, but there's also no guarantee that a device
> > which
> > produces -100dB distortion on a single tone will produce
> > the same level of distortion on multitones.
>
> Paul, did you miss the part of the post of mine that you
> responded to that said the following:
>
> "Furthermore, low order distortion can be quite audible at
> relatively low levels when the source material is
> multitones, which music always is."
>
> Again, "nonlinear distortion products" is intended to be
> taken in the most general possible even vaguely reasonable
> way.

No, I didn't miss it; in fact, I was agreeing with it. I was just pointing
out that tests with -100dB distortion products in the lab may not be
producing levels that low with real music, which is way more complicated
than any of the multitone tests now commonly used. (I use 'em too, and think
they're pretty useful, but they're only a crude map at best.)

Peace,
Paul
Anonymous
September 24, 2005 2:14:24 AM

Archived from groups: rec.audio.pro (More info?)

On Fri, 23 Sep 2005 13:57:29 -0400, "Arny Krueger" <arnyk@hotpop.com>
wrote:

>"Paul Stamler" <pstamlerhell@pobox.com> wrote in message
>news:S9XYe.301526$5N3.250179@bgtnsc05-news.ops.worldnet.att.net
>> "Arny Krueger" <arnyk@hotpop.com> wrote in message
>> news:g7Sdnc6BqtXgTa7eRVn-2w@comcast.com...
>
>>> Furthermore, low order distortion can be quite audible at
>>> relatively low levels when the source material is
>>> multitones, which music always is.
>
>>> However, audibility of nonlinear distortion products
>>> below -60 dB (0.1%) is iffy because of the ear's lack of
>>> instantaneous dynamic range, and below -100 dB (0.001%),
>>> there's simply not a chance.
>
>Note that "nonlinear distortion products" is intended to be
>taken in the most general possible way. IOW a product
>conforming to the indicated specs may be tested with *any*
>combination of tones or other kinds of signal.
>
>There's actually quite a bit of reasonably-priced audio gear
>that seems to be impossible to coax into producing
>distortion products that sum to more than -100 dB. Quite a
>bit of it is digital audio interfaces. And of course, once
>the signal is in the digital domain, distortion-free
>processing is pretty easy.

Just to pick a nit (and compare meanings of 'distortion'), the last
phrase might be "processing free of AUDIBLE distortion is pretty
easy." Most digital processing absolutely DOES cause distortion,
though it may be -140dB or more below max signal.

>
>Just to get back on topic, the audio gear I'm referring to
>with -100 dB or less nonlinear distortion products has a
>goodly number of aluminum electrolytic capacitors in the
>signal path, as well as in bypass and power supply filter
>applications.

With what impedances is that? I presume the outputs being measured
are driving a 100k or 10k ohm input inpedance, which is typical of
most modern audio equipment, but I wonder if the figures go up
significantly into 1k or 600 ohm? Any increase could of course be the
output transistor wimping out as well as the coupling capacitor
driving a much lower impedance.
Anonymous
September 24, 2005 2:06:47 PM

Archived from groups: rec.audio.pro (More info?)

"Paul Stamler" <pstamlerhell@pobox.com> wrote in message
news:hZZYe.302345$5N3.286108@bgtnsc05-news.ops.worldnet.att.net
> "Arny Krueger" <arnyk@hotpop.com> wrote in message
> news:W-ydnZSyCIcX3KneRVn-gw@comcast.com...
>>> Perhaps not, but there's also no guarantee that a device
>>> which
>>> produces -100dB distortion on a single tone will produce
>>> the same level of distortion on multitones.
>>
>> Paul, did you miss the part of the post of mine that you
>> responded to that said the following:
>>
>> "Furthermore, low order distortion can be quite audible
>> at relatively low levels when the source material is
>> multitones, which music always is."
>>
>> Again, "nonlinear distortion products" is intended to be
>> taken in the most general possible even vaguely
>> reasonable way.
>
> No, I didn't miss it; in fact, I was agreeing with it. I
> was just pointing out that tests with -100dB distortion
> products in the lab may not be producing levels that low
> with real music, which is way more complicated than any
> of the multitone tests now commonly used. (I use 'em too,
> and think they're pretty useful, but they're only a crude
> map at best.)

It is considerations like that which placed the limit
at -100 dB. If there were constraints on the test signals,
the limit could be a low as -40 dB, or even less.
Anonymous
September 24, 2005 2:08:48 PM

Archived from groups: rec.audio.pro (More info?)

"Ben Bradley" <ben_nospam_bradley@frontiernet.net> wrote in
message news:jsu8j1te96q8qpmj79ptm1l5rcm388g4hh@4ax.com

>> Just to get back on topic, the audio gear I'm referring
>> to with -100 dB or less nonlinear distortion products
>> has a goodly number of aluminum electrolytic capacitors
>> in the signal path, as well as in bypass and power
>> supply filter applications.

> With what impedances is that?

Output driving input. Audio interfaces commonly have input
impedances in the 5-10K range.

> I presume the outputs
> being measured are driving a 100k or 10k ohm input
> inpedance, which is typical of most modern audio
> equipment, but I wonder if the figures go up
> significantly into 1k or 600 ohm?

They do.

> Any increase could of
> course be the output transistor wimping out as well as
> the coupling capacitor driving a much lower impedance.

I suspect that greater nonlinearity of the active devices
primarily explains the rise in distortion.
!